4 November 2019 Equity Research Asia Pacific

Asia Technology Strategy : Can it gain tech independence?

Technology | Strategy CIC Special

Figure 1: China tech net trade (imports minus exports) waterfall, 2018 (USD bn) The Credit Suisse China Investment Conference 250 (CIC) is one of the most exclusive business 126 gatherings in the region, bringing together business leaders and entrepreneurs from 200 around the region, as well as global institutional and private investors. This year, the Credit Suisse China 150 Investment Conference will explore the theme of 80 ‘Great Expectations’ – a topic that embraces the 100 continued progression of China and the ( 141 ) advancement of some of the country’s most 50 dynamic companies against the current geopolitical ( 36 ) 8 11 12 ( 25 ) ( 3 ) and macroeconomic backdrop. - Research Analysts (52) (20) (50) Manish Nigam SPE Display Semi- Semi/IC ex Mobile Telecom Display Printed Passive Others Total Tech Equip Memory Memory Handset Equipment Circuits Component Net Imports 852 2101 7067 [email protected] Source: CEIC, Credit Suisse estimates Randy Abrams, CFA ■ Tech is the largest Chinese import; localization a national priority. Technology 886 2 2715 6366 sector forms the largest part of China's import basket (21%) and semiconductors account [email protected] for 70% of all its tech imports. This is despite the fact that China's own production of Keon Han semiconductors has grown at a 20% CAGR since 2011. This heavy dependence on 82 2 3707 3740 technology imports has led to several policy initiatives since 2000 from the Chinese [email protected] government to help grow the domestic technology industry for multiple reasons including Kyna Wong national security, assured supply availability, and an interest to continue economic and 852 2101 6950 human capital advancement growing capability in higher value add / IP areas. Rising [email protected] geopolitical issues have added urgency to these efforts. Chaolien Tseng ■ A deep-dive report. In this in-depth report, leveraging our global technology analyst team, 852 2101 6795 we outline the current market structure of various key tech sub-sectors, highlight China's [email protected] game plan to localise production and then provide our conclusions about the likelihood of Jerry Su China's success in these sectors. 886 2 2715 6361 ■ China localization: Some success; but some areas remain a work in progress. [email protected] China has achieved a lot of success in telecom equipment, hardware , Clive Cheung display, several key components and some success stories in IC design mainly in mobile 852 2101 7069 and consumer. It is devoting a lot of resources in localising semi production and design, but [email protected] so far with only modest success outside (and is likely to remain modest even in the medium term). Its dependence on imported equipment and certain key materials is likely to Please see sub-sectors for all contributing analysts remain unchanged over the medium term as well. While this report is meant more as a primer on the topic laying out the developments in key tech sectors and outlining our conclusions, on pages 4 and 5 we summarise the global tech supply chain map, indicate our estimate of China's chances of success in various areas and its likely leaders.

DISCLOSURE APPENDIX AT THE BACK OF THIS REPORT CONTAINS IMPORTANT DISCLOSURES, ANALYST CERTIFICATIONS, LEGAL ENTITY DISCLOSURE AND THE STATUS OF NON-US ANALYSTS. US Disclosure: Credit Suisse does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the Firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision.

4 November 2019

Focus charts

Figure 2: China total imports breakdown, 2018 (%) Figure 3: China semi industry revs and YoY growth Food & Live Beverages & Animals, 3% Other 1,600 1,487 40% Tobacco, 0% Manufactuered Goods, 10% 1,400 35% Non Food Raw 1,221 Materials, 13% 1,200 30% Machinery & 996 Transport 1,000 25% Equipment, 18% 843 800 707 20% Mineral Fuels & 600 Lubricants, 16% 600 15% 466 382 400 305 321 10% 255 Animal & Vegetable 200 142 5% Oils, 0% 0 0% Chemicals, 10% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Tech, 21% Textile, Rubber, China Semi Sales (Rmb bn) YoY（%） Minerals, 7%

Source: CEIC, Credit Suisse estimates Source: CEIC

Figure 4: Technology supply chain – a map of Chinese suppliers Semi Semi Cap Components Materials Equipment TV Brands Substrate: , Exposure: SMEE Panel: BOE, Tianma, CEC Panda, Fastprint, SCC Deposition: Naura, CSOT TV: , Wafer: AST, Ferrotech Casing: BYDE, Ju Teng, Tongda, FII Piotech , , JV, ESWIN, Simgui, Inspection: Grand, Touch/Fingerprint: O-Film, Truly, Q- Zing Semi, Zhonghuan Tech , Konka, Raintree TCL Lithography related: Etch: AMEC, Naura Lens: , AAC Kempur, Ruihong Photoresist processing: Acoustic: , AAC Sputtering: KFMI Kingsemi Connector: Luxshare, Everwin, FIT CMP Slurry: Anji Clean: ACM Surveillance: , Dahua Brands Chemicals: Runma, Antenna: Luxshare, Jingrui, Huayi, Sinyang LEDs: Sanan Handsets: Huawei, , Vivo, , Back-end ZTE, , Tech Foundries package & test , TCL CM / ODMs Downstream SMIC, Hua Hong, JCET, Tianshui Distributors Huali, Huatian, Tongfu, ASMC, CSMC, SiEN, China Wafer Level Synnex ( Huaqin, Wingtech, listed), Digital Silan, CanSemi, CR CSP, SJ Semi, TINNO, BYD, PC Brands Micro, Yantai Raytron Kaifa, Biwin Longcheer, USI China, VST PC / Server: Fabless IC Design Lenovo, , : HiSilicon, Spreadtrum, ASR, Pinecone, ZTE (Sanechip) Dawning, Razer, Servers/AI: HiSilicon, Cambricon, Montage, Alibaba, Eeasy Tech, Huaxinton, Big IC Distributors Huawei Fish, ThinkForce, Illuvatar, Cambricon, Bitmain, , Loongsan, eBang End GPU: Jingjia, Zhoaxin Networking Consumer MCU: Gigadevice, Ingenic, Unigroup Guoxin, SinoWealth, Silan, Goodix, Datang, HQ Mart, Will Semi, Fortune Huada, Giantec, Yixin, MindMotion, Winner Micro Tech, Wuhan P&S, INtron, OEMs RF IC: RDA, Vanchip, Huawei, Maxscend, Sanechips Apex Ace, Smart-core, Retail: Consumer: , Allwinner, , Actions Networking: Techtronics, Wisewheel, Gome,Suning Touch/Fingerprint IC: Goodix, Silead, Fortsense, Betterlife, Chipone, BYD Huawei, ZTE CMOS image sensor: Will Semi (OVT), GalaxyCore, Superpix, Artivision Sunray Online: Alibaba, Analog: Awinic, SG Micro, BYD , OnBright, Silergy, Nexperia, Huada, Chipone JD.com Driver IC: SinoWealth, Solomon Systech, Chipone Smartcard: Tongfong Guoxin, Datang, Huada, Nationz, Fudan, Hua Hong IC Memory: GigaDevice, YMTC, Changxin, Fujian Jinhua, Reliance Memory (Rambus/Giga JV), ISSI, Giantec Surveillance/Video: Huawei, Fullhan, Vimicro, Ingenic, Artosyn, Goke, Eeasy, EDA Tools Dahua, Yitu, Horizon Robotics FPGA: Gowin, Fudan Micro, Unigroup Guoxin, Huada Hyperform

Source: Company data, Credit Suisse Research

Asia Technology Strategy 2

4 November 2019

China: Can it gain tech independence? A net tech exporter; but a heavy semi importer

Import of technology products constitutes the largest proportion of total imports by China for any Tech imports are ~21% of total Chinese imports… given sector. This proportion has grown from high-teens a few years ago to 20-23% in recent years. Within technology, import of semiconductors by far constituted the largest proportion (US$311 bn; ~70% of total tech imports). Import of semis and display equipment was another important part of total imports (5%). While China is a large importer of technology products, it is … Semis are ~70% of all tech imports also a large exporter, actually exporting US$20 bn more of tech products than it imported in 2018—it is a large net importer of semiconductors and equipment but is a large net exporter of downstream products such as handsets and , and also telecom equipment. Localising semi production a national priority

While the Chinese local semis production has been growing at a 20% CAGR in recent years, Several policy initiatives since 2000 supporting the development of a local tech industry China remains a large importer of semis, and hence, the need to continue to focus on developing the local industry. Geopolitical developments in recent years, particularly the inclusion of several Chinese entities (notably Huawei, Hikvision amongst others) by the US on its restricted Entity List, further adds urgency to this localisation drive. The government has had Support from multiple sources: National IC Fund, policy initiatives, R&D and capex several policy initiatives over the years culminating in its '' initiative in 2015. subsidies, local government, and investments Substantial funding is behind the National IC Development Guideline initiative, including from private VC establishing the National IC Fund which raised US$20 bn in 2014 and is now a second round for US$29bn, along with US$120-140 bn from public/private funds. The fund has invested in creating national champions that can compete globally across memory, foundry, back-end, suppliers, and design companies. To supplement the National IC Fund, the Chinese government believes a strong policy support, growing ecosystem, and national fund backing in the will attract an even greater amount of investment from the private sector, financial institutions, and overseas investors. Local private VCs are also working in conjunction with the government and the National IC Fund. is the largest fund and is focused on establishing memory companies in China. China localisation: A mixed success story China has already shown success with the emergence of companies across the tech supply China is still far off from closing the gap on chain, with some notable ones in hardware and components. The presence of a large and semi manufacturing at leading edge (both memory and logic) and would likely still be growing hardware sector is pushing the national interest to also develop its semiconductor importing most of its semi equipment in the sector for economic benefits moving into higher value areas and lessening requirement on foreseeable future imports. A locally strong semiconductor sector also helps China's national security interest in controlling its information technology infrastructure. While we discuss in greater detail the dynamics at play in each of the tech subsectors later in the report, our conclusions for each of China will increase its dominance in display Memory semiconductors (TFT-LCD) and networking and make some these sectors are as follows: (1) : China still has some distance to progress in servers and OLED panels go before tasting any success, (2) Logic semiconductors: China has pockets of strength in IC design, backend and mature foundry nodes, but still lags in several areas, (3) Semi equipment and wafers: China is lagging in wafers and far behind in equipment and is likely to China will continue to localise a higher remain so, (4) Enterprise and servers: Chinese vendors strong in networking; server proportion of components, though would still be dependent on some imported parts over the expansion internationally may face challenge, (5) Display: China to dominate TFT panels; may medium term succeed in OLED but still lag key tools and raw materials; Korea will largely leave TFT space, (6) Components: China is largely self-reliant. Within specific components, our conclusions are: (a) Acoustics: Most parts likely to be localised, (b) Antenna: Self-reliant, work in progress to move to more advanced designs, (c) Casing: Lagging in metal casing but a leader elsewhere; still dependent on casing equipment imports, (d) Camera Module: Largely self-reliant; will increase global share, (e) CMOS Image Sensor (CIS): A credible Chinese player now, and hence, can be self-reliant, (f) Lens: Largely self-reliant, (g) MLCC: Chinese are lagging and likely to do so in the medium term too. On pages 4 and 5, we summarise the global tech supply chain map, indicate our estimate of China's chances of success in various areas and highlight Chinese companies (in bold) that are either the current or the likely future leaders within their respective sectors.

Asia Technology Strategy 3

42019 November

AsiaTechnology Strategy Global tech supply chain

Figure 5: Tech Supply Chain – China vs World Chances of China's China Supply Chain* US - Key Suppliers Non-US Key Suppliers success in 5 years

Semiconductor EUR: ASML Lithography SMEE JP: Nikon, Canon KR: Wonik IPS, TES, Eugene Tech, PSK

Deposition Naura, Piotech , JP: Semi Cap KR: Unitest, Techwing Inspection Limited Grand, Raintree KLA-Tencor, Applied Materials Equipment JP: Hitachi Hightechnologies KR: Wonik IPS, TES Etch AMEC, Naura Lam Research, Applied Materials JP: Tokyo Electron, Hitachi Hightechnologies Photoresist Processing Kingsemi JP: Tokyo Electron, SCREEN Clean ACM Lam Research, Applied Materials JP: SCREEN, Tokyo Electron, Shibaura Mechatronics Substrate Access, Fastprint, SCC KR: SEMCO, LGI KR: SK Siltron Wafer NSIG (?), AST, Ferrotech JV, ESWIN, Simgui, Zhonghuan TW: GlobalWafers, Formosa SUMCO, Wafer Works JP: Shin-Etsu Chemical, SUMCO, Ferrotec, RS Technologies KR: Dongjin Semi Lithography Kempur, Ruihong Semi Materials Limited JP: JSR, Tokyo Ohka Kogyo Sputtering KFMI KR: Soulbrain, KC Tech CMP Slurry Anji JP: Fujimi Incorporated Chemicals Runma, Jinrui, Huayi, Sinyang Photronics KR: SK Materials, Soulbrain, Wonik Materials, DNF EDA Tools Minimal Hyperform , HiSilicon, , Pinecone (Xiaomi), ZTE (Sanechip), Allwinner, Rockchip, Leadcore, KR: Mobile Processor High ASR Micro TW: Mediatek Low (servers); HiSilicon, Cambricon, Montage, Alibaba, Eeasy Tech, Huaxinton, Big Fish, ThinkForce, Servers/AI , AMD, ARM (certain technologies) EUR: ARM Medium (AI) Illuvatar, Bitmain, Zhaoxin, Loongsan, eBang GPU Limited Jingjia, Zhoaxin , AMD

Gigadevice, Ingenic, Unigroup Guoxin, SinoWealth, Silan, Goodix, Datang, Huada, EUR: NXP, Infineon,STMicroelectronics MCU Medium Microchip, TI Giantec, Yixin, MindMotion, Winner Micro (Fragmented market) JP: Renesas

TW: Win Semi, VPEC, AWSC, Richwave, Mediatek (Airoha), ACX RF IC, RF Front End Medium Vanchip, HiSilicon, Maxcend, Sanechip (ZTE), UniSOC, Will Semi, Espressif Systems Qorvo, Skyworks, Qualcomm, Broadcom JP: Murata, Taiyo Yuden

Consumer High Rockchip, Allwinner, Amlogic, Actions, UniSOC, HiSilicon (Fragmented market) AMBA (Ambarella), Microchip, Cypress Semi TW: Mediatek, Fabless IC Design Touch/FP IC High Goodix, Silead, Fortsense, Betterlife, Chipone, BYD Microelectronics TW: Egis, Elan KR: Samsung, SK Hynix, PixelPlus CMOS Image Sensor High Will Semi (OVT), Galaxy Core, Superpix, Artivision, BYD Microelectronics, SmartSens ON Semi, ST Micro EUR: Melexis JP: , , Canon EUR: STMicroelectronics, Infineon TI, , Maxim Integrated, ON Semi, Cirrus Logic, Analog / Discrete Medium HiSilicon, Awinic, SG Micro, BYD, OnBright, Silergy, Nexperia, Huada, Chipone JP: Mitsubishai, Rohm, Vishay TW: Mediatek (Richtek)

Smartcard High Tongfong Guoxin, Datang, Huada, Nationz, Fudan, Hua Hong IC (Fragmented market)

HiSilicon, Fullhan, Vimicro, Ingenic, Artosyn, Goke, Eeasy, Dahua, Yitu, Horizon Surveillance High ON Semi Robotics Gowin, Fudan Micro, Unigroup Guoxin, Huada, Anlogic, Hercules Micro (leadership FPGA Medium , , Intel, , Microchip undecided) Low (leading edge); SMIC, Hua Hong, Shanghai Huali, ASMC, CSMC, SiEN, Silan, CanSemi, CR Micro, Foundries GlobalFoundries, IBM TW: TSMC, UMC, Vanguard High (mature nodes) Yantai Raytron TW: ASE, King Yuan, SPIL, Powertech, ChipMOS, Chipbond, KYEC, Back-end High JCET, Tongfu, Tianshui Huatian, China Wafer Level CSP, SJ Semi, Kaifa, Biwin Amkor, R&D Altanova Kingpak, CHPT (probe) package & test KR: Unitest, Techwing, Lbsemicon KR: Samsung, SK Hynix YMTC, CXMT, Changxin / Gigadevice, Fujian Jinhua, Reliance Memory (Rambus/Giga Memory Limited Micron, Intel, TW: Nanya Tech, Macronix, JV), ISSI, Ginatec JP: (ex-Toshiba Memory) Note: * = Names in bold are either already or likely to be leader in supply chain in 5 years’ time

Source: Company data, Credit Suisse Research

4

42019 November

AsiaTechnology Strategy Figure 6: Tech Supply Chain – China vs World (Con’t) Chances of China's China Supply Chain* US - Key Suppliers Non-US Key Suppliers success in 5 years Others KR: Samsung, LGD Panel High BOE, Tianma, CEC Panda, CSOT, Truly, Visionox, EDO, , Olightek TW: AUO, Innolux, HannStar JP: Sharp, Japan Display, JOLED

TW: /FII/FIH, Casetek, Catcher, Everwin, Ju Teng Casing High Lens, Biel Crystal, BYDE, AAC, Everwin, Tongda, CN Innovations KR: Khvatec Touch / Fingerprint High O-Film, Truly, Q-tech, Speed, Holitech KR: Partron, Mcnex, Dreamtech, Crucialtec

KR: SEMCO,LG Innotek, JAHWA, Patron, Mcnex, Powerlogics, Cammsys ON Semi, Omnivision, Lumentum, Finisar, Vertilite, Viavi, , CCM High Sunny Optical, O-Film, Q Tech, Truly, Luxvision, Holitech, Shine, SZSeasons TW: Largan, Genius, PTOT IIVI JP: Kantatsu, Mitsumi, TDK, Alps, Shicoh, Sharp

TW: Largan, Genius Lens High Sunny Optical, AAC (?), O-film, Star JuYu, LceOptic, Huaxin KR: SEMCO, Sekonix, Elcomtec, Kolen JP: Kantatsu Components EUR: STMicroelectronics, Bosch KR: EM-Tech, Bluecom Acoustic High Goertek, AAC, Luxshare, NeoMEMS, MEMSensing, Speed Knowles, InvenSense TW: Merry, Inventec JP: Hosiden JP: Hirose, JAE, Kyocera, Panasonic, Daiichi Seiko, Iriso Connector High Luxshare, Sunway, Everwin TE Connectivity, Amphehol, Molex TW: FIT Surveillance High Hikvision, Dahua, Uniview TW: Geovision EUR: Ethertronics KR: Partron, Wisol, SEMCO Antenna High Sunway, Speed, AAC, Luxshare, JESONcom, Deman Amphenol, Skycross, Sky-Wave TW: Auden JP: Sumitomo, Murata KR: SEMCO / Samwha Low (high-end); Fenghua (?), Chaozhou Three Circle (?), Torch, Hongyuan, Sinocera, Sunlord, MLCC AVX, Kemet, Vishay TW: Yageo, Walsin Tech High (commodity end) Hongda, Tianli JP: Murata, TDK, Taiyo Yuden, Kyocera

KR: SEMCO, BH, Korea Circuit, Daeduck, Interflex TW: Taimide, Wus, Mortech, Taiflex, Thinflex, AEM, Avary, Flexium, PCB High Shengyi, Shennan Circuit, MFLEX, Victory Du Pont, TTM Career JP: UBE, Mitsubishi Gas, NOK, Fujikura, SEI, Ibiden, Kyocera

KR: APS, SFA Engineering, Wonik IPS, Viatron Equipment Low Naura, Han's Laser, Dalian Zhiyun, Liande JP: Canon, Hirata Corporationi, Tokyo Electron, SCREEN, ULVAC, V- technology, Shibaura-Mechatronics

KR: Duksan Neolux, LG Chem, Samsung, SDI, KH Vatec, SKC Kolon PI Display Jilin Optical, RuiYuan, Puyang Huicheng, Valiant, Selen, Eternal Material, Kangdexin, Materials / Components Low to medium Corning, Universal Display, Dow Chemical, Merck TW: TPK, Cheng Mei, Material, GIS Dongxu Opto, CaiHong Opto JP: New Nippon Steel, Idemitsu Kosan, Nitto Denko, Sumitomo Chem TW: Novatek, FocalTech, , Ilitek, Fitipower, Raydium Driver IC Low to medium SinoWealth, Solomon Systech, Chipone KR: Siliconworks, Magnachip, Samsung LSI EU: Osram LEDs High Sanan, HC Semitek, Canyang Opto, Changelight Cree KR: Seoul SemIConductor, Samsung, LG Innotek TW: Epistar, Lextar JP: Nichia, Osram, Toyoda Gosei ODMs / EMS High Huaqin, Wingtech, TINNO, BYDE, Longcheer, USI Flextronics, Jabil Circuit TW: Compal, Hon Hai, Inventec, , Quanta, KR: Samsung, LGE Smartphone High Huawei, Oppo, Vivo, Xiaomi, ZTE, Coolpad, Lenovo, TCL, Apple JP: Sony, Sharp TW: Acer, , MSI PC High Lenovo, Haier, , Huawei (?), Tongfang, Xiaomi (?) HP, , Apple, KR: Samsung, LG JP: Sharp Enterprise / Huawei, Great Wall, H3C, Hikvision, Inspur, Lenovo, Powerleader, Tongfang, , TW: Inventec, Quanta/QCT, Wistron/Wiwynn Medium HPE, Dell, Supermicro, Cisco Server ZTE JP: Networking OEMs High Huawei, H3C, ZTE, TP-Link Cisco, Arista, HPE, Juniper, Extreme KR: Samsung Note: * = Names in bold are either already or likely to be leader in supply chain in 5 years’ time Source: Company data, Credit Suisse Research

5

4 November 2019

Table of Contents

Focus charts 2

China: Can it gain tech independence? 3 A net tech exporter; but a heavy semi importer ...... 3 Localising semi production a national priority ...... 3 China localisation: A mixed success story ...... 3

A net tech exporter; but a heavy semi importer 8 China is by now a net tech exporter ...... 10

Localising semi production a national priority 12 Chinese IC industry policy timeline ...... 13 Made in China 2025: Semis and IT sectors the focus ...... 14 Funding of semis localisation initiative ...... 15

China localisation: A mixed success story 18 Memory semiconductors – still some distance to go ...... 18 Logic semiconductors: pockets of strength in IC Design, backend and mature foundry nodes...... 19 Semi equipment and wafers: lagging in wafers and far behind in equipment ...... 21 Enterprise and servers: Chinese vendors strong in networking; server expansion internationally may face challenges ...... 21 Display: China to dominate TFT panels; may succeed in OLED but still lags in key tools and raw materials ...... 22 Components: largely self-reliant ...... 24

Memory semiconductors 27 Current market structure ...... 27 China's game plan ...... 29 Conclusion ...... 34

Logic Semiconductors 37 IC design: China’s presence rising in consumer and mobile, still lags in higher value areas 37 Foundry: China lags on advanced capacity, more competitive on the mature nodes ...... 42 Back-end: China has used M&A to gain a stronger presence ...... 48 Funding of semi ambitions ...... 50 Conclusion: China’s presence lagging in foundry, rising in OSAT, varying in IC design ...... 54

Semi equipment and wafers 59 China's game plan ...... 61 Conclusion: Lagging in wafers and far behind in equipment...... 61

Asia Technology Strategy 6

4 November 2019

Enterprise and servers 63 Current market structure ...... 63 China's game plan ...... 67 Conclusion ...... 69

Display 71 Current market structure ...... 71 China game plan ...... 74 Conclusion: Chinese display makers becoming more meaningful but still lag key tools and raw material ...... 79

Components 84 Acoustic ...... 85 Antenna ...... 89 Casing ...... 91 Camera module ...... 95 CMOS Image sensor (CIS) ...... 101 Lens ...... 103 MLCC ...... 106 PCB/substrate ...... 108

Asia Technology Strategy 7

4 November 2019

A net tech exporter; but a heavy semi importer

Import of technology products constitutes the largest proportion of total imports by China for any given sector. This proportion has grown from high-teens a few years ago to 20-23% in recent years.

Figure 7: China tech imports as percentage of total imports

24.0% 23% 23.0% 22%

22.0% 22% 21% 21.0% 20% 20.0% 20% 19% 19% 19.0% 18%

18.0% 19% 18%

17.0% 17% 16.0%

15.0% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Source: CEIC Technology import totaled up to US$449 bn in 2018, growing 19% YoY. Tech remains the largest part of the total China imports pie. Machinery & transport equipment, Oil and other fuels, and non-food raw materials constituted the other major sectors of imports. Within technology, import of semiconductors by far constituted the largest proportion (~70%) of imports, with memory semi totaling up to US$122 bn (~27% of total tech imports) and other semis totaling up to US$189 bn (~42% of total tech imports) in 2018. Import of equipment— to produce semis and display parts—was another important part of total imports (US$23 bn; 5% of total). China also imports a large part of its telecom equipment (US$41 bn in 2018; 9% of total tech imports), while at the same time is a large telecom exporter to the world, highlighting the interdependencies of the global technology sector.

Figure 8: China total imports breakdown, 2018 (%) Figure 9: Total China tech imports breakdown, 2018 (%) Food & Live Others, 10% SPE, 3% Beverages & Animals, 3% Other Display Equip, 2% Tobacco, 0% Manufactuered Passive Goods, 10% Component, 3% Non Food Raw Printed Circuits, 3% Materials, 13% Machinery & Display, 0% Transport Semi-memory, 27% Equipment, 18% Telecom Equipment, 9% Mineral Fuels & Lubricants, 16% Mobile Handset, 0%

Animal & Vegetable Oils, 0%

Chemicals, 10% Tech, 21% Semi/IC ex memory, Textile, Rubber, 42% Minerals, 7%

Source: CEIC, Credit Suisse estimates Source: CEIC, Credit Suisse estimates

Asia Technology Strategy 8

4 November 2019

A rise in memory ASPs over the past two years, and strong growth in the market share for Chinese smartphone brands such as Huawei (which in turn consumes a lot of memory), led to a strong growth in import of memory semiconductors over the past two years—from US$64 bn in 2016 to US$122 bn in 2018. A combination of subdued growth in computing markets and a higher localisation of non-memory semis (see more details later in the report under the logic semiconductor section), non-memory semi import has been flattish in recent years—ranging from US$172 bn to US$192 bn in the past five years. Importantly, China's push to create a domestic semi and display production industry has meant that equipment imports have continued to grow rapidly from just US$8 bn in 2014 to US$23 bn in 2018.

Figure 10: Equipment, Memory and Other Semis – China import growth (YoY %)

70 YoY% 60 60

50 40 40 37 38 36

30 21 20 20 18 20 17 13 10 9 10 3 4 - (3) (10) (9) (8) (20) 2013 2014 2015 2016 2017 2018

Equipment Semi-memory Semi/IC ex memory

Source: CEIC, Credit Suisse estimates While China is a large importer of technology products, it is also a large exporter of technology products, actually exporting US$20 bn more of tech products than it imported in 2018 (more on that later). Technology exports (US$469 bn) constituted 19% of total Chinese exports in 2018, growing 11% YoY. Exports of machinery and transport equipment form ~30% of total Chinese exports with other manufactured goods accounting for another 23% of total exports. Within technology export, export of mobile handsets (US$141 bn) constitutes the largest part of the pie, followed by exports of other consumer electronics products (US$99 bn) and telecom equipment (US$77 bn). Growth in Huawei's (and also other Chinese brands) global market share has driven growth in export of handsets over the past two years, though share gains are starting to slow, given the already high share for Chinese brands, and incrementally, as the full impact of the Huawei's inclusion in the US Entity List starts to show.

Asia Technology Strategy 9

4 November 2019

Figure 11: China total exports breakdown (as of 2018) Figure 12: Mobile Handset, Telecom Equipment and Others – China import growth (YoY %)

Tech, 19% 25 YoY% 21

Textile, Rubber, 20 Minerals, 16% 17

14 15 13 10 11 9 Chemicals, 7% 10 8 7 Mineral Fuels & 4 Lubricants, 2% 5 3 Non Food Raw - Machinery & Materials, 1% Transport Food & Live (2) (2) (1) Equipment, 30% Animals, 3% (5) (3) (4) (4) (7) Other (10) Manufactuered 2013 2014 2015 2016 2017 2018 Goods, 23% Mobile Handset Telecom Equipment Others

Source: CEIC, Credit Suisse estimates Source: CEIC, Credit Suisse estimates China is by now a net tech exporter

Over the years, China has become a net exporter of technology products, though that balance has been reducing since 2015.

Figure 13: China tech net imports (USD bn)

20.0 10.5 10.0 2.3 0.0 (2.4) (2.6) (10.0) (11.5) (20.0) (20.1) (30.0) (29.0) (40.0) (37.9)

(50.0) (46.0) (49.3) (47.6) (60.0)

(70.0) (66.4) 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Source: CEIC Looking at the US$20 bn net trade surplus in technology for China, it becomes obvious that China is a large net importer of semiconductors and equipment but is a large net exporter of downstream products such as handsets and consumer electronics and also telecom equipment. Given the above, it becomes obvious why China has been focusing on growing its local semiconductor industry.

Asia Technology Strategy 10

4 November 2019

Figure 14: China tech net trade (imports minus exports) waterfall, 2018 (USD bn) 250 126

200

150

80 100 ( 141 ) 50 ( 36 ) 8 11 12 ( 25 ) ( 3 ) - (52) (20) (50) SPE Display Semi- Semi/IC ex Mobile Telecom Display Printed Passive Others Total Tech Equip Memory Memory Handset Equipment Circuits Component Net Imports

Source: CEIC, Credit Suisse estimates Telecom equipment is one high technology area where China has made significant progress over the past decade and has closed its technology gap with other global leaders, largely courtesy of Huawei and ZTE. This also provides some context to the friction that we are seeing between the US and China regarding Chinese equipment companies on the world stage. It is also worth noting, that despite China's significant success in telecom equipment, it still imported about US$41 bn of telecom equipment in 2018, underlining the point that it still depends on imported technology even in areas where it has made significant progress, to be able to even export US$77 bn worth of product. In semis, China has made some progress in developing its domestic fabless industry, though is still dependent on importing the final manufactured product in several cases. It has also reduced its import dependence in some areas by encouraging global producers to set up manufacturing within China to serve both its local demand as well as use China as an export base. Most noticeably in recent years has been the setting up of memory plants by Samsung, Hynix, and Intel in China that has resulted in a strong CAGR of memory exports from China. However, given the continued growth of domestic tech demand as well as Chinese brand's rising share of end tech products in the global market has meant that China's demand for semiconductors has outpaced the growth of its domestic semis production and has increased demand for semi parts that it does not locally produce, resulting in larger net import of semis. It also noteworthy that the strongest CAGR in import is in the area of semi and display equipment, albeit the value of imports still remain small in the context of overall tech imports— necessary import to create a domestic manufacturing base.

Figure 15: China imports CAGR, 2010-15 and 2015-18 (%) Figure 16: China exports CAGR, 2010-15 and 2015-18 (%)

40% 30% 35% 34% 25% 35% 25% 22% 30% 20% 26% 20% 25% 15% 19% 11% 20% 9% 10% 10% 15% 12% 12% 4% 4% 5% 3% 2% 2% 7% -7% 10% 7% 6% 0% 5% 2% 1% 1% 0% -1% -5% 0% -10% -5% Semi- Semi ex Mobile Telco Others Total SPE Display Semi- Semi ex Telco Others Total memory memory Handset Equip Equip memory memory Equip 2010-15 2015-18 2010-15 2015-18

Source: CEIC, Credit Suisse estimates Source: CEIC, Credit Suisse estimates

Asia Technology Strategy 11

4 November 2019

Localising semi production a national priority

As highlighted in the previous section (and also reproduced in the chart below), 70% of all tech- related imports by China are of semiconductor products, thus, making it highly dependent on foreign-sourced parts. Several of these imported semiconductors are also meant for powering its technology related exports (handsets, telco equipment, and consumer electronics). China also imports most of its equipment required for semi and display manufacturing and currently has limited capability to manufacture any of these equipment locally. Thus, not surprisingly, the focus of Chinese authorities, in the past decade or more, has been to develop a local semiconductor industry.

Figure 17: Total China tech imports breakdown, 2018 (%)

Others, 10% SPE, 3% Display Equip, 2% Passive Component, 3% Printed Circuits, 3% Display, 0% Semi-memory, 27% Telecom Equipment, 9%

Mobile Handset, 0%

Semi/IC ex memory, 42%

Source: CEIC, Credit Suisse estimates While the focus on developing a domestic Semi industry has borne some fruit – local semis production has been growing at a 20% CAGR in recent years (including semi production by global players out of their China-based facilities) – China, as discussed in the previous section, remains a large importer of semiconductors, and hence, the need to continue to focus on developing the local industry. Geopolitical developments in recent years, particularly the inclusion of several Chinese entities (notably Huawei, Hikvision amongst others) by the US on its restricted Entity List, further adds urgency to China's initiatives to localise the tech industry and reduce its import dependence.

Figure 18: China semi industry revs and YoY growth

1,600 1,487 40%

1,400 35% 1,221 1,200 30% 996 1,000 25% 843 800 707 20% 600 600 15% 466 382 400 305 321 10% 255 200 142 5%

0 0% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

China Semi Sales (Rmb bn) YoY（%）

Source: CEIC

Asia Technology Strategy 12

4 November 2019

Chinese IC industry policy timeline

The government has had several policies before, starting notably with State Council Document 18 to develop its IC industry in 2000. This policy offered favourable tax treatment for domestic IC chips and government investment in infrastructure, education, and basic research. In 2005, the US petitioned the WTO to drop the VAT rebates for China IC producers, however, some of the favourable industry policies stayed in effect, with the full document in effect through the end of 2010. The policy did lay the ground work for the domestic industry, building it up to close to 10% of industry production by the financial crisis and US$23 bn of sales, according to CCID.

Figure 19: China IC industry policies progress since 2000

Source: SEMI

China's State Council supplemented the policy with its guidelines on Scientific Technology China's policy target in 2006 was to ramp R&D and technology investments development in 2006 for the next 15 years. It set a target for R&D at 2.5% of China's GDP, with interim targets by 2020 to increase science and technology to 60% of the country's development and lowered reliance on foreign technology to less than 30%. The policy supported a Science and Technology development programme with projects to develop core devices, high-end chips, equipment and materials, China prioritised technology development and set targets to reach Top 5 in patents. The China government passed the National IC Development Guideline in 2014 and ‘Made in China 2025’ initiatives in 2015, jumpstarting the latest wave of industry development to build the local semiconductor industry, this time directing more public and private resources for the initiative. Key outcome from the guidelines is to achieve 20% of semiconductor industry revenue growth to US$143 bn by 2020 (Rmb870 bn), raising China's internal supply of silicon from one-third to half to reduce import dependence and stimulate the economy to move up the value chain from manual labour intensive to high-skilled technology intensive industries. The national policy laid out aggressive plans behind its 20% growth rate to achieve global competitiveness across manufacturing, fabless IC design, back-end package and test, materials and equipment. By 2020, the target is to move up from one third to half of its chips from domestic production, reach competitiveness in a number of advanced silicon capabilities (IoT, networking, cloud computing, big data), enter FinFet mass production and have back-end, equipment and materials solutions to be competitive enough to serve the global supply chain. Ultimately by 2030, China's target is to have a world class IC value chain with a set of top-tier global chip players.

Asia Technology Strategy 13

4 November 2019

Figure 20: China's 2014 National IC Guideline set aggressive targets for its chip industry 2015 2020 2030 Semiconductor Revenue >350bn RMB (US$55bn) >870bn RMB (20% Growth CAGR) World class IC industry value chain IC Manufacturing 32/28nm mass production 16/14nm mass production A set of leading tier 1 global semi players Reach international leading edge in key Approach international standards in some technologies (mobile, networking, cloud IC Design technologies (, networking) computing, IoT, big data) IC Package and Test >30% of sales from advanced packaging Competitive with the global leaders IC Materials 12" silicon wafers in the production line Competitive in the global supply chain Semi. Equipment 65-45nm tools in the production line Competitive in the global supply chain Source: SEMI, China's National Guideline 2014, Credit Suisse research Made in China 2025: Semis and IT sectors the focus

The Chinese government in May 2015 also laid out the first of three 10-year plans to transform China into a leading high-quality manufacturing powerhouse by 2049, transitioning the country from a volume-based labour intensive chain into a high-tech integrated manufacturing base with leading global innovation. The country is looking to follow Germany's Industry 4.0 push to move towards more automated, efficient, networked, and connected manufacturing that also integrates production, supply chain, and customers. The first ten-year initiative in the Made-in-China 2025 campaign targets strengthening China's Made in China 2025 targets industrial base to enhance the quality of the manufacturing, foster Chinese brands and improve transforming industry to a manufacturing innovation. The government prioritised ten sectors although semiconductors and high-end integrated value information technology are key focus areas in the programme. To promote the Made-in-China chain 2025 campaign, the MIIT will facilitate the industry consolidation to allocate the resource more efficiently and encourage the financial support from local government and private funds on the local semiconductor ecosystem development. The plan will combine some state planning for the framework but also market forces including IP protection and industry standard creation and broader push to participate in international standards. Localisation is another goal of the programme, with a plan to raise the domestic content of core components and materials to 40% by 2020 and 70% by 2025.

Figure 21: Improvement targets for Made in China 2025 Figure 22: Made in China 2025 domestic champions

Target Indicator 2015 2020 2025 Sector Target and focus areas Existing capability Information Technological Innovation, particularly Limited but key focus sector for creating R&D intensity as (% of sales) 0.95 1.26 1.68 Technology semiconductors leadership in technology Innovation Low and medium level automation Patents per 100m of core Numerical control tools capability. Focus on NC tool capability Already a leader in low and medium level NC 0.44 0.7 1.1 business revenues and robotics to improve manufacturing quality and tools productivity Manufacturing quality Focus on satellite capabilities and 83.5 84.5 85.5 One passenger jet C919-2,400 planes possible competitiveness index Aerospace equipment passenger jet transportation including by 2045 aircraft engines Quality Manufacturing value added NA 2% > 2015 4% > 2016 Offshore infrastructure and high-tech Existing investment and capabilities in South efficiency growth rate Marine engineering ships China Seas infrastructure Manufacturing labour productivity +/-7.5 (13th +/-6.5 (13th High speed rail equipment sales to Russia and NA Existing high speed train production. growth rate 5YP CAGR) 5YP CAGR) Malaysia. In discussions with over 20 countries Railway equipment "One Belt, One Road" projects to for potential high speed railway equipment sales improve competitiveness Broadband penetration (%) 50 70 82 including US Domestic environmental clean-up Existing environment clean-up and emission Integration of Energy efficiency and R&D digital design tool priorities and long established domestic reduction targets should support this segment. Industrialisation 58 72 84 electric vehicles penetration car manufacturing aspirations Visibility of innovation new fuel vehicles limited and IT Utilisation rate for numerical Core priorities of smart grid and smart 33 50 64 Power equipment Currently exporter of power equipment to controls in critical processes (%) city technologies

Energy consumption per unit of Key government focus on "invention" NA 18%<2015 34%<2015 New materials Research into new materials in its infancy industrial value added and "innovation" in materials C02 emissioins per unit of Rapid growth in medical equipment market of Existing success in low tech medical NA 22%<2015 40%<2015 Biopharma and medical 20% pa since 2009 from small base. Medical Environmental industrial value added devices. Likelihood of success in devices device capability currently small, low technology Improvements Water consumption per unit of biopharma and medicines less certain NA 23%<2015 41%<2015 equipment and supplies industrial value added Focus on improving quality in current Domestic industry concentrated on low power manufacturing and potential export Utilisation rate of solid industrial Agricultural machinery tractors and small to medium machinery. 85 73 79 sector. Potential for international Exports predominantly from JVs waste (%) demand less clear

Source: Refinitiv, Credit Suisse Global Strategy Research Source: Refinitiv, Credit Suisse Global Strategy Research

Asia Technology Strategy 14

4 November 2019

The Made in China programme ultimately targets making Chinese companies more competitive, production and components more localised and China firms higher up on the value chain moving towards International brand recognition and status. Funding of semis localisation initiative China National IC Fund established in August 2014 Substantial funding is behind the National IC Development Guideline initiative, including establishing the National IC Fund which raised US$20 bn in 2014 and now a second round for US$29 bn, along with US$120-140 bn from public/private funds. The fund has invested in creating national champions that can compete globally across the foundry, back-end, suppliers and design companies, with 60% of the funding to designated IC manufacturing due to the foundries central role in the ecosystem, and 40% for packaging & test, IC design, equipment, material, and local IC funds. Local Private Equity funds supplement the National IC fund To supplement the National IC Fund, the Chinese government believes a strong policy support, growing ecosystem and national fund backing in the semiconductor industry will attract an even greater amount of investment from private sectors, financial institutions, and overseas investors. The Chinese government expects 5x more investments beyond the National IC Fund. The Shanghai government has been aggressive industry development as it already has a relatively sound ecosystem of international manufacturers, local and overseas foundries, back-end and IC design. The IC fund that it established last year made its first acquisition target of ISSI in March to bring more memory IP into China. The first phase of the fund has been largely been deployed. SMIC and UMC fab expansions next year will use the second phase of the fund.

Figure 23: China semiconductor industry is supported by National IC Fund and increasing capital from private sectors China Fortune- E-Town National IC Fund Regional Funds Tsinghua Unigroup PDSTI SummitView Capital BJ Semi Fund Hua Capital Tech Capital Capital Start Date 2014 2014-2015 1988 1999 2014 2009 2011 2014 2014 - Supports China's - Shanghai $8.0bn, - Invests in semi and IC industry Beijing $4.7bn, - Focuses on - Supports China's - Supports - Supports China internet start-ups development Wuhan $4.7bn, - Build up the China emerging fabless - Focus on Ics, IT, semiconductor China's IC technology - The fund is co-set up Mandates - Encourages Sichuan $1.7bn, semiconductor IDM investments to manufacturing, development development with companies' with the Shanghai private sectors to $3.1bn, ecosystem expand SMIC's energy and medical through acquisitions development givernment mainly for invest in Wuxi $1.7bn, markets acquisitions semiconductor M&A semiconductors $150mn

RMB300bn RMB10bn RMB1.2bn RMB2bn RMB139bn RMB150bn RMB500mn RMB10bn (US$1.5bn) Fund Size (US$46bn) planned (US$1.5bn) for (US$180mn) for RMB4bn (US$310mn) for (US$21bn) (US$24bn) (US$76mn) for semiconductors in 5 years semiconductors semiconductors semiconductors

Stake investments: - Acquisition: - Investments: - Stake - Acquisition: - Shanghai with Spreadtrum, RDA, SMIC, Actions, investments: Montage - Stake ISSI, H3C Montage, SPRD, SMIC, ZTE Micro, - Stake investments: - Acquisition: - Acquisition: Investments Beijing/Shenzhen - Stake - Acquisition: ISSI Chipsbank, JCET investments: EtraSemi, CISTA, Mattson Omnivision with SMIC, Wuhan investments: TechFaith, CSMC, - Business deals: SMIC, Senodia with XMC, Nanjing Western Digital, 3 Waveguider, Fortune Tsinghua Unigroup Spreadtrum with TSMC Taiwan OSATs Techgroup, Sinosun Source: Company data, Credit Suisse estimates The Beijing IC Fund also has Rmb30 bn to acquire Omnivision and is developing a 28nm 12" JV fab with SMIC now at 17k WPM with an eventual target for 35K in the first of the three phases. The Shanghai government has also announced a JV fab with SMIC, and Ningbo is developing an R&D centre that could convert to a fab capacity focused on specialty manufacturing technology. The Nanjing government has also supported TSMC's 20k fab investment, Xiamen government formed a JV with UMC for a 50k WPM fab in that province, and Chonqing government has invested in the Globalfoundries fab in China.

Asia Technology Strategy 15

4 November 2019

Figure 24: China IC Focused Funds

Funds Amount (Rmb,bn) Amount (US$,bn) Source China Industry Investment Fund (CICIIF) 138.7 20.8 Central Government Tsinghua Unigroup 300.0 45.0 Private sectors/Central Government Beijing Semiconductor Industry Development Fund 30.0 4.5 Beijing Government Shanghai IC Fund 50.0 7.5 Shanghai Government Shanghai IC Investment Fund 28.5 4.3 Shanghai Shanghai Pudong Science and Technology Investment 10.0 1.5 Shanghai China Fortune-Tech Capital 0.5 0.1 SMIC Beijing E-Town Capital 20.0 3.0 Beijing Government/Private sectors BJ Semi Fund 4.0 0.6 Beijing Government/Private sectors Hua Capital 2.0 0.3 SMIC/Tsinghua Holdings Shenzhen IC Industry Fund 20.0 3.0 Shenzhen Government Beijing IC Overseas Fund 2.0 0.3 Beijing Xiamen Tsinghua IC development Fund 16.0 2.4 Xiamen, Tsinghua Unigroup Fujian Province Anxin Industry Fund 50.0 7.5 Fujian Province IC Fund 15.0 2.3 Guangdong Wuhan Industry Development Fund 30.0 4.5 Wuhan Sichuan Industry Investment Fund 10.0 1.5 Sichuan Nanjing Pukou IC Industry Fund 1.0 0.2 Nanjing Nanjing IC Industry Fund 50.0 7.5 Nanjing Hubei Investment Fund 30.0 4.5 Hubei Wuxi IC Industry Fund 20.0 3.0 Wuxi IC Industry Fund 200mn a year 30mn a year Tianjin Shijiazhuang IC Industry Fund 10.0 1.5 Shijiazhuang Shanxi IC Industry Fund 30.0 4.5 Shangxi Liaoning IC Industry Fund 10.0 1.5 Liaoning Summitview Capital 10.0 1.5 Private Other local government and PE 30-80bn 5-12bn Gov, Private, SOE etc. Total estimates: RMB850-950bn US$120-140bn

Source: Company data, Credit Suisse estimates Private VCs also an important form of capital Local private VCs are also working in conjunction with the government and National IC Fund. Tsinghua Unigroup is the largest fund and is focused on establishing memory companies in China. The fund had earlier invested in RDA/Spreadtrum, attracted Intel's investment in 2014 and expanded its acquisition targets into memory with a JV with ChipMOS in packaging, acquisition of XMC for NAND flash and development of a DRAM fab backed by local funds. Tshinghua has continued to invest in the Yangtze Memory Technology. Beijing ETown was involved in acquiring Mattson, a US-based smaller equipment company.

Figure 25: Top 10 Chinese semiconductor private placements: 2016-18

Source: S&P Global Market Intelligence

Asia Technology Strategy 16

4 November 2019

Other sources of funding and support Other than broad level and significant funds available from the National IC Fund, local private equity funds and private VCs, these semiconductor players in China also stand to gain funding and help from the following sources, and we discuss these in more detail later in the report under the logic semiconductors section: o Setting up joint ventures between fab/foundry operators and local governments o Providing R&D and equipment subsidies o Incentivising talent to move to China and improving local talent output; increasing local R&D spend o Easing access to capital markets Setting up localised standards to support industry development The Chinese government also believes that large domestic demand from the local market provides it an opportunity to support the local semiconductor industry by setting up localised industry standards. The government successfully promoted TD-LTE in the communications segment to build its domestic equipment industry early in the 3G era to pave the way for its more competitive position on 4G and now 5G. It now believes that it is in a leading position to drive standards in 5G development and has issued licenses and is targeting an aggressive domestic roll-out to build its local equipment and smartphone brands. It also set up a new standard in the financial cards market to support the local smartcard fabless in 2014.

Asia Technology Strategy 17

4 November 2019

China localisation: A mixed success story

China already has companies emerge across the tech supply chain, with more success in hardware and components. In tech hardware, China's smartphone brands have taken 55-60% of the global smartphone market, has leading PC brand Lenovo, networking suppliers Huawei and ZTE and leading regional TV brands in TCL, Changhong, Skyworth, Konka, Haier and Hisense. The component sector has also been seeing competition rise across acoustics, casing, lens/camera module, optical components and connectors. The country is also leading in EMS, with most of the industry's TV and notebook production in China. The presence of a large and growing hardware sector is pushing the national interest to also develop its semiconductor sector for economic benefits, moving into higher value areas and lessening requirement on import, and also national security controlling its processing, communication and storage links.

Figure 26: Technology supply chain – a map of Chinese suppliers Semi Semi Cap Components Materials Equipment TV Brands Substrate: Access, Exposure: SMEE Panel: BOE, Tianma, CEC Panda, Fastprint, SCC Deposition: Naura, CSOT TV: Skyworth, Wafer: AST, Ferrotech Casing: BYDE, Ju Teng, Tongda, FII Piotech Changhong, Haier, JV, ESWIN, Simgui, Inspection: Grand, Touch/Fingerprint: O-Film, Truly, Q- Zing Semi, Zhonghuan Tech Hisense, Konka, Raintree TCL Lithography related: Etch: AMEC, Naura Lens: Sunny Optical, AAC Kempur, Ruihong Photoresist processing: Acoustic: Goertek, AAC Sputtering: KFMI Kingsemi Connector: Luxshare, Everwin, FIT Smartphone CMP Slurry: Anji Clean: ACM Surveillance: Hikvision, Dahua Brands Chemicals: Runma, Antenna: Luxshare, Sunway Jingrui, Huayi, Sinyang LEDs: Sanan Handsets: Huawei, Oppo, Vivo, Xiaomi, Back-end ZTE, Coolpad, Tech Foundries package & test Lenovo, TCL CM / ODMs Downstream SMIC, Hua Hong, JCET, Tianshui Distributors Shanghai Huali, Huatian, Tongfu, ASMC, CSMC, SiEN, China Wafer Level Synnex (Taiwan Huaqin, Wingtech, listed), Digital Silan, CanSemi, CR CSP, SJ Semi, TINNO, BYD, PC Brands Micro, Yantai Raytron Kaifa, Biwin Longcheer, USI China, VST PC / Server: Fabless IC Design Lenovo, Inspur, Mobile Processor: HiSilicon, Spreadtrum, ASR, Pinecone, ZTE (Sanechip) Dawning, Razer, Servers/AI: HiSilicon, Cambricon, Montage, Alibaba, Eeasy Tech, Huaxinton, Big IC Distributors Huawei Fish, ThinkForce, Illuvatar, Cambricon, Bitmain, Zhaoxin, Loongsan, eBang End GPU: Jingjia, Zhoaxin Networking Consumer MCU: Gigadevice, Ingenic, Unigroup Guoxin, SinoWealth, Silan, Goodix, Datang, HQ Mart, Will Semi, Fortune Huada, Giantec, Yixin, MindMotion, Winner Micro Tech, Wuhan P&S, INtron, OEMs RF IC: RDA, Vanchip, Huawei, Maxscend, Sanechips Apex Ace, Smart-core, Retail: Consumer: Rockchip, Allwinner, Amlogic, Actions Networking: Techtronics, Wisewheel, Gome,Suning Touch/Fingerprint IC: Goodix, Silead, Fortsense, Betterlife, Chipone, BYD Huawei, ZTE CMOS image sensor: Will Semi (OVT), GalaxyCore, Superpix, Artivision Sunray Online: Alibaba, Analog: Awinic, SG Micro, BYD , OnBright, Silergy, Nexperia, Huada, Chipone JD.com Driver IC: SinoWealth, Solomon Systech, Chipone Smartcard: Tongfong Guoxin, Datang, Huada, Nationz, Fudan, Hua Hong IC Memory: GigaDevice, YMTC, Changxin, Fujian Jinhua, Reliance Memory (Rambus/Giga JV), ISSI, Giantec Surveillance/Video: Huawei, Fullhan, Vimicro, Ingenic, Artosyn, Goke, Eeasy, EDA Tools Dahua, Yitu, Horizon Robotics FPGA: Gowin, Fudan Micro, Unigroup Guoxin, Huada Hyperform

Source: Company data, Credit Suisse Research While we discuss in greater detail the dynamics at play in each of tech subsectors later in the report, we summarise our conclusions for each of these sectors in the following paragraphs. Memory semiconductors – still some distance to go

While there are many Chinese entities that have indicated interest to engage in production of memory semiconductors, we conclude that currently there are only two likely players that will pioneer the memory semi development in China: Changxin Memory Technologies (CXMT) for DRAM and Yangtze Memory Technologies (YMTC) for NAND. Overall, for both DRAM and NAND, we see very little supply impact from China by 2020.

Asia Technology Strategy 18

4 November 2019

In DRAM, while CXMT's first fab is designed up to 120K wafers based on announced 19nm technology plan, the initial ramp of 20K to 30K wpm has already been delayed by over a year. CXMT is going to provide output only on 19 nm (we cannot yet confirm if this is an industry accepted 19nm) by next year, but by mid-2020, 19 nm would not be competitive to industry leaders producing mostly on 1y nm and with Samsung potentially moving on to 1z nm. Also, within such a small production scale and lack of engineers, it is unlikely that CXMT can simultaneously work on various DRAM architectures such as mobile, PC, Server, etc. Most likely, CXMT, even if it is successful in commercially producing 18nm DRAM, will find itself competing in the consumer specialty DRAM segment which would require less advanced nodes and speed. Therefore, the Taiwanese consumer DRAM makers would be the first in line in terms of competitive threat, however, as we said, only if CXMT is successful. From a capacity planning perspective, 20K to 30K wpm on higher nodes does not represent overall DRAM industry supply threat. However, if CXMT were to be successful and ramp to its maximum 120K by end of 2022, then it could represent as high as ~8% of global DRAM wafer capacity, although likely significantly less in terms of bit shipment market share. While we believe CXMT remains the designated DRAM champion in China, in recent weeks, we have seen some signs, given the relatively slow progress in DRAM at CXMT that China may be looking to back another group (partly funded by Gigadevice – 603986.SS) as well for DRAM development. YMTC's 3D NAND mass production roadmap is less clear. Having 64L TLC design at this juncture of NAND industry development puts its technology far behind industry leaders who are already producing on 92L/96L and investing in 128L. The industry leaders are already dismantling 6xL capacity as any hope of return to profitability fades, given the sharp fall in ASPs over the past two years. This technology (6x layers) would have no cost advantage and has limited use in the end-market applications. Although YMTC has begun R&D of 128L NAND and targets risk production from 4Q20, given its history, we expect more time is needed. It may be working on the successful of 64L development for a while. While the long-term goal of ramping to 300K wpm has been indicated by the company, it would first need sizeable customer wins in order to produce NAND in that magnitude. YMTC has announced that it began sampling its NAND products with various client SSD makers but success of any qualification is unknown currently. Capacity ramp schedule would be difficult to project without confirmation of a commercially viable product. Logic semiconductors: pockets of strength in IC Design, backend and mature foundry nodes The Chinese semiconductor sector is seeing varying degrees of success establishing self- sufficiency in its domestic ecosystem due to high barriers in the advanced manufacturing and chip design. In foundry, market share has been stable around 9% the past decade, despite National IC fund investments and heavy attention on this sector as the centre of China’s industrial development. TSMC’s wide lead on process technology, and increasingly high-end advanced packaging, allows it to address the high-end compute, mobile and networking demand. The Chinese foundries have kept up with market growth, but largely by leveraging their more legacy 8” and mature 12” process nodes to address high-volume consumer and wireless/wired connectivity, and specialty high-volume applications (CMOS image sensors, power management, bank cards, specialty memory). In the back-end space, China has increased share following the M&A of overseas companies, and through a combination of aggressive pricing, targeting of SiP with high sales contribution but low margin, and addressing the growing base of local fabless and international companies with supply chains in China. The companies, following acquisition, can address the mainstream high-volume packaging—including flip chip and wafer level packaging—though lag foundry leader TSMC integrating high-end applications on its silicon interposers and high density fan-out process. Market share has been rising, and now approaching 20%, although profitability is still a concern with China players operating at low- to mid-single-digit operating margins. In IC design, traction has been steady, to lift share from 4% in 2010 to 16% in 2018, although it has been concentrated in select areas of the semiconductor space. Market share is the highest in mobile segment (processor, modem, wireless power management, connectivity,

Asia Technology Strategy 19

4 November 2019

CMOS sensors), wired connectivity (led by Huawei) and some new high-volume applications seeing strong support in China (crypto-currency and AI). Market share is still below 5% in approx 70% of the overall industry – most notably processors, memory, GPU, FPGA, analog, MCU and RF. Figure 27: China semiconductor industry continues to outgrow global peers

Semiconductor sales (US$mn) China market share (%) 300,000 30%

250,000 25%

200,000 20%

150,000 15%

100,000 10%

50,000 5%

0 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 China foundry China back-end China fabless Global foundry Global back-end Global fabless China foundry share China back-end Share China fabless share

Source: Company data, Credit Suisse estimates Likely growth areas over the next three years: 1. Continued growth in its core wireless-related semis. We expect China to have the most growth from its existing strong hold in wireless, through higher content in mobile processors, connectivity, image sensor and gradual traction in RF. The move to 5G networks should drive both content and volume growth. 2. AI could be a new high-volume category for fabless. Replicating their success in the crypto-currency ASIC market, we expect these companies to target AI inference and some cloud-acceleration applications, which could drive another revenue cycle addressing the evolution toward smarter products capable of voice recognition and image/object detection. 3. Growth from a low-base in analog, RF, and MCUs. China has an opportunity to target more mature mainstream analog and discretes, and MCUs, including at domestic companies looking for a local-source alternative to the global IDM. The presence of local foundry capacity with Hua Hong’s 8” and new 12” mature fab and back-end capacity should help. The Asian RF foundries like Win Semi and AWSC along with Sanan’s development should also help the emerging RF suppliers to target some of the discrete RF components and mature 2G-4G cellular layers and Wi-Fi RF. 4. Local foundries to grow on specialty applications/mature nodes. We expect China’s foundries to expand their position with the growing local IC design companies using their mature lines (power management, , CMOS sensor, digital consumer). Hua Hong’s expansion of its Wuxi fab would double its revenue base in the next five years on still-mature technology nodes, while SMIC should keep its mature lines relatively full with the growth of the domestic market. The advanced technology may, however, take more time. 5. China back-end takes a larger share of local customers. China’s back-end sector is seeing increasing opportunities as it leverages the more advanced technology capabilities picked up from its acquisitions to target domestic companies adding local sources. Profitability though, may still lag a bit due to on-going competition. 6. A few longer term (three-to-five years) possibilities: a. SMIC’s aggressive push on technology at the expense of near-term profitability, heavy government support, and drop-off of advanced technology competition from UMC and GF gives it a chance to emerge as a capacity source for more advanced applications. SMIC will need to endure high capex and difficult profitability but may emerge in three to five years as a more Asia Technology Strategy 20

4 November 2019

competitive foundry capable of supplying some of the more advanced mobile and networking applications. b. China’s design houses are working on alternatives to Intel/AMD CPUs using ARM and RISC-V processors and NVIDIA accelerators with a host of AI inference and training chips. Within three to five years, the push for a local source—for national security and economic reasons—could allow local suppliers to gain some traction. c. We would view the build-up of more local teams and domestic diversification as allowing local suppliers to gain further market share in broad markets in MCUs and analog. Semi equipment and wafers: lagging in wafers and far behind in equipment

There are two relevant Chinese equipment companies, but both with negligible shares currently: AMEC (Advanced Micro-Fabrication Equipment) which develops and manufactures dielectric and TSV etch tools for semiconductor manufacturers, and MOCVD tools for LED makers; and Naura (Naura Technology Group) which has been set up with a mission to establish China’s capabilities in various equipment fields and is developing a wide range of semiconductor equipment (including etch, PVD (physical vapour deposition, CVD (chemical vapour deposition), diffusion, cleaning tool, UV curve, indexer, gas measuring control, ALD (atomic layer deposition) etc.). For equipment, we estimate China players are at least 5-10 years behind global suppliers, or even longer for areas where there is no volume production by China companies so far. We expect the 5-10 years gap to continue in the foreseeable future, given the long development and complex qualification process required by chipmakers. That said, given China’s strong momentum in domestic replacement, we expect some China players to shorten their technology gaps vs global players (i.e., AMEC), while some may require an even longer time. Importantly, price of equipment is not a critical issue to change suppliers in the equipment market as production yield is quite an important parameter for semiconductor makers. Similarly, we think local Chinese wafer makers may not be able to close the technology gap with the wafer majors and significantly boost their market share over the next one to three years. Given also the size of the market shares commanded by Japanese, European, Taiwanese and Korean makers, we do not envisage any noteworthy change in market share distribution. Our view is based on the scale of the technological barriers. At this point, we have ascertained the following major developments, although in all three cases we understand there are unresolved technological issues. (1) Zingsemi is shipping 300mm test wafers to customers, but these are not ready for use in commercial mass production. (2) Having hired engineers from outside of Japan, RS Technologies has improved its 200mm wafers technology, but not enough to catch up to the five market leaders. (3) Ferrotec has increased the competitiveness of its product by licensing in technology from GlobalWafers, but appears to mainly ship test and prime wafers and to a few China chipmakers that are not particularly fussy about product quality. As China chipmakers get up and running, demand growth in the mature segment of the semiconductor market that does not require high-end wafers should provide business opportunities for China wafer-makers. Enterprise and servers: Chinese vendors strong in networking; server expansion internationally may face challenges

For the networking market, we expect to see Chinese vendors take more share over time, and China vendors will gain more shares long-term in Ethernet switch from the US vendors, with in the higher networking speeds of 100GB+ from the low level today (currently Chinese vendors also increasing pressure to the higher ports has 10% market share at this speed), and we expect to see more pressure on the market share speed segment of US-based vendors as political tensions persist. Where we see additional opportunities for Chinese vendors are European developed markets, where Germany recently announced that Asia Technology Strategy 21

4 November 2019 they will not ban Huawei from their 5G build-outs, giving Chinese vendors more global opportunities for the networking stack. For overall US switching, we expect to see Cisco, Juniper Networks, and HPE to continue dominating US share, with limited opportunity in China given the political tensions. Finally, given the degree of R&D investments and supply chain ramp taking place in China, we expect Chinese vendors to continue dominating in China driven by domestic government policy. China is already nearing self-sufficiency in servers from a hardware vendor perspective, with Chinese vendors accounting for >80% of the domestic market, per IDC. Given the government’s China First strategy and its intention to break away from the US supply chain, we expect China server vendors will continue to expand their dominant position domestically. On the international front, Chinese vendors’ share remains relatively small at ~6%—70% of which is by We expect China vendors to gain more shares Lenovo, who purchased IBM’s x86 business in 2015. Inspur has been making good progress in servers especially domestically with policy support; however, they may see more internationally, supported by its JDM business model and alignment to OCP standards; however, obstacles internationally from recent security we believe rising security concerns—especially after a Bloomberg report in mid-2018 and trade tension speculated on a hacking incident allegedly involving China-produced motherboards used in Supermicro’s servers—will slow the progress of Inspur’s expansion at US-based customers in both cloud/enterprise segments. By the same token, Huawei should also see more obstacles expanding internationally, especially for the roll-out of its 5G infrastructure, and we believe it could take longer for Lenovo’s DCG business to break even. We also see limited room for other Asian vendors to meaningfully displace more tradition market leaders; Fujitsu, which is the largest, only has 2% global x86 revenue share. On the other hand, ODM Direct vendors supplying cloud operators, specifically Quanta and Wiwynn, will continue to take share from traditional server brands, leveraging their enhanced customisation and hardware-design capabilities, on top of lower total cost of ownership. For supply chain localisation, we expect Chinese server vendors will increase their sourcing from local component-makers in the next few years; nevertheless, we expect a majority of the BOM should still be dominated by the US suppliers. Specifically, we see 40-50% of the total server BOM associated with the (Intel/AMD); 20-30% in memory (Micron, Samsung, SK Hynix, etc.); and a majority of the remaining from storage (WD and Seagate). While we China will take long time to achieve supply chain localization with key components believe China vendors are gradually catching up in their design capabilities and production supplies still dominated by US suppliers know-how—as their learning curve scales along with international market share expansion—we do not expect China to have a complete back-up solution, especially on the chipset; however, Dell focusing on expanding leading shares; we do expect sourcing of memory to shift to Japanese/Korean players, as they move away HPE focusing more on protecting profitability from Micron. Domestic sourcing will be encouraged, if CXMT/YMTC are able to deliver a successful product, but our analysts believe that it is unlikely over the next five years (outside of some consumer-level parts). Beyond the CPU, the other major obstacle to true Chinese self- sufficiency in servers is operating systems where Windows and continue to dominate.

Display: China to dominate TFT panels; may succeed in OLED but still lags in key tools and raw materials China to dominate TFT-LCD Chinese display panel makers have obviously caught up on producing TFT-LCD panels, especially for the commodity and lower-entry-barrier products. We expect their market share to further increase as they ramp up new fabs, especially amid Korean panel makers’ strategy change. China’s central government started pushing the construction of new panel fabs in 2009. Since then, leading players for large-size display applications like BOE and CSOT have built new facilities with support from the central/local government. Moreover, smaller players like CEC- Panda, HKC, CHOT, etc., were also supported by different provincial governments and constructed several new fabs in the past few years. As of 1H19, Chinese panel makers accounted for 41% of area shipment for LCD TV, 36% for NB/tablet, and 30% for monitor.

Asia Technology Strategy 22

4 November 2019

Although the industry has been in oversupply, Chinese panel makers are still building new large size panel fabs (mostly Gen 10.5) and will begin mass production in the next one to two years. Koreans to exit TFT-LCD; focus on OLED We fully expect Korean LCD makers to exit the TFT-LCD panel manufacturing over time. Most Chinese LCD makers’ share could further increase as Korean players exit the market LCD products are no longer profitable and Korea's strategy is to shift up on the technology curve to pure OLED production base while phasing out the existing LCD facilities. Both Chinese and Taiwan TV LCD panels are at sufficient quality, where major TV brands such as Samsung and LG Electronics (LGE) have been expanding panel procurement for TVs at lower price points. High-end TV panels will move to OLED TVs for both brands, with LGE enjoying a head start. For small TFT-LCD and LTPS LCD display, Samsung began phasing out production from 2014. Today, Samsung produces no smartphone LCD displays internally. Instead, its early shift toward the future of OLED displays enabled the company to enjoy near-monopoly status for more than a decade. OLED offers less competitors and is a more difficult technology to master. The near- monopoly status also enables the company to keep tight control over its equipment and material suppliers. OLED has grown into a W25 bn revenue per annum, with much more stable 10% to 12% sustainable OPM for the company. Similarly, LG Display has closed its P2, P3 and P4 lines in Paju and Kumi with most of the AP2 line used to make LTPS-LCD now being converted for the usage of OLED production. While the company still runs a profitable business in the oxide based LCDs for AAPL's PC monitors and , the mobile LTPS display has also been phasing out as Apple began to adopt more OLED displays into its iPhone models, and LGE’s smartphone volume has shrunk considerably. As a LG Group strategy, developing OLED TVs was emphasised first, followed by flexible mobile OLED transition following Apple's iPhone shift. Chinese good with rigid OLED; but lagging in flexible OLED Chinese panel makers started to spend significant capex from 2017 to expand OLED capacities, triggered by Apple's adoption of OLED panel along with a massive subsidy from the government. Several panel makers have announced accelerating their OLED capacity build-up plans including BOE, Tianma, EverDisplay (privately held), Truly, Visionox, Royole (privately held) and CSOT. We estimate over Rmb400 bn of investment in OLED panel manufacturing. We see a majority of projects are setup with government bodies, with listed companies holding approx. ~10% stake; this could help lessen the financial burden for listed companies. Chinese OLED panel makers rely on overseas suppliers for raw material and equipment. Among Chinese OLED panel makers, Everdisplay and Tianma were successful in delivering rigid smartphone OLED panels in 2016. Everdisplay has started with non-smartphone applications such as smart watches, industrial and other handheld devices, and taken 2 years for a ramp up. Tianma only shipped a small volume in 720p resolution to begin with, but now their Wuhan OLED fab can deliver rigid smartphone OLED panels at a reasonable yield rate. BOE started mass production of flexible OLED smartphone panels from 4Q17 but ramping took long time without a flagship model. We see Chinese OLED panel makers are struggling for yield improvement and cost reduction. Longer-term, we expect more Chinese OLED panel makers could deliver rigid OLED panels, while BOE remains the leader in flexible OLED panels in the China market. China to rely on Japan, Korea, and the US for key raw materials, especially OLED We believe Chinese panel makers might use more local material and components for manufacturing in the next few years, although they will still rely on overseas vendors for certain high-end components that require more fundamental research in material science. We believe material/components such as ultrathin cover glass (Corning’s Gorilla glass or Asahi’s Dragontrail glass), large size glass substrate (Gen 10.5), OLED polariser (film-type circular or liquid crystal pattern), high-end optical film (quantum dot film), OLED materials, etc., will still be

Asia Technology Strategy 23

4 November 2019 dominated by overseas suppliers, given their know-how in raw material, formulation, and end- customers’ preference. From a long term perspective, China would evolve its domestic supply chain to get a meaningful contribution from local panel makers and reduce its reliance on overseas suppliers, especially US vendors. For OLED materials/components, we believe it will take longer for Chinese panel makers to establish their own supply chain, primarily due to Samsung’s black box strategy to protect its core technology. OLED itself is a much more difficult technology to master in general, requiring a lot of trial and error due to the nature of its analogue manufacturing processes. Furthermore, the pioneer and the leader, Samsung Display has wanted to strictly control the supply chain, including building up its IP portfolio on production methods, equipment advancement and upgrading organic materials science. In particular, for OLED materials, some key suppliers such as Universal Display, Dow Chemical, Idemitsu Kosan, Duksan Neolux, etc., are already dominant players in sub-segments where Samsung Display is unlikely to supply advanced materials to Chinese panel makers. Components: largely self-reliant Acoustics: Most parts likely to be localised We see China has low reliance on imported acoustic components but it still relies on overseas codec chips and MEMS chips in mid- to high-end smartphones. We expect Chinese players to continue gaining share in key areas, after having successfully gained dominant share in Apple’s acoustics supply chain for speakers and receivers. AirPods production is also expected to be largely divided between Luxshare and Goertek. For MEMS microphone market, Chinese players are quickly catching up with Knowles. Goertek and AAC are producing MEMS IC, and looking to increase the self-supply ratio of MEMS IC. We expect it will take longer for Chinese players to realise domestic substitution for MEMS IC, and it is likely to be a medium term effort (3-6 years). For the smart speakers market, we already see the market share increase from both the end-product perspective (with Baidu, Alibaba and Xiaomi entering the top 5 global brands), as well as the supply chain (with increasing share being taken by Chinese players such as AAC and Goertek). Antenna: work in progress to move to more advanced designs We believe China will strengthen its domestic value chain and ecosystem by moving upstream of the FPC antenna industry, somewhat overlapping with the PCB industry. We think the industry/enterprise investment will move from FPCB and antenna module assembly to CCL midstream in the next 1-3 years. Besides, local governments/ research institutes are likely to continue investing in R&D for 5G antenna design and materials/resin development, etc. In the long term, China might close the gap with Japan/US chemical companies on resin and film manufacturing. Casing: lagging in metal casing but a leader elsewhere; still dependent on casing equipment imports Chinese players are already the market leaders, in terms of market share and technology, across different casing materials for Android smartphones and tablets. For the Android supply chain, BYDE leads metal, plastic, and glass casings, and BYDE is aggressively pursuing an aggressive pricing and capacity expansion strategy. In terms of quality and segment, FIH and AAC are targeting high-end metal mid-frame casing, BYDE focuses on the mass market, and Tongda focuses on the mid-to-low-range segment. Lens Tech is expanding its share in Android glass casing with new products delivering rising precision, strength and more colour options; Tongda leads in ‘glastic’ casing and innovating with ICM and 3D glastic casing materials. For Apple glass cover/casing also, Chinese companies (Lens Tech and Biel Crystal) are the market leaders, with advanced new products on rising precision from unibody design and higher strength/more colour options. However, Chinese players are trying to get into Apple metal Asia Technology Strategy 24

4 November 2019 casing but still lag behind their peers in Taiwan, and it’s unlikely Chinese players could get into the Apple metal casing supply chain in the short-term due to lagging in technology advancement and precision manufacturing. In the Android supply chain, we see low reliance on overseas suppliers. Domestic vendors dominate in plastic, metal, glass and ceramic casing. However, reliance on overseas equipment makers is still high, as the casing manufacturing process involves multiple steps, and requires proprietary machines with specific/high-entry-barrier technology. Some of these can only be sourced from foreign suppliers so far, such as for certain coating, photolithography and silk printing machines. However, BYDE has developed and adopted some in-house equipment including CNC machines, and is also adopting more general equipment that can be used for multiple materials/product lines. Camera Module: Largely self-reliant; will increase global share Chinese players already have well-established capability in CCM (multi-cam) and handset lens, and we think in the near-term (within 3 years), Chinese players will likely establish their leadership in 3D sensing and advanced optical solutions, such as WLO and periscope-style products. However, in the CCM value chain, key components such as CIS and VCM are still largely dominated by Japanese and Korean players, despite some Chinese players making progress in these products; e.g., Win Semiconductor for CIS and ZET, JSS Optical Technology and Shanghai B.L. Electronics for VCM. For handset lens, Sunny Optical is an important supplier to Chinese branded smartphones. In the long-term, we believe will see Chinse players establish their ecosystem and build a domestic supply chain for optical components, including CIS, high-end lens and VCM, and gradually increase their share and technology leadership. Hence, Chinese local camera modules makers such as Sunny Optical, Q Tech etc. are already taking a major share at local Chinese handset makers (i.e., Huawei/Oppo/vivo/Xiaomi) not just for mid/low-range phones but for high-end flagship phones as well. Meanwhile, a growing Chinese camera modules industry, as well as its improving technology curve, is offering better growth opportunities out of China, by penetrating into the Samsung handset supply chain, for instance. Considering rapidly rising multi-cam penetration at SEC—even for mid/low-range phones (i.e. Galaxy A series), relevant camera module makers (both Korea or China camera modules suppliers) are all enjoying the total addressable market growth through shipment volume as well as blended-ASP hikes. However, in the mid to long term, we believe the existing Korea handset supply chain will likely face more headwinds amid intensifying competition at the major client (SEC), either through market-share loss or ASP-cut pressures. Recently, Samsung is even testing the ODM outsourcing strategy to Chinese makers. While we believe total smartphone volumes outsourced will remain controlled (~20 mn in 2019, and up to 40 mn in 2020) and only applicable to the very low-end A series (which has little chance of ever becoming profitable). However, it still seems a downside risk to the existing Korea camera-module supply chain (mainly for mid/low- range) such as Partron, Mcnex, Powerlogics, etc. In contrast, the strategy will likely give better growth opportunities for local Chinese camera modules makers, in our view. For high-end flagship phones at Samsung, SEMCO may remain as key supplier for longer, considering the need for key optics technology protection. However, more ASP-cut pressure seems inevitable, given emerging alternative options from Chinese module makers with a better technology readiness. CMOS Image Sensor (CIS): a credible Chinese player now CIS is one of the very few semiconductor areas where China has a top-tier player. We think OmniVision’s (now owned by a Chinese company, Will Semiconductor) CIS technology could be 1-2 years behind Sony, but at a similar level to Samsung. In the automotive segment, OmniVision is ahead of Sony and Samsung, and only behind On Semi. Driven by the geopolitical tension, OmniVision has been taking over the mid/high-range spec smartphone segment from Samsung since mid-2019, and also growing its share in the China automotive Asia Technology Strategy 25

4 November 2019 market. OmniVision’s gap vs Sony should continue for 1-2 years as most OEMs customise CIS chips for flagship models, so it will take time for OmniVision to regain position in the premium segment. In five years, we expect OmniVision’s technology to be slightly behind, but close to Sony, and its market share to grow from 12% in 2018 (per Gartner data) to high-teens or twenties percentage in 2023. However, in the near term, we expect existing CIS suppliers to continue expanding production capacity to cater to the growing demand. Sony plans to increase monthly capacity to 130,000 by March 2021 (up from 105,000 in Apr–Jun 2019, based on 300mm wafer). Over the medium and longer term, we see Sony focusing on the functional fusion of sensors for automotive applications (to integrate data taken from cameras, LIDAR and millimetre-wave radar). Through its alliance with Microsoft, the company is also targeting a business model that generates recurring income by channelling development resources into CMOS image sensors with embedded edge AI processing capabilities. Lens: Largely self-reliant The overall reliance on government is small, and leaders such as Sunny have become No. 2 in terms of market share in the smartphone lens market, with continuous R&D investment of their own. The biggest competitors include Largan and Genius, which have entered the Apple supply chain (Sunny not yet). We believe Chinese players are gradually improving their capabilities in lens, and investing in R&D for high-end models, and should be able to penetrate the high-end market over the next 1-3 years. MLCC: The Chinese are lagging and likely to do so in the medium term too Despite the fact that China is one of the largest MLCC-consuming markets, the self-sufficiency rate is relatively low. Currently, MLCC supply is mainly dominated by Japan (Murata/TDK/Taiyo Yuden/Kyocera), Korea (SEMCO / Samwha Capacitor) or Taiwan manufacturers (Yageo / Walsin Tech). A growing number of Chinese players are trying to break into the market— Fenghua Advanced Technology (000636.SZ), Tianli Holdings Group Limited (0117.HK), Chaozhou Three-circle (300408.SZ), Fujian Torch Electron Technology (603678.CN)—but their combined global market share is only still around ~5%. These Chinese companies are largely addressing MLCC demand from consumer electronics, mainly for the mid/low-range segment, (i.e., larger size / low capacitance) where leading JP or KR MLCC makers are withdrawing for production optimization (i.e., A/V, PC, appliances or low-end handsets). In the MLCC arena, while Japanese-owned MLCC makers and SEMCO are working to lock in technology by manufacturing materials and equipment in-house, Taiwanese- and Chinese- owned makers are dependent on Japanese materials and equipment suppliers for the majority of their raw materials and production equipment. This makes it highly unlikely that Chinese MLCC makers will close the technology gap to the Japanese and Korean frontrunners anytime soon. Our focus in the near term is on how they can shrink the gap between themselves and Taiwanese makers.

Asia Technology Strategy 26

4 November 2019

Memory semiconductors

Keon Han Current market structure 82 2 3707 3740 [email protected] Manufacturing DRAM locally is historically symbolic Chaolien Tseng 852 2101 6795 Historically, not only did the DRAM industry follow a distinctive pattern of capex, but it also [email protected] produced the country's industrial policy cycle. Having the ability to adopt, develop and Hideyuki Maekawa 81 3 4550 9723 manufacture DRAM semiconductors on-shore usually symbolised a country's "arrival" into the [email protected] upper echelons of high-tech manufacturing. This was the case where the US, the originator of DRAM, once dominated the industry but was displaced by Japan over time. Korea has now displaced Japan, and dominates DRAM production. Taiwan also once had a thriving DRAM industry, which eventually collapsed under the weight of a perpetual need for capital and an unforgiving need to succeed in developing the next generation of technology through constant R&D. The result was similar for the Japanese DRAM industry. Micron, in the end, became the consolidator for the Taiwanese and Japanese DRAM assets by acquiring the deeply-discounted residual assets. The DRAM has industry consolidated further since the financial crisis. Samsung, Hynix and Micron now combine to control 96% of the industry capacity – creating an oligopolistic structure. This followed the bankruptcies that forced Elpida, Qimonda and Promos out of the industry. The remaining, smaller, Taiwan producers such as Nanya, Winbond and shifted to a more concentrated strategy away from advanced DRAM production of DRAM for specialty applications.

Figure 28: The DRAM industry now has the fewest producers in its long history

100% 90% 80% 70% 60% 50% 40% 30% 20% 10%

0%

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017 2018

Samsung Micron Hynix Nanya Elpida Qimonda ProMOS Powerchip Others

Source: Company data, Credit Suisse estimates

Asia Technology Strategy 27

4 November 2019

Figure 29: Global DRAM revenue M/S in 2019E – Korea dominates Nanya Others 2% 2%

Micron 23% Samsung 44%

SK Hynix 29%

Source: DRAM eXchange, Credit Suisse

Figure 30: Global DRAM production- Consumer segment likely Figure 31: Specialty (Consumer) DRAM market share in 2019E China's focus area – Threat to Taiwan first

Consumer Graphic China IDM Other Taiwan 8% 5% vendors fabless vendors 3% 3% SK Hynix 10%

PC Nanya 13% Technology Mobile Samsung 39% 41% 13%

Server Micron 33% 14% Winbond 18%

Source: DRAM eXchange, Credit Suisse Source: DRAM eXchange, Credit Suisse Different history for NAND The initial battle among various types of flash memories finally settled on NAND flash design due to its economical scalability required in anticipation of data storage demand explosion. NAND's role is to alter the way data is stored—from hard disk drives, which will eventually be displaced by flash-memory-based data storage. NAND storage device penetration—such as USB, UFS and SSD—continues to rise as the dollar-per-bit storage continues to fall over time. Other advantages—such as weight, smaller form factor, faster read/write speed, less heat generation, higher storage density per area—all make NAND preferable vs HDDs as a storage medium as the price gap narrows. While the concept of NAND flash memory was developed by Toshiba and Sandisk (now Western Digital), Samsung was quick to seize and replicate Japan's success under IP licensing from both companies. Micron joined hands with Intel (but are now going their separate ways) in order to compete in the industry, while Hynix continues to try and remain a relevant player. All NAND makers pay some IP royalties to Toshiba and Sandisk, but the amounts vary depending on the strength of the individual company's patent portfolio filing that payment could be netted out. Today, global NAND production is dominated by Samsung and Toshiba/Sandisk with combined M/S of 67%. Micron and SK Hynix are smaller producers, with

Asia Technology Strategy 28

4 November 2019

Intel carving out a niche in the enterprise SSD segment. The majority of NAND is consumed by either the smartphone or SSD segments in both PCs (client) and Servers (enterprise).

Figure 32: Global NAND M/S in 2019E – more balanced competitors

Intel Others 9% 1%

SK Hynix 10% Samsung 35%

Micron 13%

WDC Toshiba 14% 18%

Source: DRAM eXchange, Credit Suisse

Figure 33: NAND Flash demand—concentrated in handsets and SSDs 100%

80%

60%

40%

20%

0% 1Q19 2Q19 3Q19E 4Q19F 1Q20F 2Q20F 3Q20F 4Q20F Others Solid State Drives Digital Still Cameras USB Flash Drives Portable Media Players Tablet PCs Handsets

Source: DRAM eXchange, Credit Suisse China's game plan China's turn next in DRAM/NAND Competition in the memory industries, both DRAM and NAND, is tough, requiring ever- increasing capital. The next-generation technology development is not only more difficult, but more time-consuming, to a point that Moore's Law has clearly slowed down in DRAM, although re-acceleration is seen in NAND due to the advent of the 3D NAND architecture. Having the capital is no longer a guarantee of success, given these barriers and technology difficulties. The DRAM industry has consolidated to a point where the three remaining producers are comfortable with the status quo in market share, with enough pricing control to give them sustainable profits even through a cyclical downturn. With the end-market demand now enjoying ~15-20% CAGR, the industry seems mature enough that it can no longer accommodate another major supplier who could again disrupt the balance of the market.

Asia Technology Strategy 29

4 November 2019

For NAND, the major players are still jockeying for a leading position, with each supplier very capable of producing quality NAND chips. The NAND industry is still not ripe for consolidation, as the weaker NAND players partially fund their loss-making (and now cash-burning) NAND businesses with highly profitable DRAM products (in the case of SK Hynix and Micron). NAND memory is technically a side business for Intel, while Toshiba has restructured its semiconductor division with fresh capital injection through a consortium of private equity investors. While Samsung is operationally slightly loss-making, NAND is sufficiently cash-flow positive, given the company’s earlier commitment to shift to 3D NAND production, its technology lead, and broader customer base. The rationale for China's move into memory semiconductors So, facing these industry dynamics, what is the motivation for China to join the DRAM/NAND race; and can China succeed as a late entrant? If so, when can China influence the DRAM/NAND industries and what are the implications for the incumbents? It is already clear that China is at the onset of developing its own DRAM and NAND industry. It has already selected its DRAM champion in CXMT, and YMTC for NAND. Initial fabs have been built, with some WFE installation in order to test-produce both DRAM and NAND products. The key motivation for China to develop its own memory semiconductor industry could revolve around the various reasons below. 1. Memory is part of the IC industry development. In 2014, the China government issued a National IC Development Guideline. It is segmented into four broad areas of IC industry development: (1) design; (2) manufacturing; (3) packaging/test and equipment; and (4) materials. In short, advancement of the semiconductor industry value chain is a formally stated industrial policy. While China has been somewhat successful in developing its own logic/foundry semiconductor industries through SMIC, Grace Semiconductor (in the past), etc., memory semiconductor development has not been explicitly stated. However, capital allocation toward the memory industry seems to be the natural outcome given the: (1) attractive size of the addressable end-market; (2) break-away from 100% reliance on imports; (3) commoditised nature of DRAM/NAND chips, which helps capital allocation to be more focused; and (4) perceived relative ease of memory design/manufacturing compared to logic (the memory industry has other, complicated, industry dynamics). Also (ironically), the memory industry is attractive given it has fewer competitors (at least in the case of DRAM).

Figure 34: 2014 China government national IC development guideline Key words Development strategy Near term: Mobil smart terminal and network communication: mobile terminal, digital TV, network communication, smart wearable and OS (1) Supply chain collaboration Mid-term: Cloud computing, IOT, and big data Design (2) M&A Break-through: Smart card/electricity/transportation, satellite navigation, industrial control, financial electronics, automotive, medical and embedded software Speed up the 45/40nm foundry capacity ramp up and 32/38nm line set up, 3D process R&D, capability set up of 22/20nm and 16/14nm (1) CAPEX Mfg foundry production capability (2) Advanced node Excel the specialized production line such as Analog and mix circuit, MEMS, high voltage and RF (1) M&A Pkg/Test CSP, WLP, TSV and 3D stack (2) Evolution (1) Key equipment/material break-through R&D in key equipment such as lithography, etching and implant R&D of Equipment/Material (2) Mfg synergy key material such as photoresist and large wafer Source: Government of China, Credit Suisse

Asia Technology Strategy 30

4 November 2019

Figure 35: Memory semi value as % of global total semi value 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Memory Other semis

Source: SIA, Credit Suisse 2. Lacking domestic source. As the world's largest tech manufacturing base, China is either importing or relying on supply from the domestically-based foreign producers— such as NAND chips from Samsung's Xian fab or DRAM chips from Hynix's Wuxi fab. Overall semiconductor imports in China, including memory semiconductors, reached US$260 bn in 2017 and exceeded the value of oil imports that year. Domestically produced chips made up less than 20% of that value, which partly drives the motivation to rely less on foreign imports going forward. Over time, China has internalised the manufacturing of tech components from lower value chains such as PCBs, HDIs, LEDs, TFT-LCD panels, etc. Its logic and foundry industry development is ongoing. Memory semiconductor is the next area of focus. 3. Diversify supply risk. Without domestic memory chip supply, China companies often cite their businesses being negatively impacted by volatile memory prices and supply shortages. For example, during the 2016 – 2018 up cycle, DRAM and NAND prices surged on a global shortage, which damaged many of the Chinese smartphone OEMs’ cost structure. Also, some Chinese smartphone makers perceive that Samsung controls the supply and price of DRAM and NAND as a competitive tool to keep Chinese smartphone makers in check. These are some of the reasons for Chinese companies to support the development of a domestic memory chip supplier. Past failed attempts to acquire/invest in global companies force internal development The 2015 attempt by China's Tsinghua Unigroup (a controlling shareholder of YMTC) to acquire Micron for US$23 bn, is well documented. The acquisition attempt was not realistic, because the US authorities—such as Committee on Foreign Investment in the US (CIFIUS)—would have blocked the deal on concerns of national security, given that Micron was the last-remaining US memory semiconductor maker. Therefore, the potential deal never reached a formal offer stage. Later in 2015, Tsinghua Unigroup offered to buy 20% of SK Hynix for ~US$5.3 bn. One of the conditions required SK Hynix to also set up a NAND fab in China, where Tsinghua was to become one of the manufacturers of Hynix's NAND products. This offer was also rejected by Hynix. No incumbent producer so far has shown any willingness to either set up JVs or engage in technology/engineering agreements with Chinese entities. These rejections are logical, given the pains the remaining DRAM makers have gone through to survive to the end, in order to finally enjoy a more profitable industry. Furthermore, the last thing the NAND industry would need is another large supplier, in what already looks like a crowded field that is already facing potential industry consolidation. While acquiring or investing in global companies, setting up JVs, or acquiring IPs by entering into engineering agreements would have been an easier path of entry; most global competitors are keen to keep China out of the industry as long as possible, due to the potential supply disruption risk. Asia Technology Strategy 31

4 November 2019

CXMT and YMTC the selected memory champions While there are many Chinese entities that have indicated interest to engage in production of memory semiconductors, we conclude that currently, there are only two likely players that will pioneer the memory semi development: Changxin Memory Technologies (CXMT) for DRAM, and Yangtze Memory Technologies (YMTC) for NAND. The memory market is highly cyclical and requires endless capital investment. Returns are not guaranteed, and could be very low for a very long period of time, with laggards seldom reaching a level of technology sufficient to become profitable). From the China central government’s perspective, it may not be desirable to redundantly invest heavily in the same technology areas; such was the case in the TFT-LCD and LED industries that rapidly created overcapacity conditions, even among Chinese start-ups. Most of these companies have yet to reach profitability, after 10 years of operation, and are now consolidating. Furthermore, while some provincial governments expressed support and have a desire to co-invest in new memory start-up companies, they may lack a sufficient capital base to support projects sustainably. Additionally, the engineering talent pool remains thin, which has been one of key issues in China for its general semiconductor industry advancement. Memory is no exception. There are simply not enough NAND and DRAM engineers in China to rapidly expand production, because the industry is at an infant stage. For China, instead of targeting a certain global market share within the next five years, we think the priority will be to establish its own supply chain and fulfil the domestic consumption needs. YMTC has been engaged in developing NAND since 2015. It is most likely that YMTC’s technology will still be behind Samsung and Toshiba for the foreseeable future. Therefore, while the technology may not be cost-competitive compared to the global leaders, the initial goal of supplying the domestic demand could be satisfied, if the final product has the quality to be consumed in the local end-markets. CXMT’s initial goal as a DRAM maker is similar to YMTC’s. It wants to capture some of the domestic DRAM supply, likely in the area of specialty DRAM. If this stage is successful, it can plan to broaden the range of products to compete in broader global markets. So far, NAND development appears more advanced than DRAM. From a global competition perspective, we do not see Chinese memory producers disrupting the current market structure, either in DRAM or NAND given its current ramp-up schedule, lack of product commercialisation, and having no clear next-generation technology roadmaps. CXMT 19nm DRAM close to mass production Founded in May 2016, CXMT’s DRAM development started with Qimonda, according to the company. Qimonda is the previous global leading DRAM tech provider, the inventor of Buried Word Line, and pioneered the advanced DRAM technology. CXMT said it has transferred 10 mn technical files (2.8TB of data) from Qimonda. Besides Qimonda, CXMT indicated close partnerships with industrial leaders such as AMAT, ASML, KLA, LAM, Tech Insights, and TEL. It also jointly released technical papers with ASML at the SPIE conference. The company claims it has spent over US$2.5 bn so far on R&D expenses and capex. CXMT’s first-generation technology is 19nm DRAM. Initially, the company had aimed to begin production in 2018 but the schedule was delayed for a year. It has been sampling products with some potential domestic customers. If successful, mass production could begin in early 2020. Its next process node will be 17nm DRAM, although there is no clear timeline. CXMT’s target customers include large and small domestic OEMs. It plans to produce LPDDR4 for mobile devices as well as DDR4 for PC and Server use eventually. Large OEMs likely will require more time for qualification. Given CXMT's plan to only produce on a small scale initially, any order wins from smaller OEMS could be sufficient to fill their initial capacity. CXMT’s current DRAM fab is built for maximum 120k wpm. In early 2019, it aimed to build 20- 30k/month capacity by mid-2019, but the schedule has been delayed to early 2020. Its longer- term plan includes building another fab capable of 120k wpm production capacity, but currently there is no timeline when the second fab will begin construction.

Asia Technology Strategy 32

4 November 2019

YMTC prioritises on 64L qualification and 128L R&D YMTC developed 32L MLC NAND in 2015 and it collaborated with Spansion for its first generation 32L NAND. It then built a fab in early 2017, began tool move-in from mid-2018 and started risk production from 1Q19. That said, as the technology didn’t have any cost advantage, only a small volume was produced. Currently it has only around 10k/month capacity for 32L NAND production and its yield is likely 70%+. YMTC is now in trial production of 64L TLC NAND and sampling with domestic SSD companies. While its 32L NAND is mainly consumer grade, its 64L NAND is trying to qualify with Huawei for enterprise grade now. Currently its fab capacity is about 30k/month capacity, including the 10k capacity for 32L NAND. If YMTC should pass 64L NAND qualification with Huawei or another top domestic customer, and improve its fab yields to 85%+, it could begin another round of financing to support more capacity build. That said, we note that it is unclear if it will enter volume production on 64L, even if qualified, given the cost constraints. 64L TLC NAND would not have any cost advantage at the current market price, compared to global memory makers already moving on to 92L/96L and 128L production. While the company believes that the cost structure of its 64L TLC NAND will be sufficient enough from a positive cash flow point of view, we understand that it could decide to develop 128L before entering volume production. YMTC’s 64L NAND is based on its Xtacking technology. YMTC proclaims that Xtacking’s independent processing eliminates thermal and stress impacts and enables higher NAND I/O speed. Xtacking 64L TLC NAND enables more efficient use of die area and smaller die size (by 20-30%) than conventional 3D NAND. Xtacking could achieve higher bit density, the best bit density of all TLC 64L products in the market, according to the company's presentation material. Xtacking adopts a modular approach to accelerate process development. Periphery and array wafers are produced independently, and then stacked. Development time is reduced by at least three months, according to the company. It also estimates manufacturing cycle time is reduced by 20-25%, as CMOS and array wafers are manufactured at the same time. While each NAND producer will proclaim advantages of its own design, ultimately performance, cost advantage and reliability will be determined by the customer choosing the product. While we do not have details of its royalty payment, it is believed that YMTC pays royalty to some international memory companies to use their IP. YMTC’s next generation technology could be 128L, but there is no clear timeline. YMTC has begun R&D for 128L NAND, and risk production is scheduled for 4Q20. Additionally, the company has also kicked off another project to build an initial DRAM R&D team with 300 engineers, while using XMC’s existing facility as its DRAM pilot line. The drive for its DRAM project is the same as its NAND development; YMTC is expected to help the country to achieve the mission of memory self-sufficiency. Their long-term plan is to build nine fabs in 10 years, targeting 15% market share in 8-10 years. The company has about 4,000 employees, with 1/3 being R&D engineers, while XMC has 1,200 employees.

Figure 36: YMTC’s 3D NAND R&D journey Year Highlights 2014 3D NAND kicked off in XMC 2015 9L test chip validation 2016 32L test chip tape-out. YMTC founded 2017 32L product tape-out. 1st Si validation. YMTC 32L 64Gb MLC 3D NAND flash ES 2018 64L product tape-out. 32L production. 64L First silicon 2019 64L TLC NAND sampling and trial production Source: Company data, Credit Suisse

Asia Technology Strategy 33

4 November 2019

Conclusion How big can China be in DRAM and NAND, in 1 to 3 years? For DRAM and NAND, we see very little supply impact from China by 2020. In DRAM, while CXMT's first fab is designed for up to 120K wafers based on announced 19nm technology plan, the initial ramp of 20K to 30K wpm has already been delayed by over a year. Additionally, only providing output on 19 nm (we cannot confirm if this is industry standard 19nm yet). By mid- 2020, 19nm would not be competitive to industry leaders who will be producing mostly on y nm with Samsung potentially moving on to 1z nm. Also, with such a small production scale and lack of engineers, it is unlikely CXMT can simultaneously work on various DRAM architectures such as mobile, PC, Server, etc. Most likely, CXMT will likely find itself competing in the consumer specialty DRAM segment which would require less advanced nodes and speed. Therefore, the Taiwanese consumer DRAM makers would be the first in line, in terms of competitive threat, only if CXMT is successful. From capacity planning, 20K to 30K wpm on higher nodes does not represent overall DRAM industry supply competitive threat. However, if CXMT were to be successful and ramp to its maximum 120K by the end of 2022, it could represent as much as ~8% of global DRAM capacity, although likely significantly less in terms of bit shipment M/S. YMTC's 3D NAND mass production roadmap is less clear. Having 64L TLC design at this juncture of NAND industry development puts its technology far behind industry leaders that are already producing on 92L/96L, and investing in 128L. The industry leaders are already dismantling 6xL capacity as any hope of return to profitability fades. This technology would have no cost advantage and have limited use in the end-market applications. Although it has begun R&D for 128L NAND and targets risk production from 4Q20, given its history, we expect more time is needed as it may be working on the successful development of 64L for a while. While the long-term goal of ramping to 300K wpm has been indicated by the company, it would first need sizeable customer wins in order to produce NAND in that magnitude. YMTC has announced that it began sampling its NAND products with various SSD-maker clients but success of any qualification is unknown currently. Capacity ramp schedule would be difficult to project without confirmation of a commercially viable product. Key issues to consider Government ready for long-term investment, and losses

Semiconductors require long-term, sustainable investment in both R&D and capex. We believe that the Chinese government is aware of such industry dynamics and is prepared for (or accepts) losses for years in order to build its own memory semiconductor ecosystem. The LCD industry development and support for BOE is a good example. BOE, after purchasing its initial TFT-LCD technology from SK Hynix (Hyundai Electronics back then) developed its LCD business for over a decade but remains loss making operationally today (excluding government subsidy). While not efficient on a return-on-capital metrics, China has succeeded in developing its own formidable TFT-LCD industry with BOE as its champion. This ensures at least a steady source of TFT-LCD supplies for its smartphone and TV market, the world's largest. Not having to worry about trade frictions and other potential supply limitations are also important considerations. The fallout from the perpetual over-investment in the industry is now having the effect of driving historically-strong LCD makers such as and LG Display to abandon their TFT-LCD production. Sources of Capital The capital support from the government includes bank loans and equity investments. The capital from the private sector is often in the form of equity investment. The central government can help top-priority companies secure bank loans at lower, subsidised interest rates than what is available through normal commercial bank channels. In the past decade, many provincial governments and companies have generated profits from selling government owned real estate assets. Given the recent slowdown in the real-estate markets, provincial governments and larger

Asia Technology Strategy 34

4 November 2019 companies need to turn to other industries to re-invest the proceeds, with direct equity investment being the preferred method. From 2017-19, many of these entities have chosen to invest in the new semiconductor projects. Sources of capital include: China Development Bank, Hubei Government, Wuhan Government, China IC Fund, Tsinghua Unigroup, etc. Addressing the lack of engineering resources The semiconductor engineer talent pool is shallow in China, and this is especially true in memory semiconductors. Various DRAM project start up attempts were spearheaded by pooling Taiwanese and Japanese engineers and managers, from companies that were consolidated or finally exited the DRAM industry. DRAM and NAND chip design engineers from Samsung or Hynix are highly coveted. While incentives such as huge pay increase, several free trips home every year, and heavily subsidised apartments are offered to critical engineers, the Taiwanese engineers are normally highly prized given the common language and cultural background. Acceleration of the memory industry development could have a honing-out effect of some Korean engineering talent, which has been stated as a key risk by both Korean manufacturers and government policy makers. Nevertheless, the integration of various engineers who have never worked together before, in a brand new fab setting, will pose some challenges and high cost burden of maintaining mostly foreign labour in the beginning. Potential IP disputes With the lack of any known acquisition of major DRAM/NAND IPs, China is essentially starting from scratch. This brings the burden of developing its own technology, while not infringing on countless patents already filed in DRAM and NAND, in both design and manufacturing processes. Generally in the technology IP battle, new producers are seldom blocked initially. Rather, the lawsuits requesting for blocking sales come when products actually become competitive. In China's case, the current sources of chip design in DRAM and NAND are actually unknown. Bringing in chip designers from abroad nearly always risks patent infringement. New Chinese start-ups have been busy announcing the success of their own chip designs, but any IP infringements will not be known until the final products show up in the market place. Nevertheless, any disputes maybe silent for several years given the initial production plans are relatively small in scale, on lagging technological designs and largely targeted to niche domestic buyers. Even if scale grows significantly enough to impact global supply, if the consumption is largely based domestically, not much IP protection is expected. Only when products find their way into final exportable items such as smartphones or servers can competitors contest any type of patent infringements, if any, which could be years away. Sustainable equipment support Development of memory industry requires a high degree of support from global memory WFE makers. Historically, China has been a big investor in any area it chooses to endeavour into, such as in the TFT-LCD and LED industries. Therefore, development of the Chinese memory industry could be a sustainable and lucrative event for global semi equipment makers. However, as seen the in case of Fujian Jinhua's, the entire DRAM project was halted to a point of extinction by the action of the US government banning any US semi equipment sales to the company. While the reason is purely company-specific, involving IP theft from Micron, similar perception of risk will linger for CXMT and YMTC, particularly if any IP infringements are to be found. Therefore the Chinese companies are accelerating their own R&D and IP development and increasing the number of patent filings/acquisitions. Support from customers From a pure business perspective, there is no clear reason why even a domestic buyer of DRAM/NAND would bother supporting new memory start-ups. In the case of Korea, it really did not have a large domestic demand to fulfil and the memory semi industry was developed purely as an export item. While the long-term view of having a reliable domestic producer is in the interest of both the Chinese government and system integrators, final products will not be made using inferior memory components in the name of nationalistic sentiment. There will be different levels of quality requirement depending on end-usage, but without an explicit national

Asia Technology Strategy 35

4 November 2019 campaign to incentivise support for domestic suppliers—such as import limitations, tariffs, or subsidies to the buyers—we see no reason why customers would take that product performance risk. Such is the case today with flexible , where many Chinese smartphone makers are abandoning BoE and purchasing Samsung products in order to make more globally competitive smartphones. What if China fails? China failing in DRAM/NAND development would be industry neutral, while success could bring about a global over-supply risk in the future. Currently, the consensus expectation is for China to succeed in NAND on a limited basis, while DRAM development would likely to prove more challenging. Ever since the final stage of industry consolidation, DRAM makers are generating highest profits in the history of the industry. Any hope and chance of DRAM producer's stock price multiple expansion in the future will likely be eliminated. Investors continue to debate if the DRAM industry's profit margins and sustainable FCF have shifted upwardly on a sustainable basis to justify a higher valuation multiples. Any entry by China in mass scale will likely erode that confidence. China's success in NAND will likely trigger industry consolidation in the future, particularly for the highest-cost producers such as Micron and SK Hynix. This would be true if Chinese companies are willing to invest through 10 years of losses, driving capital depletion in the weakest producers.

Asia Technology Strategy 36

4 November 2019

Logic Semiconductors

To move up the value chain, China has also been developing its foundry, back-end and fabless Randy Abrams, CFA 886 2 2715 6366 IC design sectors to a varying degree of success—barriers are the highest in foundry due to the [email protected] high capital intensity, established ecosystem of IP, design and customer base at the existing Chaolien Tseng foundries and manufacturing challenges ramping up an ever more advanced technology node 852 2101 6795 every two years where equipment companies also put their highest focus and earliest [email protected] investments enabling the leaders. The IC design sector also has varying barriers by application due to requirements for the integration of multiple IPs, support for complex and evolving industry standards, design complexity and know-how on advanced geometries and also established customer relationships. Back-end assembly and test has traditionally been higher in advanced packaging due to IP, customer service and technological capability, although M&A has accelerated the Chinese back-end suppliers’ entry. We profile China's varying success catching up across the logic foundry, back-end and IC design sectors. IC design: China’s presence rising in consumer and mobile, still lags in higher value areas Current market structure The China IC design market has had a steadier and gradual penetration as compared with the flat market share in foundry and M&A-accelerated growth from the China OSATs. The local fabless companies are benefiting from a combination of domestic R&D and wafer subsidy support, recent desire for localised and non-US sources of technology, rise of China brands in mobile and consumer and a gradually improving IP and engineering base of talent supplemented by foreign hires and returning Chinese engineers from overseas corporates and academia. At the policy level, China is targeting a number of the high growth areas, including mobile communications, networking, cloud computing, and AI/machine learning. China fabless has grown in importance in the global rankings, improving from only 1 company in the top 50 fabless in 2009 (Hi-Silicon) to 11 in the top 50 in 2018 (HiSilicon, Unigroup, ZTE Micro, ISSI, CIDC, Nari Smart Chip, Datang, GigaDevice, Montage, Rockchip and Allwinner). According to iSuppli, China fabless also represented 5 of the top 10 growing companies in 2018, with ISSI #1 up 32% YoY to US$645 mn, Allwinner #3 up 29% YoY to US$207 mn, Hi-Silicon #4 up 25% YoY to US$5,880 mn, and GigaDevice #9 up 18% YoY to US$360 mn. The five other fastest growing fabless were NVIDIA, AMD, MegaChips, Nordic Semi and Elite Semiconductor.

Figure 37: China has 11 of the top 50 fabless Figure 38: China represents 12% fabless share

Source: IC Insights Source: IC Insights We estimate the top 20 fabless companies in China reached US$16.8 bn sales in 2018, representing 16% global fabless share of our global fabless sample of US$104 bn sales. China fabless has been outgrowing the global average, showing a 23% CAGR since 2010 vs 5% for the Top 10 global fabless and 6% for the broader fabless sector.

Asia Technology Strategy 37

4 November 2019

Figure 39: China's IC design growing at a 23% CAGR to take 16% market share in 2018

China 2010 2011 2012 2013 2014 2015 2016 2017 2018 CAGR IC Design Companies revenue revenue revenue revenue revenue revenue revenue revenue revenue 2010-18 Major products Shenzhen HiSilicon Technologies $652 $1,032 $1,178 $2,120 $2,950 $3,299 $3,881 $4,480 $6,035 25% Networking/Set-tops Bitmain $0 $0 $0 $0 $0 $137 $278 $2,518 $4,268 NM Cryptocurrency mining machine Spreadtrum Communications $346 $674 $725 $1,050 $1,110 $1,640 $1,866 $1,587 $1,286 23% Mobile processors RDA Microelectronics $191 $289 $391 $380 $339 In SPRD In SPRD In SPRD In SPRD NM Connectivity and RF Goodix NA $13 $86 $102 $127 $167 $448 $545 $522 NM Fingerprint ICs ZTE Microelectronics NA NA $25 $45 $470 $520 $600 $775 $520 NM Networking / Wireless CEC Huada Electronic Design $74 $128 $123 $169 $183 $504 $507 $368 $463 27% Smart card ICs Silan Microelectronics NA $238 $203 $263 $263 $300 $343 $402 $425 NM Communications ASICs BYD Microelectronics NA $144 $125 $105 $275 $345 $424 $403 $383 NM Image Sensor / fingerprint Galaxycore $124 $190 $201 $295 $367 $350 $385 $365 $380 15% Image Sensors Beijing Nari Smartchip Microelectronics $180 $220 $280 $350 $400 $436 $480 $408 $347 13% Industrial ICs Vimicro $91 $64 $76 $65 $100 $276 $302 $293 $346 16% Consumer ICs Tongfang Guoxin NA $97 $90 $142 $180 $187 $198 $247 $330 NM Smart card / ASIC GigaDevice Semiconductor NA $75 $97 $96 $93 $178 $215 $240 $283 NM NOR Flash Memory Montage $0 $0 $0 $100 $150 $200 $220 $225 $270 NM Memory interface design $117 $128 $103 $152 $115 $218 $228 $235 $242 9% Mobile processors NA $50 $168 $246 $185 $181 $187 $179 $206 NM Tablet/audio processors Fuzhou Rockchip Electronics $58 $82 $108 $216 $152 $152 $191 $179 $182 16% Tablet/audio processors Datang Microelectronics Technology $74 $92 $116 $151 $131 $146 $132 $86 $89 8% Smart card ICs Beijing Ingenic NA NA NA NA NA $11 $16 $26 $37 NM Acquired Omnivision Top 20 China fabless $2,153 $3,768 $4,359 $6,357 $7,928 $9,601 $11,262 $13,731 $16,808 23% YoY Growth 75% 16% 46% 25% 21% 17% 22% 22% Top China fabless market share 3.8% 6.0% 6.4% 8.4% 9.3% 10.7% 12.7% 14.1% 16.2% BRCM $6,818 $7,389 $8,006 $8,305 $8,388 $8,472 $13,292 $17,665 $20,029 9% Networking/Broadband AVGO $2,093 $2,336 $2,364 $2,520 $4,269 $4,300 In BRCM In BRCM In BRCM NM RF/Storage/Networking LSI (Acquired by Avago) $1,870 $2,044 $2,506 $2,370 In Avago In Avago In BRCM In BRCM In BRCM NM Storage/Networking QCOM QCT $7,204 $9,828 $13,177 $17,211 $19,291 $16,008 $15,415 $17,029 $16,581 10% Mobile Processors NVDA $3,543 $3,998 $4,280 $4,130 $4,682 $5,010 $6,910 $9,714 $11,716 9% Graphics MediaTek $3,605 $2,956 $3,358 $4,586 $7,030 $6,718 $8,549 $7,832 $7,902 11% Mobile/Digital Home AMD (+ ATYT) $6,668 $6,748 $5,616 $5,481 $5,668 $4,111 $4,272 $5,253 $6,506 -5% Processors/Graphics XLNX $1,896 $2,437 $2,315 $2,256 $2,485 $2,447 $2,334 $2,485 $2,658 3% Programmable Logic MRVL $2,808 $3,612 $3,393 $3,169 $3,404 $3,707 $2,726 $2,393 $2,416 0% Networking/Connectivity ALTR $1,954 $2,064 $1,783 $1,733 $1,932 $1,971 $2,010 $2,050 $2,091 0% Programmable Logic Novatek $1,153 $1,193 $1,253 $1,397 $1,784 $1,602 $1,416 $1,548 $1,820 3% Driver ICs/Display Top global fabless $39,613 $44,605 $48,052 $53,158 $58,933 $54,346 $56,924 $65,969 $71,719 5% YoY Growth 13% 8% 11% 11% -8% 5% 16% 9% Top global fabless market share 69.5% 70.5% 70.6% 69.9% 69.4% 60.5% 64.3% 67.6% 69.0% China top 20 fabless $2,153 $3,768 $4,359 $6,357 $7,928 $9,601 $11,262 $13,731 $16,808 23% Overseas fabless revenue $54,833 $59,526 $63,688 $69,656 $77,045 $80,172 $77,288 $83,883 $87,197 4% Global fabless revenue $56,986 $63,294 $68,047 $76,013 $84,973 $89,773 $88,550 $97,614 $104,005 6% YoY Growth 11% 8% 12% 12% 6% -1% 10% 7% Source: Gartner, PwC, Credit Suisse estimates IC design traction varies by application China has outgrown the global industry, although at a 16% market share in fabless and with no China has had more major established IDMs joining manufacturing and design, the success and market penetration success in mobile, varies by application with a number of major semiconductor categories barely addressed by the networking and consumer local supply chain.

Figure 40: China has grown its IC design share to 16% in 2018

Fabless sales (US$mn) Top China fabless market share (%) $100,000 30% $90,000 27% $80,000 24% $70,000 21% $60,000 18% $50,000 15% $40,000 12% $30,000 9% $20,000 6% $10,000 3% $0 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 Top 20 China fabless Overseas fabless revenue Top China fabless market share

Source: Company data, Credit Suisse estimates Areas with reasonable success for Chinese companies

■ Mobile communications. IC design in China is led by Huawei’s chip division HiSilicon, which alone represents 35% of China’s fabless sales. HiSilicon currently supplies about 70% of Huawei’s smartphones including the mid-high-end and its newly launched Mate 30 Asia Technology Strategy 38

4 November 2019

5G smartphone with internal Kirin 990 modem built on TSMC 7nm+ as the first high volume customer on that node. Huawei has also internalised its RF transceiver, low-end power amplifier and connectivity to lessen reliance on the US supply chain, following its placement on the entity list. Elsewhere in mobile chipsets, UniSOC (formerly Spreadtrum) has somewhat lagged on 4G after developing a competitive presence on 2G/3G, although it is still pushing ahead with commercialising its 12nm modem in 1H20 and developing its first 5G SoC on 7nm by the end of 2020. UniSOC also acquired RDA which also gives it capability on the connectivity and RF. Beyond these mobile chipset providers, China has developed a competitive footprint in other mobile peripherals including CMOS image sensors (Omnivision, GalaxyCore), fingerprint IC (Goodix, Silead) and RF (Vanchip, RDA).

■ Networking. HiSilicon also leads China’s networking IC sector with internal supply to Huawei’s base stations and infrastructure including its own high-end ASIC which is displacing some of the FPGA content. ZTE Microelectronics and Datang have also emerged as infrastructure suppliers to their parent companies and Montage has developed a leading memory interface solution for cloud computing systems.

■ Consumer electronics. Lower entry barriers for mature high volume consumer products have also enabled entry of a number of Chinese designers, including several application processor suppliers that initially reached a high volume on tablets and are currently moving into connected audio and video and IoT applications including Rockchip, Allwinner, Actions, AM Logic, JieLi and BES.

■ Crypto-currency/AI inference. China’s local IC design companies including Bitmain, Pansemi and eBang commanded 95%+ of the crypto-currency ASIC market. The companies along with start-up Cambriconare currently using their processing and low power design capability and software algorithm work in crypto-currency into AI inference processors for smart buildings, smart home and smart city applications.

■ Flash memory. China has worked with several sources of IP for flash memory. Gigadevice produces NOR flash and NAND flash through SMIC and Huali. YMTC is also developing 3D NAND flash by leveraging multiple international chipmakers’ NAND flash IPs. Integrated Silicon Solution Inc (ISSI) is also being acquired by Ingenic. These together will bring China’s capacity in flash memory. Ingenic is in the process of acquiring ISSI. YMTC also has recently showcased a 64-layer sample for its own SSD for its volume production planned in 1H20. ChangXin (CXMT) has also been sampling its 19nm DRAM and has an initial volume production planned in early 2020.

■ Smart card ICs. China is developing its local smart card IC vendors to embed flash on a secure card for use in transportation, social security/National ID cards, and financial IC cards. The company has qualified six vendors for mobile payments, including Tongfang Guoxin, Datang, Huada, NationZ and Hua Hong IC. Despite success in these high volume markets in mobile, networking and consumer, China is China's share is still low in a number of large chip segments including processors, analog, still a small player in a number of strategic and high value areas in the industry, with low-single- graphics and sensors digit share in a number of critical semiconductor categories including microprocessors, graphics processors, , programmable logic, MEMs sensors, analog, automotive components, memory, storage ICs and display drivers. We estimate that these markets where China has less than 5% market share represent about US$333 bn of sales, and over 70% of area in the semiconductor industry for China to target through building up its in-house teams and supplementing with hiring overseas talent, M&A or equity stakes to accelerate development.

Asia Technology Strategy 39

4 November 2019

Figure 41: China’s success higher in mobile, consumer and wired applications, low in analog, RF processors, memory, FPGAs and GPUs

US$mn China share (%) $100,000 20% $90,000 18% $80,000 16% $70,000 14% $60,000 12% $50,000 10% $40,000 8% $30,000 6% $20,000 4% $10,000 2%

$0 0%

RF

DSP

MPU

MCU

PMIC

GPUs

Wired

FPGA

NAND

DRAM

Analog

Sensors

Discretes

DriverIcs

Connectivity

DiscreteAPs

CMOS sensors

ASICs Crypto&

Discretemodems Mobile processors Overseas China China Share

Source: Company data, Credit Suisse estimates Areas where Chinese companies still lag . CPU – dominated by Intel and AMD. The processor market is dominated by Intel and AMD, with their x86 architecture as the basis for running the PC and server operating systems. China’s opportunity in the core processor is to leverage ARM’s core for servers tailored to China domestic applications. Huawei’s Kunpeng server chipset is the highest performing ARM core and may be its solution to mitigate the impact from its ban on US technology. Zhaoxin has also licensed the x86 core as a JV with Via and the Shanghai Municipal Government for a home-grown processor based on Intel’s well-established x86 instruction set for compute. is also targeting MIPs 64 bit CPUs for general purpose computing. . FPGA – Xilinx and Altera/Intel lead. Xilinx and Altera have over 80% market share, followed by Microchip (acquired Microsemi/) and Lattice each with about 5% market share, and negligible share for the Chinese players. The FPGA market has higher barriers due to the software platforms used to programme the FPGAs and now increasing amount of embedded IP and requirement to manufacture high-end FPGAs at large die size (more difficult to use) on the advanced technology nodes. Several Chinese players including Gowin, Fudan Micro, Unigroup Guoxin and Huada are developing FPGAs. . Analog/Discretes – Chinese design houses focusing initially on high volume IT. China is following the route of Taiwan analog companies a decade prior by focusing on high volume IT markets initially in computing, mobile, consumer/IoT, display and LEDs and later focusing on auto/industrial. China has several emerging companies including SG Micro, Silan, and Silergy and ZTE’s Microelectronics (SaneChips) that are gaining traction and growing both as a lower cost alternative for global companies or a local source domestically in China. In standard products, Wingtech’s purchase of Nexperia’s standard products group also gives it a high position in discrete. . Graphics – barriers high, a couple of Chinese design houses trying to address. The graphics has had high barriers as both NVIDIA and AMD have built their GPUs as platforms for gaming, in close collaboration with the leading game developers and in extensive marketing to build up the consumer base for their systems. The graphics requires a long design cycle, large die size, complex architecture and advanced manufacturing and marketing capabilities, raising the barriers to entry. China has a couple of IC design companies working on graphics including Jingjia Micro developing a high performance GPU Asia Technology Strategy 40

4 November 2019

starting with 28nm (vs NVIDIA on 12nm and AMD on 7nm). Zhoxin has also integrated Via’s GPU technology for an integrated processing unit with graphics.

Figure 42: China’s 2018 market share was still relatively low in many of the large semiconductor categories Market Size US$mn Overseas players Share Chinese players Share Market Size US$mn Overseas players Share Chinese players Share Samsung Electronics 42.8% YMTC <1% Robert Bosch 13.5% Goodix 4% SK hynix 29.5% CXMT <1% Infineon Technologies 7.8% Silan <1% DRAM 99,872 Micron Technology 22.9% Fujian Jinhua <1% STMicroelectronics 7.8% Shanghai Belling <1% Sensors 10,489 Nanya Technology 2.8% TDK 4.9% Yangjie <1% Overseas subtotal: 98.0% Chinese subtotal: <1% Knowles 4.4% Samsung Electronics 37.8% YMTC <1% Overseas subtotal: 38.4% Chinese subtotal: <5% Western Digital 15.5% GigaDevice <1% Qualcomm 54.0% HiSilicon Technologies 15% Toshiba Memory 14.6% ISSI / Ingenic <1% Mobile MediaTek 13.8% UniSoC Technologies 4% 14,829 11.1% Processor Samsung Electronics 13.8% Allwinner <1% NAND 57,951 SK hynix 11.0% Overseas subtotal: 81.6% Chinese subtotal: <20% Toshiba 4.9% NVIDIA 81.4% Jingjia Micro <1% Intel 4.1% GPUs 8,262 AMD 18.1% Zhaoxin <1% Overseas subtotal: 98.9% Chinese subtotal: <1% Overseas subtotal: 99.5% Chinese subtotal: <1% Intel 91.2% HiSilicon <1% Broadcom 26.9% HiSilicon Technologies <1% AMD 5.4% Cambricon <1% Qualcomm 18.6% UniSoC Technologies <1% NXP 0.8% Alibaba <1% MediaTek 6.7% Maxscend <1% MPUs 54,151 Connectivity 11,362 Texas Instruments 0.5% Montage <1% NXP 5.1% Goodix <1% Marvell 0.8% Eeasy Tech <1% Cypress Semiconductor 4.0% Will Semi <1% Overseas subtotal: 98.8% Chinese subtotal: <1% Overseas subtotal: 61.3% Chinese subtotal: <1% Renesas Electronics 18.2% CEC Huada 1% Broadcom 36.4% Sanechips Technology 7% NXP 17.4% Gigadevice <1% Intel 9.2% Microchip Technology 14.1% Unigroup Guoxin <1% Marvell Technology Group 8.4% Wired 22,122 MCUs 19,211 STMicroelectronics 11.2% SinoWealth <1% HiSilicon Technologies 7.0% Texas Instruments 9.7% Silan <1% Realtek Semiconductor 3.0% Infineon Technologies 9.0% Goodix <1% Overseas subtotal: 64.0% Chinese subtotal: 7% Overseas subtotal: 79.7% Chinese subtotal: <2% Skyworks Solutions 22.4% Sanechips <1% Texas Instruments 41.7% Qorvo 17.3% Vanchip <1% NXP 22.6% Qualcomm 14.7% Xinyi Semi <1% DSP 1,453 Analog Devices 21.6% RF 11,831 Broadcom 13.9% Smarter Micro <1% Overseas subtotal: 86.0% Chinese subtotal: <1% NXP 5.2% Lansus <1% Xilinx 51.1% Unigroup Guoxin <1% Murata Manufacturing 4.4% UniChip <1% Intel 35.8% Fudan Micro <1% Overseas subtotal: 77.8% Chinese subtotal: <1% FPGA 5,683 Microchip Technology 5.2% Huada <1% Texas Instruments 15.7% HiSilicon Technologies 3% Lattice Semiconductor 5.0% Gowin <1% Qualcomm 9.8% Silergy 3% Overseas subtotal: 97.1% Chinese subtotal: <1% Dialog Semiconductor 8.4% SG Micro <1% PMIC 11,431 Samsung Electronics 28.4% Huada 3% Infineon Technologies 6.7% Shanghai Belling <1% Novatek 16.1% SinoWealth <1% STMicroelectronics 6.4% Silan <1% Himax Technologies 9.5% Solomon Systech <1% Overseas subtotal: 47.0% Chinese subtotal: <10% Driver IC 6,060 Silicon Works 9.1% Chipone <1% Intel 27.5% Bitmain Technologies 5% Synaptics 6.7% STMicroelectronics 4.6% HiSilicon Technologies 4% Overseas subtotal: 69.9% Chinese subtotal: <5% Infineon Technologies 3.6% Montage Technology 1% Texas Instruments 29.0% ASICs 24,801 ON Semiconductor 3.0% Analog Devices 20.3% 2.8% Maxim Integrated 4.6% Texas Instruments 2.7% Analog 24,401 ON Semiconductor 4.1% Overseas subtotal: 44.2% Chinese subtotal: <15% Cirrus Logic 3.0% Apple 28.7% HiSilicon 2% STMicroelectronics 2.5% MediaTek 11.8% Unisoc N/A Overseas subtotal: 63.5% Chinese subtotal: <1% Broadcom 7.6% Infineon Technologies 14.4% Nexperia 5% Discrete AP 25,264 AMD 6.9% ON Semiconductor 10.4% Jilin Sino-Micro 1% NVIDIA 5.3% STMicroelectronics 5.9% HiSilicon <1% Intel 3.4% Mitsubishi 5.6% SG Micro <1% Overseas subtotal: 63.8% Chinese subtotal: <5% Discretes 22,731 Vishay 5.4% Silan <1% Intel 39.3% UniSoC Technologies 9% Rohm 5.2% Shanghai Belling <1% Discrete Qualcomm 37.0% 4,153 Toshiba 5.0% Will Semi <1% modem MediaTek 6.1% Overseas subtotal: 51.9% Chinese subtotal: <10% Overseas subtotal: 82.5% Chinese subtotal: <10% Sony 47.0% Will Semiconductor 11% Samsung Electronics 20.1% Galaxycore 3% CMOS 12,281 ON Semiconductor 5.9% BYD Microelectronics <1% sensors SK hynix 2.3% SmartSens <1% Overseas subtotal: 75.2% Chinese subtotal: <1% Source: Gartner, Company data, Credit Suisse estimates . MCU – a broad fragmented market dominated by global IDMs makes penetration slow. China has a low-single-digit share in the MPU market dominated by global IDMs including Renasas, NXP, Microchip, STM, TI and Infineon. The MCU customer base is in thousands and relies on a large catalog of products and takes time to build up the customer base. China has multiple MCU suppliers including Gigadevice, Ingenic, Silan, Datang and Huada and can use Hua Hong/SMIC’s embedded flash capability. The MCU companies can target the large domestic China market looking for a domestic source and also some of the higher volume markets such as home appliances and toys.

Asia Technology Strategy 41

4 November 2019

. RF – advanced specialty technology requires an experienced team. RF design is increasingly complex supporting multiple bands across 2G-5G, integration of multiple components and use of specialty compound semiconductor materials requiring experienced radio frequency chip designers. China’s RF suppliers (HiSilicon, RDA, Maxscend, Vanchip) are now competitive on 2G and 3G but only supply the entry-level 4G. The supply chain will also rely on Murata (PA filters) and Taiyo Yuden (BAW filters) to mitigate risk of US technology exposure. . Display drivers. Taiwan and Korea lead in driver ICs with established position from years of supplying their domestic panel makers. Market share is led by long-time suppliers Novatek, Himax, Synaptics, FocalTech, Magnachip and Samsung’s internal division. China does have a few suppliers seeking entry including SinoWealth, Galaxycore, Solomon Systech and Chipone, though it lags in capacity support (at times tight on 8”) and panel makers’ confidence relative to the established suppliers already with a large purchasing scale and design advantage. Foundry: China lags on advanced capacity, more competitive on the mature nodes Current market structure China has been committed to the foundry sector for more than a decade, with Hua Hong founded in the late 1990s and SMIC and Grace Semiconductor in 2000 charged with creating local champions for contract semiconductor manufacturing. Despite consistent government support, several industry cycles of investment and management teams, the China foundry sector's share in 2018 remains 9.5%, virtually unchanged from the 9.6% share it held in 2008, although has clawed back 170 bp of share since bottoming in 2015 at 7.8%.

Figure 43: China foundries led by SMIC and Hua Hong, but still modest in the global picture CAGR Foundry 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2011-2018 SMIC $1,354 $1,070 $1,554 $1,319 $1,702 $2,069 $1,970 $2,229 $2,914 $3,101 $3,351 14.2% Hua Hong (plus Grace) $513 $357 $620 $610 $572 $585 $665 $651 $721 $808 $929 6.2% CSMC $142 $151 $225 $230 $212 $210 $341 $328 $418 $456 $492 11.5% Shanghai Hauli $0 $0 $0 $0 $37 $111 $305 $280 $310 $587 $676 NM XMC $0 $0 $0 $0 $0 $180 $210 $200 $250 $250 $300 NM ASMC $136 $95 $130 $124 $137 $128 $132 $120 $116 $124 $180 5.5% Shanghai Belling $23 $14 $20 $20 $13 $15 $15 - - - - NM China foundry sales (US$mn) $2,168 $1,687 $2,549 $2,303 $2,673 $3,298 $3,638 $3,808 $4,730 $5,326 $5,928 14.5% YoY Growth -22.2% 51.1% -9.7% 16.1% 23.4% 10.3% 4.7% 24.2% 12.6% 11.3% China foundry share (%) 9.6% 8.4% 9.0% 8.0% 7.9% 8.4% 7.9% 7.8% 8.8% 9.0% 9.5% TSMC $9,575 $8,396 $12,141 $12,359 $14,621 $17,213 $22,396 $25,100 $29,451 $32,155 $34,208 15.7% GlobalFoundries $1,743 $2,422 $3,662 $3,005 $4,013 $4,121 $4,352 $5,016 $5,495 $6,176 $6,209 10.9% UMC $3,188 $1,786 $1,873 $2,484 $3,596 $2,900 $3,203 $3,261 $4,579 $4,909 $5,021 10.6% Samsung Foundry $515 $486 $390 $770 $1,295 $3,200 $3,339 $2,621 $3,700 $4,475 $4,721 29.6% Japanese IDMs $1,265 $972 $1,095 $992 $940 $954 $1,111 $1,261 $1,291 $1,289 $1,318 4.1% Powerchip Logic $190 $140 $149 $431 $614 $862 $917 $985 $987 $1,057 $1,152 15.1% TowerJazz $436 $298 $509 $613 $639 $505 $828 $961 $1,249 $1,387 $1,303 11.4% Vanguard $490 $400 $546 $514 $579 $697 $755 $736 $802 $819 $959 9.3% Dongbu $432 $370 $512 $483 $478 $464 $540 $590 $667 $611 $615 3.5% IBM $566 $383 $500 $545 $634 $495 $519 $280 - - - NM Magnachip $310 $265 $410 $338 $390 $400 $400 $291 $274 $320 $327 -0.5% X-Fab $368 $211 $317 $278 $297 $290 $310 $290 $461 $582 $588 11.3% Silterrra $157 $160 $180 $180 $160 $172 $160 $150 $130 $120 $190 0.8% Other $1,185 $2,165 $3,473 $3,460 $2,789 $3,778 $3,385 $3,542 $184 $131 $133 -37.2% Global foundry sales (US$mn) $22,588 $20,141 $28,305 $28,754 $33,716 $39,349 $45,852 $48,891 $53,999 $59,357 $62,673 11.8% YoY Growth -10.8% 40.5% 1.6% 17.3% 16.7% 16.5% 6.6% 10.4% 9.9% 5.6% Source: Company data, Gartner, Credit Suisse estimates

Asia Technology Strategy 42

4 November 2019

Key barriers to entry

■ High capex requirements. TSMC's capex at US$10-12 bn per year is a 5x multiple over SMIC's recent US$2-2.5 bn range. Equipment cost for the most advanced leading-edge capacity is now US$150-200 mn per 1,000 WPM capacity, limiting SMIC to 10-15,000 capacity additions per year at its current budget, paling in comparison to TSMC's installed 400,000 wafers per month capacity at 28nm and below. While SMIC has used its Shanghai and Beijing JVs to share the capex and initial losses, the capex, even including those JV investments, still implies that the scale gap only widens in the coming years with TSMC continuing to outspend. The higher scale base also translates to a wide difference in revenues. SMIC's revenue base is still only 8% of TSMC's and is also playing out in investment capability, with operating cash flows also only 6% of TSMC's operating cash flow and capex staying around 20% of TSMC's overall capex. At a US$2-2.5 bn budget, SMIC can add about 15,000 14nm wafers, enough to generate US$500 mn on the advanced node, still less than 1.3% of TSMC's projected 2020 sales.

Figure 44: SMIC <10% of TSMC's sales and cash flow Figure 45: TSMC capex is much higher than that of SMIC's US$mn (%) Capex US$mn Capex/Sales (%) 30,000 18% 14,000 100% 16% 90% 25,000 12,000 14% 80% 20,000 12% 10,000 70% 10% 60% 15,000 8,000 8% 50% 10,000 6% 6,000 40% 4% 30% 5,000 4,000 2% 20% 2,000 0 0% 10%

0 0%

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019E

2020E

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019 2020 SMIC Operating Cashflow TSMC Operating Cashflow SMIC OCF as % of TSMC OCF SMIC Sales as % of TSMC Sales TSMC Capex SMIC Capex TSMC Capex/Sales SMIC Capex/Sales Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates

■ Capacity additions committed from the government, but not yet needed. Contrary to memory where fabs are kept loaded to minimise unit costs, foundry requires specific wafer demand from unique customers that have qualified product, all subject to having effective and competitive capacity. SMIC has available government commitments for it to have another 35k Beijing JV capacity, 70k Shanghai JV capacity, and 35k available in Shenzhen with only 10k built out as management will only ramp up as customer commitments come through. Our tracker of China capacity shows only 191k WPM of capacity built out by foundries in China, out of the 565k available capacity planned.

Asia Technology Strategy 43

4 November 2019

Figure 46: China has sizeable committed capacity, but only gradually building it out Max Company Location Node 4Q17 4Q18 4Q19 Status Capacity SMIC Beijing 90nm-28nm 46.0 42.0 50.0 50.0 Fully built SMIC Shanghai 45-28nm 17.0 10.0 8.0 20.0 Fully built SMIC Beijing 2A JV 40nm 29.0 33.0 36.0 35.0 Ramping SMIC Beijing 2B JV 40nm 0.0 0.0 0.0 35.0 Ramping SMIC Shenzhen 90-40nm 6.8 6.8 6.8 35.0 Pilot Line SMIC Shanghai JV 28-14nm 0.0 0.0 6.8 70.0 Future SMIC capacity build-out in China: 98.8 91.8 107.5 245.0 Hua Hong/Huali Shanghai F5 55-40nm 35.0 35.0 35.0 35.0 Fully built Hua Hong/Huali Shanghai F6 28nm 0.0 0.0 0.0 40.0 Construction Hua Hong Wuxi F7 90-65nm 0.0 0.0 5.6 40.0 Planning TSMC Nanjing 16nm 0.0 10.0 10.0 80.0 Starts 2Q18 UMC Xiaman: F12X 40-28nm 11.5 17.0 17.0 50.0 Ramping Powerchip 90nm 0.0 6.0 6.0 40.0 Starts 2018 GlobalFoundries 22nm FDS 0.0 10.0 10.0 65.0 Starts 4Q18 Yuexin Guangzhou 40-28nm 0.0 0.0 0.0 30.0 Starts 2019 Other foundries capacity build out in China 46.5 78.0 83.6 380.0 Total capacity build out in China 145.3 169.8 191.1 625.0

Source: Company data, Credit Suisse estimates

■ Technology gap remains high. Technology and scale shows up as a sizeable gap both on mature and leading nodes. SMIC has done quite well to fully load 8" and its mature 12” capacity with specialty and 2nd wave applications (fingerprint IC, smart cards, power management, image sensors, NOR flash, RF transceivers) but is still only 20% of TSMC's revenue base. On the advanced nodes, SMIC has only 3% of sales on 28nm and no sales below that node, keeping its sales base limited in areas that account for 70% of TSMC's sales. SMIC is no longer aggressively targeting 28nm expansion due to low profitability, although will expand its 3k WPM pilot line on 14nm to 15k WPM over the next year as customer orders ramp up, giving that node a chance to scale to US$500 mn sales, though still only one-sixth the size of TSMC’s 16nm capacity and not able to access the 110k capacity being built on 7nm and 50k on 5nm.

Figure 47: TSMC's revenue on legacy nodes is still much Figure 48: TSMC has a significant leading-edge scale higher even with SMIC’s progress advantage over SMIC US$mn US$mn 3,500 30,000 3,000 25,000 2,500 20,000 2,000 15,000 1,500 10,000 1,000

500 5,000

0 0

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018 2019 TSMC - 0.18/0.15um SMIC - 0.18/0.15ium TSMC - 65nm and below SMIC - 65nm and below Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates

■ Ecosystem development has posed a barrier. The challenges of ramping up a new node every two years and increasingly a half node every year requires development across equipment, materials, design software (EDA), IP developers, foundry and customers, keeping an advantage for the incumbents ahead on technology. TSMC has noted a rising number of titles now approaching 21,000 IP blocks qualified on its manufacturing platform and a stepped up rise in IP qualification, architecture design and system verification. The Chinese foundries are working to put in place that ecosystem, but the size of revenue base

Asia Technology Strategy 44

4 November 2019

with TSMC at US$35 bn vs SMIC at US$3 bn places a huge gap in resources committed, even with the government and IC fund backing and subsidy support.

Figure 49: TSMC qualifying a large IP pool on its process Figure 50: TSMC has built a wide ecosystem of partners

Source: Credit Suisse (TSMC – 2018 ARM Forum) Source: Credit Suisse (TSMC – 2018 ARM Forum) TSMC captures a large share of the opportunity China IC design growth helps the foundry sector, though TSMC capturing high share The foundry opportunity from domestic IC design companies in China has increased at a 23% CAGR to US$8.5 bn, representing almost 15% of the foundry market and an opportunity for all the players. Due to its scale and leadership on advanced capacity, TSMC has grown even faster, increasing its share of China local customers from 20% in 2010 to over 60% in 2018. We estimate TSMC’s China sales currently at US$5.5 bn vs US$1.8 bn for SMIC and US$500 mn for Hua Hong.

Figure 51: TSMC gaining share at the Chinese fabless Figure 52: China foundry share flat due to TSMC’s lead

Foundry sales in China (US$mn) Foundry sales (US$mn) China foundry market share (%) $10,000 70% $80,000 24% $9,000 $70,000 21% 60% $8,000 $60,000 18% $7,000 50% $50,000 15% $6,000 40% $40,000 12% $5,000 23.2% CAGR from 2010-2018 $30,000 9% $4,000 30% $3,000 20% $20,000 6% $2,000 $10,000 3% 10% $1,000 $0 0% $0 0% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 China foundry sales (US$mn) Global foundry sales (US$mn) China foundry share (%) TSMC SMIC UMC Hua Hong TSMC's China share Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates TSMC also has a 12” fab to address the China opportunity TSMC is also building out a fab in Nanjing China in four phases. The first of four phases is TSMC currently investing in capable of ramping up to 20k WPM at an initial cost of US$3 bn. The company will also have a a 12" fab in China design service centre to support local IC design companies. The revenue contribution will be modest at only 4% of sales once built out, but can allow TSMC to stay competitive addressing local customers for its 16nm FinFET process as SMIC starts its Shanghai line for its 14nm in 4Q19.

Asia Technology Strategy 45

4 November 2019

Figure 53: China foundry share small vs its opportunity Figure 54: China trying to build a locally sufficient chain

Source: SMIC Source: SMIC China players SMIC/HuaHong invest to raise competitiveness, TSMC still the lead supplier SMIC – high spending mode to improve its technology position. China’s domestic foundries still have about 9% market share, led by SMIC and Hua Hong. SMIC is in a high investment phase, reversing course from its strategy a few years back to fill its mature 12” and 8” with differentiated technology and gradually migrate to the advanced nodes. The company has hired Dr. Liang, formerly at TSMC and Samsung, to improve its foundry process and R&D capability. Since the stepped up efforts, it is spending at a high capital intensity and R&D level and growing its ability to receive government R&D grants to advance its process towards 14nm and n+1 technology (~8nm) to target more advanced mobile and high performance computing applications. The company’s profitability is only slightly positive even with the government support, with management conceding a need to take time for sustainable solid profitability or cash flow, but investing for a long-term improvement in technology competitiveness that can lift its performance long term. Hua Hong – focused on specialty technology on mature 8”/12”, privately held Huali pushing advanced technology. China's other top foundry Hua Hong has carved out a role in China's development as a specialty foundry focused on leveraging more mature technology on 8” and 12”. The foundry achieved good scale on 8” foundry manufacturing through the merger of Grace and Hua Hong Semiconductor with differentiated specialty technology and particular capabilities both in applications using embedded flash (over 40% of sales) and in supplying domestic Chinese customers (over 50% of sales). The company’s technology roadmap focuses production on a wide range of specialty applications including microcontrollers, power management, RF, smart cards, discretes and MEMS. The company has licensed Super Flash from SST, a Microchip subsidiary, and SONOS from Cypress to develop a competency in eNVM (embedded Non-Volatile Memory) used in smart cards (SIM cards, bank ICs, mobile payments, ID cards, social security cards) at 35% of sales and targeted to reach 50% of sales long term.

Asia Technology Strategy 46

4 November 2019

Figure 55: SMIC sales from China customers Figure 56: Hua Hong sales from China customers US$mn US$mn $600 $160 $140 $500 $120 $400 $100 $80 $300 $60 $200 $40 $20 $100 $0

$0

Jun-14

Jun-15

Jun-16

Jun-17

Jun-18

Jun-19

Mar-14

Mar-15

Mar-16

Mar-17

Mar-18

Mar-19

Sep-14

Dec-14

Sep-15

Dec-15

Sep-16

Dec-16

Sep-17

Dec-17

Sep-18

Dec-18

1Q09

3Q09

1Q10

3Q10

1Q11

3Q11

1Q12

3Q12

1Q13

3Q13

1Q14

3Q14

1Q15

3Q15

1Q16

3Q16

1Q17

3Q17

1Q18 3Q18 1Q19 China Japan Asia Europe North America China Eurasia Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates Hua Hong is now moving ahead more aggressively on the first 10k WPM of its new 12” fab for first volume production late this year and maintains plans to ramp up 40k WPM 12” fab by 2022 to double capacity to US$2 bn. Its sister company Huali is also building a 2nd advanced 12” fab with an eventual capacity for 40k for 28-14nm.

Figure 57: Hua Hong's technology portfolio Figure 58: Hua Hong has grown embedded flash (eNVM) US$mn $250

$200

$150

$100

$50

$0

Jun-14

Jun-15

Jun-16

Jun-17

Jun-18

Jun-19

Mar-14

Mar-15

Mar-16

Mar-17

Mar-18

Mar-19

Sep-14

Dec-14

Sep-15

Dec-15

Sep-16

Dec-16

Sep-17

Dec-17

Sep-18 Dec-18 Embedded flash Logic & RF Discrete Analog & PM Standalone NVM Others Source: Hua Hong Semiconductor Source: Hua Hong Semiconductor Aside from the leading players, China is also investing in several communal IDMs which would combine manufacturing, chip design and partnerships with other IDMs and IC design companies to address growing market opportunities in the industry. Dr. Chang, former SMIC founder and Chairman, is building up a commune IDM in Qingdao China. The company will form a partnership with system companies and chip designers to target an application to develop a chip to jointly sell to the customer base. Requests are market driven, coming from end users and manufacturers for SiEn to design, develop and produce the wafers for customers’ needs. In the early 1990s, HP and Canon had collaborated with TI and the Economic Development Board (EDB) to form a TECH company on the same business model. SiEn plans to build 1 12” fab, 1 8” fab and a photomask shop. The 8” fab will ramp up initially to 30k WPM and 12” fab will ramp up to 40-45k WPM, with the first phase set at 10k WPM. The company plans pilot start from August 2019-December 2019 and production start during 2020-21. The company also has land to develop living quarters and a school in an adjacent area to assist in talent retention. Staff levels are up to 300 as of March 2019, with 30% having PhD and Masters’ degree and 48% bachelor’s degree. The company has started up projects for Microcontrollers, OLED Driver ICs, IoT chips for home appliances, and analog/power management chips for industrial electronics. The company believes that several global IDMs will partner with it for some new chip designs.

Asia Technology Strategy 47

4 November 2019

Back-end: China has used M&A to gain a stronger presence Current market structure Since 2013, the back-end sector has seen the faster emergence of China OSAT (Outsourced semiconductor assembly and test) players following the local players’ acquisition of overseas companies to gain capability and customers. Some of the key M&As in the sector included JCET’s acquisition of Stats-Chippac, which moved its market share up from 3% to 10%, Tongfu’s acquisition of AMD’s factories which gave it a high growth CPU customer and advanced flip chip packaging and final test capabilities, and Tianshui Huatian closed on Flip Chip International for flip chip bumping and wafer level packaging. The wave of acquisitions in the 2013-15 period lifted the back-end companies’ market share from 5% in 2012 to 20% in 2019, making it a competitive force and source of self-sufficiency on mainstream and reasonably advanced technology (flip chip, fan-out packaging) for China. The one area of lag is still in the high-end advanced packaging, where TSMC has also emerged over the past five years by supplying the high density fan-out (InFO) process for Apple’s application processor and CoWoS (silicon interposers) for advanced graphics integrating high bandwidth memory and high-end FPGAs.

Figure 59: China accelerated share gains with M&A in 14-15, though now stabilising a bit Back-end sales (US$mn) China % of Back-end $35,000 35%

$30,000 30%

$25,000 25%

$20,000 20%

$15,000 15%

$10,000 10%

$5,000 5%

$0 0%

2007 2008 2009 2011 2012 2014 2015 2016 2017 2019 2010 2013 2018 China back-end Total back-end sales China % of back-end

Source: Company data, Credit Suisse estimates The Chinese OSATs after their aggressive M&As were initially aggressive to expand capacity and gain share even at the expense of pricing and margins operating near break-even levels and helping trigger several points of gross margin erosion for the whole sector. Due to that expansion and also taking on some high dollar value but low margin SiP projects, China companies have grown at a 25% CAGR (19% excluding M&A) to reach 20% market share.

Asia Technology Strategy 48

4 November 2019

Figure 60: China’s back-end sector has grown at a 25% CAGR to reach 20% share in 2019 (+20% CAGR excluding M&A) Sales comparison 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 '06-'10 CAGR 10-'18 CAGR JCET $311 $343 $347 $535 $582 $703 $830 $1,043 $1,713 $2,875 $3,540 $3,607 $3,226 21% 27% Tianshui Huatian $105 $107 $114 $172 $203 $257 $399 $536 $616 $824 $1,039 $1,081 $1,228 20% 26% Tongfu Micro $167 $171 $181 $255 $251 $252 $288 $339 $369 $689 $967 $1,093 $1,225 13% 20% China back-end $583 $621 $642 $962 $1,035 $1,213 $1,517 $1,918 $2,698 $4,388 $5,545 $5,781 $5,679 18% 25% YoY Growth 18% 6% 3% 50% 8% 17% 25% 26% 41% 63% 26% 4% -2% China % of back-end 3% 3% 4% 4% 4% 5% 6% 7% 9% 15% 18% 18% 19% China organic back-end $583 $621 $642 $962 $1,035 $1,213 $1,517 $1,918 $2,253 $2,951 $3,944 $4,121 $4,092 18% 20% YoY Growth 18% 6% 3% 50% 8% 17% 25% 26% 17% 31% 34% 4% -1% ASE $3,081 $3,007 $2,608 $3,992 $4,345 $4,401 $4,829 $5,273 $4,873 $4,982 $8,035 $8,230 $8,040 7% 9% YoY Growth 0% -2% -13% 53% 9% 1% 10% 9% -8% 2% 61% 2% -2% % of back-end 15% 14% 15% 17% 18% 18% 19% 19% 17% 17% 26% 26% 26% Amkor $2,739 $2,659 $2,179 $2,939 $2,776 $2,760 $2,956 $3,129 $2,885 $3,894 $4,186 $4,316 $3,910 2% 5% SPIL $1,968 $1,925 $1,729 $2,026 $2,084 $2,188 $2,336 $2,743 $2,613 $2,640 N/A N/A N/A 4% N/A Powertech $744 $993 $908 $1,202 $1,342 $1,407 $1,267 $1,321 $1,339 $1,500 $1,960 $2,072 $2,303 23% 7% TSMC $246 $318 $270 $466 $654 $772 $1,006 $1,384 $1,596 $1,913 $2,247 $2,496 $2,548 18% 23% UTAC $756 $711 $601 $925 $981 $978 $748 $734 $878 $875 $874 $888 $901 10% -1% ChipMOS $698 $522 $352 $546 $620 $651 $652 $726 $627 $571 $589 $613 $649 -3% 1% Chipbond $178 $168 $159 $399 $450 $508 $533 $584 $532 $536 $605 $621 $671 33% 6% KYEC $441 $465 $382 $561 $516 $496 $495 $537 $540 $623 $647 $820 $941 8% 5% J-Devices $474 $498 $523 $549 $576 $532 $825 $923 $822 N/A N/A N/A N/A 5% N/A Carsem $372 $370 $275 $394 $360 $356 $350 $336 $360 $378 $384 $389 $395 3% 0% Walton $315 $258 $256 $286 $279 $280 $299 $330 $249 $272 $310 $317 $324 -1% 1% Unisem $285 $373 $300 $433 $380 $354 $315 $319 $323 $340 $345 $350 $355 23% -3% Orient Semi $281 $327 $265 $357 $369 $360 $322 $456 $507 $490 $457 $467 $478 5% 3% AOI Electronics $246 $248 $211 $234 $304 $331 $343 $363 $392 $336 $414 $385 $394 0% 6% Formosa Advanced $274 $313 $276 $376 $404 $360 $302 $304 $276 $264 $259 $265 $271 17% -4% Kingpak $3 $4 $30 $96 $58 $42 $42 $40 $81 $73 $71 $78 $91 82% -3% Major Players (ex-China) $13,103 $13,157 $11,322 $15,782 $16,500 $16,777 $17,619 $19,502 $18,894 $19,687 $21,385 $22,307 $22,273 7% 4% YoY Growth 8% 0% -14% 39% 5% 2% 5% 11% -3% 4% 9% 4% 0% % of back-end 64% 62% 66% 67% 69% 68% 70% 72% 66% 66% 69% 71% 73% Other suppliers $6,914 $7,430 $5,186 $6,850 $6,489 $6,536 $5,946 $5,710 $6,895 $5,836 $3,878 $3,337 $2,530 1% -9% YoY Growth 5% 7% -30% 32% -5% 1% -9% -4% 21% -15% -34% -14% -24% Total back-end sales $20,600 $21,209 $17,150 $23,593 $24,024 $24,526 $25,082 $27,130 $28,487 $29,911 $30,808 $31,425 $30,482 5% 4% YoY Growth 7% 3% -19% 38% 2% 2% 2% 8% 5% 5% 3% 2% -3% Source: Company data, Credit Suisse estimates, (2014 revised to show the impact of JCET + Stats) In the past year, the China suppliers became a bit more selective on bidding for SiP projects, spending on capex and more restricted from large M&As on greater national protectionism, and therefore may have more limited share gains. Key opportunity would be to continue outgrowing the industry with the emergence of more local IC design companies. China’s key OSATs intend on growing their position We expect the China OSATs to continue to show outgrowth and target a combination of growing domestic opportunities and International customers seeking either more localised position or a lower priced solution: Figure 61: Comparison of ASE-SPIL with JCET and Tongfu US$mn Revenue GM OPM FCF Top customers & revenue % 2018 2018 2018 2018 3711.TW ASE/SPIL 13,187 17% 7% 70 Apple 19% (including USI module revenue) ASE/SPIL 8,114 21% 10% 178 QCOM 8%, Broadcom 7%, Huawei HiSilicon 6%, Mediatek 6%, NVIDIA 4%, Infineon 3%, STM 3%, ex-USI NXP/Freescale 3% 600584.SS JCET 3,608 11% 1% (300) Apple 23%, QCOM 7%, Broadcom 4%, Huawei HiSilicon 4%, Spreadtrum 3%, Bitmain/Cannon 3%, Samsung/Hynix 1-2% 002156.SZ Tongfu 1,092 16% 1% (230) AMD 45%, MediaTek 7%, Broadcom 3%, Huawei HiSilicon 2%, STM 2%, TI 2%, NXP 2% Source: Company data, Credit Suisse 1. JCET – targeting to be China’s strongest competitor and key global player. JCET through its acquisition of Singapore headquartered Stats-Chippac became #3 in back-end scale, after ASE/SPIL and Amkor. The company gained both advanced technology with the fan-out, flip chip and advanced test and global customers including Apple, Qualcomm and HiSilicon. The company had been growing revenue over the past few years, adding low margin SiP business to Apple, but is currently refocusing on improving its competitiveness and execution to better serve its International and local Chinese customers in the core assembly and test business. The company is looking to jump start sales as it is tracking flat to slightly down this year, with below average margins (12% GMs/1% OpMs).

Asia Technology Strategy 49

4 November 2019

2. Tongfu – advancing its capacity leveraging its AMD acquired assets. Tongfu Micro is leveraging the acquisition of AMD’s China back-end operations in 2016, giving it close to 50% of its revenue and more advanced operations, along with flip chip assembly and final test. The company is spending at a high R&D level to develop fan-out, driver IC, memory and packaging to support 7nm processors. Tongfu has been improving its customer base with good position at global customers Mediatek, Broadcom, Huawei, STM, TI and NXP. The company is currently seeing accelerated growth including with Huaweidue to the trade tension, Mediatek due to its share gains on low-mid-end smartphones, and recently qualified Infineon’s high-end automotive chips. The company is on pace to grow 15-20% on the back of growth from AMD and its better position at China and International customers, though margins are still relatively low at 13% GMs and 1% OpM in 2019. 3. TSHT - focused on mainstream packaging and better profitability. Tianshui Huatian (TSHT) traditionally has a higher level of profitability focusing on specialty ICs including RF, CMOS image sensors, NOR flash and sensors. The company also acquired Unisem (22% of sales in 1H19) which gave it access to international customers Qorvo, Skyworks and Broadcom/Avago. The company expects opportunities from domestic Chinese customers, replacing overseas headquartered OSATs ASE and Amkor. CS projects 15-20% YoY.

Funding of semi ambitions

Other than broad level and significant funds available from the National IC Fund, local private equity funds and private VCs (see earlier in the report), these semiconductor players in China also stand to gain funding and help from the following sources. JV fabs ease the investment burden for advanced manufacturing The China municipal funds have been helping the foundries build out capacity through JV fabs which allow the foundry operator to have operating control, but the JV contributes up to half of the capex outlay. During the first few years of fab investment, the high depreciation burden ensures losses but benefits the foundry P&L by sharing the loss of that operation with the JV partners. The contracts often allow the foundry to inject more capital over time to assume a larger ownership once the fab turns more profitable. Key JV fabs in the foundry sector include SMIC’s JV 12” fabs in Shanghai and Beijing, Hua Hong’s JV fab in Wuxi and UMC’s JV in Xiamen. GlobalFoundries had established a JV fab in Chengdu although we believe it is now up for sale with the foundry focusing more on generating positive free cash flow and paring off some of the more speculative investments.

Asia Technology Strategy 50

4 November 2019

Figure 62: Fab investment in China across logic IC and memory from domestic and foreign companies Origin Wafer Installed Next phase Final Capacity Company Fab Site Segment Node (nm) Country size (WPM) planned (WPM) China Fujian Jinhua Fujian 12" DRAM 25nm On Hold 0 60,000 China Hua Hong JV Wuxi 12" Foundry 90-55nm 10,000 10,000 40,000 China Huali Micro Pudong 12" Foundry 28-14nm 10,000 10,000 40,000 China CXMT Hefei 12" DRAM TBD 5,000 40,000 125,000 China SiEn Qingdao 8" Foundry 110nm+ 0 0 30,000 China SiEn Qingdao 12" Foundry 55/40nm 0 0 40,000 China SMIC JV Beijing 12" Foundry 40/28nm 35,000 0 35,000 China SMIC JV Shanghai 12" Foundry 14nm 3,000 12,000 70,000 China SMIC Shenzhen 12" Foundry 55nm 6,800 0 40,000 China Tsinghua Unigroup Nanjing 12" NAND/DRAM 2x nm TBD 0 100,000 China YMTC Wuhan 12" 3D NAND 3D NAND 20,000 60,000 300,000 Upcoming fabs by domestic companies 89,800 132,000 880,000 US Alpha & Omega Chongqing 8" Discretes 130nm+ 25,000 0 25,000 US GlobalFoundries JV Chengdu 12" Foundry 22nm FDS On Hold 0 65,000 Korea Hynix Wuxi C3 12" DRAM 1x nm 0 60,000 130,000 US Intel Dalian 12" NAND 3D NAND 80,000 20,000 100,000 Taiwan Powerchip Hefei 12" Logic 90nm 6,000 5,000 40,000 Korea Samsung Xian 12" 3D NAND 3D NAND 100,000 20,000 120,000 Taiwan TSMC Nanjing 12" Foundry 16nm 10,000 10,000 80,000 Taiwan UMC JV Xiamen 12" Foundry 28nm 17,000 0 50,000 Upcoming fabs by overseas companies 238,000 115,000 610,000 Source: Company data, Credit Suisse research R&D and equipment subsidies help manage the high cost of advanced technology development China also provides support in term of grants for the manufacturers tied to R&D projects and also investment grants that can offset interest expense. Our analysis of SMIC shows a steadily rising amount of government grants in the past five years since China’s National Guideline for the IC industry in 2014 and its Made in China 2025 programme were implemented. The company’s income statement recognition of grants for R&D, interest and other income has increased from US$40-50 mn range during 2011-15 to US$200mn in 2018. The funding for R&D, as the company also steps up meeting the national goal for FinFET commercialisation by 2020, has also accelerated, ramping up from US$250 mn on the balance sheet during 2013- 15 to over US$600 mn in 2018, with new funding up from US$$40 mn in 2015 to US$265 mn in 2018.

Figure 63: SMIC government grants rising since 2015 Figure 64: SMIC’s government funding of R&D rising

US$mn % of net income US$mn 250,000 250% 700,000

600,000 200,000 200% 500,000 150,000 150% 400,000

100,000 100% 300,000

50,000 50% 200,000

100,000 0 0% 2011 2012 2013 2014 2015 2016 2017 2018 0 P&L grants - other operating income P&L grants - interest expense 2011 2012 2013 2014 2015 2016 2017 2018

P&L grants - R&D expenses Grants as % of Net Income Government funding on the balance sheet New government funding on R&D projects

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates

Asia Technology Strategy 51

4 November 2019

Technology and IP from existing countries and companies supplemented by growing in-country talent China is also looking to follow in the footsteps of other countries that developed their semiconductor industry leveraging the insights of existing players. The electronics chain has had three major migrations of electronics’ value and production, with the US leading in the 1950s- 60s, a first migration to Japan and West Germany in the 1970s-80s, a second migration to the Asian Tigers in the 1990s-00s (South Korea, Taiwan), and a migration with evolution to China in the 2010s. The Chinese lead companies have been aggressively offering packages to existing engineers from established companies to accelerate their build-up of know-how. According to H&L Management Consultants, a recruitment firm, more than 300 senior engineers from Taiwan moved to mainland Chinese chip makers in 2018 and more than 1,000 since Beijing set up its National IC fund in 2014. Figure 65: Technology transfer through regions has been on-going since the 1950s

Source: Hua Hong Group China has doubled its gross expenditure on R&D from 1% to 2% in the past decade (closer to the 3-3.5% for the US and Korea) and currently is in the lead in granted and total patents. China is also now graduating 7.5mn citizens from college vs. 3.3mn in the US and may raise its global share of graduates from 18% in 2010 to 29% by 2020. In addition, China established the Thousand Talents program in 2008 to attract foreign researchers and incentivise the return of Chinese scientists from abroad with a background in STEM. Figure 67: China has overtaken the EU in R&D spend as a % of Figure 66: China has more applications and granted GDP (OECD data) patents than any other country

Source: Refinitiv, Credit Suisse Global Strategy research Source: Refinitiv, Credit Suisse Global Strategy research M&A activity another source of development, though now more difficult The Chinese supply chain went through an aggressive series of M&As through 2012-15 and some selective deals in the past few years to build up its capabilities in acquiring established Asia Technology Strategy 52

4 November 2019 foreign companies. The high degree of activity has attracted attention and also caused government regulators to slow the approvals of deals, including the US CFIUS on the grounds of national security concerns. Before the M&A window narrowed, China completed deals including RDA (RF/Connectivity/DTV), Spreadtrum (Mobile), Montage (Set-tops) ISSI (specialty memory), Omnivision (CMOS image sensors) and NXP's RF Power and Standard Products divisions (Nexperia). In the back-end, JCET acquired Stats Chippac to become the #3 global back-end supplier, Tianshui Huatian closed on Flip Chip International (flip chip bumping and wafer level packaging) and Tongfu Micro acquired two AMD advanced packaging facilities in China. In equipment, the E-Town Dragon Fund has also acquired Mattson, an equipment supplier of dry strip, etch, RTP and annealing used in front-end semiconductor manufacturing. ARM also spun off its China unit into a locally controlled company with better local access.

Figure 68: China supply chain has completed M&A to improve its technology position, although some larger deals cancelled Announced Value Announced Value Target Acquirer Deal status Target Acquirer Deal status date (US$mn) date (US$mn) 12-Apr-12 Leadcore, Uniscope, Yousi Datang $289 Completed 15-Apr-16 Tongwei Tongwei Solar $770 Completed 12-Jul-13 Spreadtrum Tsinghua Unigroup $1,780 Completed 14-May-16 NXP Standard Products Wise Road Capital $2,750 Completed Shandong Luyitong Intelligenet 11-Nov-13 RDA Tsinghua Unigroup $907 Completed 8-Jun-16 Hangzhou Jianan Yunchi $459 Completed Eletronics 1-Apr-14 NXP's auto IP Datang & NXP JV $60 Completed 26-Jan-17 YaGuang Technology Chengdu Yaguang Electronics $496 Completed 11-Jun-14 Montage Technology PDSTI $693 Completed 6-Jun-17 Jinglong Group JA Solar $1,408 Completed 14-Aug-14 OmniVision Will Semiconductor $1,900 Completed 30-Jan-18 GigaDevice Silead $424 Completed 3-Sep-14 Gerad Wafer Level CSP $53 Completed 5-Jun-18 Undisclosed ARM Technology (China) $775 Completed Asia Pacific Resources 25-Sep-14 Tsinghua Unigroup Intel $1,470 Completed 27-Sep-18 Shunfeng Photovoltaic $692 Proceeding Development Investment 6-Nov-14 STATS ChipPAC JCET, SMIC, IC Fund $780 Completed 13-Jul-15 Micron Technology Tsinghua Unigroup $23,000 Cancelled 16-Dec-14 FlipChip Int'l (FCI) Huatian $41 Completed 30-Sep-15 Western Digital Unisplendour (Tsinghua) $3,775 Cancelled 20-Jan-15 Sunplus' STB unit Availink Inc. $11 Completed 30-Sep-15 Pericom Montage Technology $430 Cancelled 20-Jan-15 Availink Inc. Sunplus' STB unit $10 Completed 30-Oct-15 Powertech Tsinghua Unigroup $600 Cancelled 12-Feb-15 SMIC National IC Fund $400 Completed 7-Dec-15 Mattson E-Town Capital $300 Cancelled 12-Mar-15 ISSI Uphill (SummitView) $764 Completed 9-Dec-15 Fairchild Hua Capital, CR Semi $2,460 Cancelled 28-Apr-15 Mic Led Light Huatian $8 Completed 11-Dec-15 ChipMOS Taiwan Tsinghua Unigroup $368 Cancelled 11-May-15 Static Control (SCC) Apex Microelectronics $63 Completed 11-Dec-15 Taiwan OSAT company Tsinghua Unigroup $1,700 Cancelled 19-May-15 HP's H3C HK Unisplendour $2,300 Completed 23-May-16 AIXTRON Grand Chip Investment $749 Cancelled 28-May-15 NXP's RF Power JAC Capital $1,800 Completed 26-Jul-16 Henan Yicheng New Energy Ldk Solar Hi-Tech $395 Cancelled 13-Jul-15 Silex Microsystems GAE N/A Completed 11-Jan-17 Qinchengda Investment Eging Photovoltaic $426 Cancelled 13-Aug-15 ISSI's Chingis MediaTek $27 Completed 22-Nov-17 Huaxin Investment Goodix $428 Cancelled 15-Oct-15 AMD's 2 ATMP facilities Tongfu Micro $371 Completed Proposed: $56,511 24-Nov-15 ZTE Microelectronics National IC Fund $380 Completed Completed: $21,189 Source: Company data, Credit Suisse estimates Companies currently view large-scale M&As as more difficult as either US may block the deal or China may ask for concessions. Companies are wary of larger deals currently in light of US CFIUS and Presidential actions to block Broadcom-QCOM and China’s Mofcom failure to approve Qualcomm-NXP. We expect the industry to still pursue acquisitions but political pressure increasingly will limit large scale global M&As or Chinese companies gaining IP quickly through acquisition, forcing it to maintain targets on smaller companies or acquiring IP or seeking lower profile JVs. Capital markets: new listings provide capital and recognition to China’s suppliers. We expect 2019-20 to see one of the largest waves of semiconductor IPOs in China in history, with strong support from the central/local governments and SOEs for the new Sci-Tech Innovation Board, a capital market in China for emerging companies to list. The board may allow much earlier-stage companies to list, including some small-scale, loss-making China semiconductor suppliers, later this year. As per our earlier analysis, A-share companies have traded at a 200%+ premium to the A- share market in the past five years so the high valuations could give a good source of funds for companies even at a relatively low profitability. There have been seven semiconductor companies’ IPOs in A-share between March and July 2019, including equipment (AMEC), materials (Anji Micro), IC design (Montage, Amlogic, Expressif, Beken, Maxscend). There will Asia Technology Strategy 53

4 November 2019 be more A-share semi IPOs in late 2019-20, including but not limited to IC design (UNISOC, Giantec, Fudan Micro), IDM (China Resources Microelectronics and Yantai Raytron), wafer (National Silicon Industry Group), IC design service (VeriSilicon) and equipment (Accotest, Kingsemi). BYD is also reported to have planned the spin-off of its semiconductor business (BYD Microelectronics) to list in A-share. Conclusion: China’s presence lagging in foundry, rising in OSAT, varying in IC design The Chinese semiconductor sector is having varying degrees of success establishing self- sufficiency in its domestic ecosystem due to high barriers on the advanced manufacturing and chip design. In foundry, market share has been stable around 9% in the past decade, despite National IC fund investments and heavy attention on this sector as the centre of China’s industrial development. TSMC’s wide lead on process technology and increasingly high-end advanced packaging allows it to address the high-end compute, mobile and networking demand. The Chinese foundries have kept up with market growth, largely by leveraging their more legacy 8” and mature 12” process nodes to address high volume consumer and wireless/wired connectivity, and specialty high volume applications (CMOS image sensors, power management, bank cards, specialty memory). In the back-end space, China has increased share following M&As of overseas companies and through a combination of aggressive pricing, targeting of SiP at high sales contribution but low margin, and addressing the growing base of local fabless and International companies with supply chain in China. The companies following acquisition can address the mainstream high volume packaging including flip chip and wafer level packaging though lag foundry leader TSMC integrating high-end applications on its silicon interposers and high density fan-out process. Market share has been rising and now approaching 20%, although profitability is still a concern with China players operating at low-mid single digit operating margins. In IC design, traction has been steady to lift share from 4% in 2010 to 16% in 2018, although has been concentrated in select areas of the semiconductor space. Market share is the highest in mobile areas (processor, modem, wireless power management, connectivity, CMOS sensors), wired connectivity (led by Huawei) and some new high volume applications seeing strong support in China (crypto-currency and AI). Share is still below 5% in areas representing about 70% of the industry, most notably processors, memory, GPU, FPGA, analog, MCU and RF. Figure 69: China semiconductor industry continues to outgrow global peers

Semiconductor sales (US$mn) China market share (%) 300,000 30%

250,000 25%

200,000 20%

150,000 15%

100,000 10%

50,000 5%

0 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 China foundry China back-end China fabless Global foundry Global back-end Global fabless China foundry share China back-end Share China fabless share

Source: Company data, Credit Suisse estimates What happens in the coming years? A few key trends we expect in the next 1-3 years: . China’s semiconductor expansion in its core wireless market. We expect China to have most of its growth from its existing strong hold in wireless, higher content in the mobile processor, connectivity, image sensor and gradual traction into RF. The move to 5G networks will allow higher content in mobile devices and stimulate more

Asia Technology Strategy 54

4 November 2019

spending in the infrastructure, with benefit for the local ecosystem from launching in the first wave vs lagging on 4G by a couple years. . AI emerges as a new high volume category for fabless. The China fabless were successfully dominating the crypto-currency ASIC market by leveraging advanced foundry capacity at TSMC and design service companies in Taiwan (GUC, ), and eventually Samsung and its design service ecosystem. We now see those companies targeting AI inference and some cloud acceleration applications, which could drive another revenue cycle, addressing the evolution towards smarter products capable of voice recognition and image/object detection. . Growth from a low-base in analog, RF, and MCUs. China has an opportunity to target more mature mainstream analog and discretes and MCUs, including at domestic companies looking for a local source alternative to the global IDM. The presence of local foundry capacity with Hua Hong’s 8” and new 12” mature fab and back-end capacity should give it a base of competitive manufacturing to gain traction and grow sales from a low base in these areas. The Asian RF foundries like Win Semi and AWSC along with Sanan’s development should also help the emerging RF suppliers in targeting some of the discrete RF components and mature 2G-4G cellular layers and Wifi RF. . Local foundries continue to grow on specialty applications, still limited advanced technology success. We expect China’s foundries to expand their position with the growing design companies on their mature lines (power management, flash memory, CMOS sensor, digital consumer). Hua Hong’s expansion of its Wuxi fab would double its revenue base in the next five years on still mature technology nodes, while SMIC should keep its mature lines relatively full with the growth of the domestic market. The advanced technology may take more time, as SMIC’s management admits that profitability on its upcoming 14nm FinFET ramp may still lag due to time gap from TSMC’s launch in 2015 (now nearly depreciated) and requiring development of n+1 and beyond to gain competitiveness and materially better pricing/profitability. . China back-end takes more share of local customers. China’s back-end sector is seeing increasing opportunities as it leverages the more advanced technology capabilities picked up from its acquisitions to target domestic companies adding local sources. Although, profitability may still lag a bit due to the on-going competition from ASE’s higher scale post-merger with SPIL and from TSMC continuing to take some of the highest value advanced applications with its wafer level system integration. We also highlight a few trends possible beyond three years for China: . A possible advanced technology expansion for SMIC. SMIC’s aggressive push on technology at the expense of near-term profitability, heavy government support, and drop-off of advanced technology competition from UMC and GF gives it a chance to emerge as a capacity source for more advanced applications. SMIC will need to endure high capex and difficult profitability but may emerge in 3-5 years as a more competitive foundry capable of supplying some of the more advanced mobile and networking applications. . Domestic data centre silicon. China’s design houses are working on alternative to Intel/AMD CPUs using ARM and RISC-V processors and NVIDIA accelerators with a host of AI inference and training chips. Within 3-5 years, the push for a local source for national security and economic reasons could allow local suppliers to gain some traction. . More penetration into broad markets particularly in China. We would view more build-up of local teams and domestic diversification to allow local suppliers to gain further market share into broad markets in MCUs and analog.

Asia Technology Strategy 55

4 November 2019

. Still limited success in some high barrier applications. We still view barriers on the platform in graphics, FPGAs with software and high-end integrated RF for 5G requiring a higher hurdle of engineering talent and incumbency in market position to gain strong traction. Opportunities and challenges We summarise the opportunities and challenges for China in achieving self-sufficiency. The key opportunities include: . Foundries grow scale on the mature nodes. Key opportunities for China foundries are to address and continue grow on the mature nodes requiring lower capital intensity and with large domestic base of customers, with Hua Hong/SMIC already having >50% of sales from local Chinese customers. . SMIC on the advanced nodes. SMIC has an opportunity if it can bear the high expense in the next few years to leap frog UMC and GlobalFoundries on technology and address more of the advanced mobile, networking and compute applications. . Back-end gaining from localisation. The Chinese back-end suppliers can continue to gain share with their local supply chain and growing domestic market leveraging the M&As they completed before the acquisition window narrowed with the recent political sensitivity. . IC design - leverage ARM and RISC-V for server, networking, mobile, connectivity. The IC design companies have lower entry barriers in some markets through the support of ARM and the growing RISC-V consortium, with more access to license the core IP for processing and connectivity applicable to data centre, networking, mobile and IoT markets. . Embrace advanced packaging, SiP and chiplet architectures. The foundry, EDA companies and back-end suppliers are providing more tools for integrating IP across smaller more manageable and easier to fabricate chiplets and joining in packaging, a possible opportunity to avoid the complexity in designing monolithic large system on chips. China may be able to target further opportunities if more systems take a system in package (SiP) or chiplet approach. While China can capitalise on some of these opportunities, it also faces some risks: . Wide gap to TSMC in capex and R&D scale. TSMC’s 5x spending advantage on capex and R&D and 10x advantage on revenue and even higher magnitude of free cash flow likely keeps the resource, technology and ecosystem gap wide. . Overseas fabless and IDMs also have a scale advantage. The Chinese will also face competition from established IDMs and fabless with embedded base of qualified design wins, global customers and also IP blocks that allow them to integrate some technology into the main chip – opposite trend of chiplets but still prevalent, for example, the mobile SoC companies now integrating the AI processing unit. . Global customer access may drop. The Chinese may gain share at local customers, but similar creation of two supply chains and overseas security concerns/US pressure may limit some of the market opportunities for Chinese suppliers into global customers. . Potential US restrictions on advanced technology. A risk for China would come from the widening of export restrictions of US technology or advanced manufacturing technology or equipment. The reliance on advanced tools for the fab or EDA tools would severely hamper the local supply chain.

Asia Technology Strategy 56

4 November 2019

Global alternatives to supply China if the domestic chain falls short We believe the attempt for China to localise its supply chain or limit supply from the US if the trade tensions continue to escalate will lead it to reach out to overseas non-US suppliers for its development. We highlight Asian suppliers potentially gaining share displacing Western companies: . Foundry: TSMC with only independent source of advanced manufacturing should remain key supplier to the Chinese design companies on silicon for advanced applications in mobile, networking, and high performance computing. . Back-end: We expect ASE, with 15% of sales in China and the largest industry scale at 30% market share in the back-end plus 77% owned USI Shanghai subsidiary, to also supply the local supply chain. King Yuan is also well placed for localisation with 15-20% of sales from Huawei’s chip division and 10-15% from Mediatek. . Mobile chipsets: Mediatek has a good opportunity, with Hi-Silicon’s mobile chipset captive to Huawei and UniSOC still lagging on mobile processors. Mediatek has a local advantage relative to Qualcomm if China continues to look for diversification from US supply sources. . Analog: We expect Silergy and Mediatek’s Richtek division to see opportunities to gain share, with Silergy competing against global analog companies and Mediatek having opportunity to bundle its own power management content on its platforms. . Probe cards: Chunghwa Precision competes with US suppliers such as R&D Altanova on probe card PCBs for advanced semiconductors, so it is already gaining a bit of share in Huawei following the customers’ US entity list restrictions. . RF: In RF, the Japanese suppliers Murata and Taiyo Yuden can gain content share from US suppliers, including Qorvo, Skyworks, Qualcomm and Broadcom. In Taiwan Win Semiconductor, VPEC and AWSC have higher market share, with local fabless versus the US IDMs. . Test equipment. In test equipment, Chroma and Anritsu have an opportunity to gain some share from US supplier Keysight from diversification. In semiconductor test, Advantest also can be an alternative Japanese test source to its rival Teradyne. . VCSEL. In VCSEL, Osram and Chinese based Vertilite can gain share from US suppliers Lumentum and Finisar in the local market.

Asia Technology Strategy 57

4 November 2019

Figure 70: China IC Design by application type

Mobile processor Ticker MCU Ticker Memory Ticker Huawei HiSilicon Private Gigadevice 603986 Gigadevice / ChangXin 603986 Unigroup Spreadtrum-RDA (Unisoc) Private, IPO Ingenic 300223 YMTC / XMC (Tsinghua Unigroup) Private Allwinner 300458 Unigroup Guoxin Microelectronics 002049 Fujian Jinhua Private Rockchip Private SinoWealth 300327 Reliance Memory (JV of Rambus and Gigadevice) Private Leadcore Private Silan 600460 Montage Technology 688008 ASR Micro Private Goodix 603160 Unigroup Guoxin Microelectronics 002049 Pinecone (Xiaomi) 1810 HK Datang Microelectronics Private ISSI / Ingenic 300223 Sanechips (ZTE Microelectronics) Private Huada Semiconductor Private Giantec Private, IPO Cloud, Server, NPU, AI Giantec Private, IPO SSD/eMMC controller Huawei HiSilicon Private Yi-xin Private Huawei HiSilicon Private Cambricon Private MindMotion Private Goke Microelectronics 300672 Montage Technology 688008 WinnerMicro Private Jiangsu Huacun Electronic Technology Private Alibaba BABA Fingerprint sensor & touch controller Tsinghua Dera Private Eeasy Tech Private Goodix 603160 SinoChip Semiconductors Private Huaxintong Private Silead / Gigadevice 603986 Yeestor Private Big Fish (Xiaomi) 1810 HK Fortsense Private StoreArt Private ThinkForce Private Betterlife 835288 Power, MOSFET, IGBT Iluvatar Private Chipone Private Huawei HiSilicon Private Video, audio, surveillance, image BYD Microelectronics (within BYD, to IPO) 1211 HK SG Micro 300661 Huawei HiSilicon Private CIS (CMOS image sensor) Silan 600460 Fullhan 300613 OmniVision / Will Semi 603501 Shanghai Belling 600171 Vimicro Private Galaxycore Private Nexperia / Wingtech 600745 Ingenic 300223 SuperPix / Will Semi 603501 Silergy 6415 TT Artosyn Private BYD Microelectronics (within BYD, to IPO) 1211 HK Will Semi 603501 Goke Microelectronics 300672 SmartSens Private Yangjie 300373 Eeasy Tech Private DDI (Display driver IC) BYD Microelectronics (within BYD, to IPO) 1211 HK Dahua 002236 SinoWealth 300327 Good-ARK 002079 YITU Private Solomon Systech 2878 HK Xiaocheng Tech 300139 Horizon Robotics Private Chipone Private Sun.King Power Elec 580 HK Sanechips (ZTE Microelectronics) Private RF, BLE, WiFi, connectivity Unigroup Spreadtrum-RDA (Unisoc) Private, IPO Bestechnic Private Huawei HiSilicon Private Huada Semiconductor Private FPGA Unigroup Spreadtrum-RDA (Unisoc) Private, IPO Tianyi Hexin Private Gowin Private Maxscend 300782 Beijing Zhisi Microelectronics Private Fudan Micro 1385 HK Goodix 603160 Sanechips (ZTE Microelectronics) Private Unigroup Guoxin Microelectronics 002049 Will Semi 603501 Chipone Private Huada Semiconductor Private Espressif Systems 688018 Global Semiconductor Limited Private Anlogic Private Siflower Private Jiangsu CAS-IGBT Technology Private Hercules Micro Private Etra Semi Private Secure IC for bank card, , security card DeePhi (Acquired by Xilinx) XLNX Tianyi Hexin Private Nationz Technologies 300077 CPU WinnerMicro Private Unigroup Guoxin Microelectronics 002049 Zhaoxin Private Xinyi Semi Private Shanghai Fudan 1385 HK Montage 688008 Power amplifier Huada Semiconductor Private Loongson Private Sanechips (ZTE Microelectronics) Private Giantec Private, IPO GPU Vanchip Private LED driver, MEMS sensor, discrete Jingjia Micro 300474 Xinyi Semi Private Silan 600460 Zhaoxin Private Kangxi Communication Private Shanghai Belling 600171 Cryptocurrency ASIC Smarter Micro Private Yangjie 300373 Bitmain Private Lansus Private Automotive Canaan Private UniChip (Will Semi's subsidiary) 603501 NavInfo / AutoChip 002405 Ebang Communication Private Aerospace, satellite IC BYD Microelectronics (within BYD) To IPO STB Orbita 300053 Horizon Robotics Private Goke Microelectronics 300672 UniStrong 002383 Huawei HiSilicon Private Amlogic 688099 Drone, robot SoC Sanechips (ZTE Microelectronics) Private Availink Private Artosyn Private Allystar Private Sanechips (ZTE Microelectronics) Private SiC (Silicon Carbide) Yi-xin Private Yangjie 300373

Source: Credit Suisse Research

Asia Technology Strategy 58

4 November 2019

Chaolien Tseng Semi equipment and wafers 852 2101 6795 [email protected] Current market structure—equipment Hideyuki Maekawa 81 3 4550 9723 The global semiconductor equipment market is dominated by a few major players, including [email protected] Yoshiyasu Takemura Applied Materials (the US), KLA (the US), LAM Research (the US), ASML (Europe) and Tokyo 81 3 4550 7358 Electron (Japan). There are a few suppliers in Korea, but those mainly serve Samsung and [email protected] Haas Liu Hynix. In China, China has only two established equipment companies, Naura and AMEC, 886 2 2715 6365 though both scales are very small compared with global semicap. [email protected]

Figure 71: China equipment suppliers still very small vs global players China Player Applied Lam Tokyo KLA Materials ASML Research Electron Tencor AMEC Naura Lithography 0.2% 83.1% - - - Photoresist Processing (Track) - - - 88.5% - Etch, Clean, and Planarization 19.0% - 34.7% 25.2% - ~1% 0~1% Deposition 37.6% - 21.0% 12.7% - ~40% MOCVD Ion Implanter (Doping Equipment) 68.2% - - - - Process Control 11.5% 5.9% - - 51.1% Manufacturing Automation and Control 4.0% - - - - Total Wafer Fab Equipment 18.5% 16.4% 15.1% 15.1% 5.5% ~1% ~1% Source: Company data, Gartner, Credit Suisse estimates Current Chinese players AMEC (Advanced Micro-Fabrication Equipment Inc. China): Founded in 2004 and headquartered in Shanghai, AMEC develops and manufactures dielectric and TSV etch tools for semiconductor manufacturers and MOCVD tools for LED makers. It has shipped etch tools to many semiconductor companies, including TSMC, SMIC, YMTC and Hynix. In December 2015, AMEC announced that its etch tool passed TSMC qualification for 5nm and the company expected a bigger opportunity at 5nm than at 7nm. In memory, AMEC indicated that etch tool can be used in 64L NAND production. For MOCVD, the company claims that it is one of the Top 2 suppliers, with the other one being Veeco. As of June 2019, AMEC had applied 1,280 patents, including 1,118 invention patents. It has been granted 961 patents, including 814 invention patents. Naura (Naura Technology Group Co., Ltd): Naura is formed through the consolidation and restructuring of Beijing Sevenstar Electronics and Beijing North Microelectronics in 2016 and the group was renamed Naura in 2017. Different from AMEC which was founded by a group of industry veterans, Naura and its previous entities assumes the nation’s mission to establish China’s capabilities in various equipment fields, from semiconductors, new energy resources, new materials and other fields. The more recent and important mission of Naura is to develop equipment for semiconductor manufacturing, as China has barely any domestic supplies. Naura develops a wide range of semiconductor equipment, including etch, PVD (physical vapor deposition), CVD (chemical vapor deposition), diffusion, cleaning tool, UV curve, indexer, gas measuring control, ALD (atomic layer deposition) and so on. That said, we note it has only 0-1% market share in these tools. Further, while AMEC’s etch has been qualified by TSMC’s most advanced node, Naura is still far behind and its equipment is more used in production of matured nodes. First, the company is too ambitious in development of a wide range of tools. Insufficient R&D talents issue is an industry-wide problem, with no exception to Nuara. Second, due to its history and management culture, Naura is more like an SOE (state-owned enterprise). Its SOE culture is tough for it to recruit and keep leading R&D researchers to develop advanced semi equipment.

Asia Technology Strategy 59

4 November 2019

Figure 72: AMEC’s key customers by product Equipment Customer type Major customers Etch IC manufacturers, OSAT TSMC, SMIC, UMC, Huali, Hynix, YMTC, Winbond, China WLCSP, GF, Bosch, STM MOCVD LED chip, power discrete Sanan, HC Semitek, Canyang Opto, Changelight manufacturers Source: Company data, Credit Suisse Research

Figure 73: AMEC revenue by etch and MOCVD Figure 74: AMEC shipments by etch and MOCVD Rmb mn Shipments 1,400 200

1,200 150 1,000 800 100 600 400 50 200 - 0 2016 2017 2018 2016 2017 2018 Etch MOCVD Etch MOCVD

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates Current market structure—silicon wafers The wafer market is dominated by the five majors, which collectively control over 90% of the market. Shin-Etsu Chemical and SUMCO are the two companies that have taken the lead in wafers for cutting-edge nodes. Both companies have about 30% share of this market. For Globalwafers, the company also gained share to 18% since the acquisitions of Topsil and SunEdison in 2016, which also allow them to have global production base and support leading edge foundries and memory makers. There have been some regional players establishing joint ventures in China. Ferrotec has established a 200 mm production base in Shanghai licensing technologies from GlobalWafers since 2017. They further set up another joint venture as a sales agent for the 200 mm production base, with 60% ownership by GlobalWafers and 40% by Ferrotec. Ferrotec also has plan to build a 12” raw wafer fab to support local semiconductor demand. Additionally, RS Technologies has invested in GriTek (GRINM Semiconductor Materials) in Beijing to make 200 mm prime silicon wafers. GriTek is 45% owned by RS Technologies, 49% by GRINM and 6% by Fujian Kuramoto.

Figure 75: China silicon wafer suppliers far behind global suppliers China players Market share Global players Market share Asian players to displace Western NSIG 0% for 12". 1-2% Shin-Etsu 30% ✔ for 8" Tianjin Zhonghuan <3% SUMCO 27% ✔ Semiconductor Regional players in Market share GlobalWafers 18% ✔ China Ferrotec – 5% for 8” Siltronic 15% GlobalWafers JV GRINM Semiconductor ~1% for <8”? SK Siltron 10% ✔ Materials Source: Siltronix, Credit Suisse estimates

Asia Technology Strategy 60

4 November 2019

For China domestic players, local wafer companies include NSIG and Tianjin Zhonghuan Semiconductor. They are working to boost their product competitiveness by recruiting engineers from overseas with detail below. NSIG (National Silicon Industry Group): NSIG is formed through the consolidation of three entities, including Zing Semiconductor, Shanghai Simgui and Okmetic. The company also tried to acquire Siltronic in 2016 for technology and capacity though didn’t come through due to concerns on IP leakage and the industry was at the inflection point for pricing upturn from more balanced supply/demand following years of loss. Currently, Zing Semi develops 12” silicon wafer. It was qualified to supply test wafers to some customers in 2017, but our channel checks suggested that it hasn’t yet begun real volume production. Its first phase of 100k 12-inch wafers/month capacity was built in 2018 and it has begun building the next phase of 200k capacity. Zing Semi plan to reach 1 mn capacity in next few years. It recorded Rmb25 mn/Rmb188 mn revenue in 2017 and 11M18. Shanghai Simgui produces <8” EPI and SOI wafers and it licenses SOI technology from SOITEC. Based in Finland, Okmetic is the seventh-largest silicon wafer supply for wafers used in manufacturing MEMS, sensor, discrete and analog circuits. Its revenue is over €85 mn. NSIG’s consolidated revenue was Rmb694 mn and Rmb1,010 mn in 2017 and 2018, mainly from the sales of 8” wafers. Tianjin Zhonghuan Semiconductor: Zhonghua is another China company which has indicated strong ambition in developing and producing similar wafers. The company has 300k WPM 8” raw wafer capacity in Tianjin in 2019 and targets to grow to 500k WPM in 2020. In addition to 8” raw wafer capacity, the company also targets to grow its 12” raw wafer capacity in its Yixin fab which is still under qualification by its customers. Although the company claims its raw wafer technology and production capability has been qualified by 60 8” customers and is under qualification by 20 12” raw wafer customers, we believe most of the raw wafers they produce is still only for testing purpose and its technology and quality are lag by its global peers by three to five years. China's game plan

China government, either central or provincial, does not have separate policies specifically supporting semi equipment or wafers. In general, for the semiconductor industry, the IC Development Guideline is the top-level policy document. That said, it is widely recognised that domestic customers (chip manufacturers) are encouraged to adopt equipment or wafers made by China suppliers. China companies across all semiconductor sub-sectors have made efforts to recruit international experienced talents to accelerate their R&D and expand their business. For raw wafers, China government appears to subsidise around 20% of the purchase price when China chipmakers purchase locally manufactured wafers. In addition, we believe local government would encourage foreign raw wafer suppliers to set up fabs locally to ensure sufficient supply amid macro uncertainty from the trade war. Conclusion: Lagging in wafers and far behind in equipment Equipment still lagging 5-10 years behind global peers For equipment, we estimate China players are at least five to ten years behind global suppliers, or even longer for areas where not yet any volume production by China companies. We expect the five-to-ten-year gap to continue in the foreseeable future, given the long development and complex qualification required with chipmakers. That said, given China’s strong momentum of domestic replacement, we expect some China players to shorten their gaps vs global players (i.e., AMEC), while some may require even longer time. Further, we remind that semiconductor makers are so keen on production yield and they have relied on current global equipment suppliers to improve yields. It has been quite difficult to change the suppliers unless China suppliers prove a huge technology advantage (less than 10% Asia Technology Strategy 61

4 November 2019 of performance gap may not be a trigger to change supplier). Price of equipment is not a critical issue to change suppliers in the equipment market as production yield is quite an important parameter for semiconductor makers. Opportunities and challenges for China equipment and wafer suppliers are similar to all the other semiconductor sub-sectors. Their great opportunities now are mainly driven by the strong momentum of import substitution and domestic chipmakers are willing to try domestic solutions. Key challenges are insufficient R&D talent in China and IP infringement risk. China wafer makers unlikely to close technology gap Although raw wafer is seen as a more commoditised product in the semiconductor supply chain, we believe local Chinese wafer makers are unlikely to be able to close the technology gap with the wafer majors and significantly boost their market share over the next one to three years. Given also the size of the market shares commanded by Japanese, European, Taiwanese and Korean makers, we do not envisage any noteworthy change in market share distribution. In addition, we believe the raw wafer customers wouldn’t risk to switch raw wafer suppliers only based on pricing as the raw wafers are essential and important for semiconductor manufacturing, while only representing a small portion of the manufacturing (10-15% for 8” foundries and mid-to-high-single-digits for 12” foundries). At this point, we have ascertained the following major developments, although in all three cases we understand there are unresolved technological issues. (1) Zingsemi is shipping 300 mm test wafers to customers, but these are not ready for use in commercial mass production. (2) Having hired in engineers from outside, RS Technologies has improved the technology of its 200 mm wafers, but not enough to catch up to the five market leaders. (3) Ferrotec has increased the competitiveness of its product by licensing in technology from GlobalWafers, but appears mainly to ship test and prime wafers and to a few China chipmakers that are not particularly fussy about product quality. As China chipmakers get up and running, growth in demand in the mature segment of the semiconductor market that does not require high-end wafers should provide business opportunities for China wafer makers.

Asia Technology Strategy 62

4 November 2019

Enterprise and servers

Jerry Su Current market structure 886 2 2715 6361 [email protected] US leadership in server declining Matthew Cabral 1 212 325 1754 The global x86 server market had been on a favourable uptick trend with total units and vendors’ [email protected] revenue growth of 6% and 14% CAGR from 2013-2018, respectively, driven primarily by the Sami Badri 1 212 538 1727 growing infrastructure build-out from increasing cloud service penetration. Coming off [email protected] significantly above-trend 15%/35% unit/revenue growth in 2018, we expect the x86 market to Kyna Wong pause in 2019 as buyers digest excess capacity. However, longer term, we continue to be 852 2101 6950 believers in the rising value of data structurally increasing compute/server demand. US brands, [email protected] including HP and Dell, continue to dominate the x86 market with 12.8% (or 15.7% including H3C) and 19.3% unit share, respectively. However, both of their market shares have been in decline (Dell by 3.9 pp and 12.9 pp by HP/H3C from 2013 to 2018), due to the rise of ODM US brands market share declining from the rise of ODM Direct and China competition Direct business models, particularly in the public cloud space, given concerns on total costs of ownership supported by cloud operators’ in-house design and software capabilities. During the same period, Chinese vendors’ market share has grown by 16.6 pp to reach 27.2% in 2018, driven by: (1) an increase in China’s domestic share of the overall server market (6.9 pp); (2) an increase in Chinese vendors’ share within the domestic market (4.4 pp); and (3) Chinese vendors’ growth outside China (5.3 pp), most of all of which is from Lenovo who purchased IBM’s x86 business in 2015. Supportive government policy has also been a key enabler of growth.

Figure 76: US vendors HP and Dell still leading the market Figure 77: …due to the rise of new ODM Direct business despite overall shares declining… model and China vendors expansion x86 server market share % by total shipments x86 server market share % by total vendor revenues

30% 30% 27.2% 28% 26.9% 26.0% 25% 26% 24.4% 24.2% 25.2% 25.5% 22.9% 24% 20% 19.7% 20.6% 18.1% 22% 21.6% 19.3% 15% 17.8% 15.5% 20% 19.8% 19.3% 14.8% 18% 17.1% 10% 16% 13.3% 8.2% 14% 5% 12.8% 12% 0% 10% 2014 2015 2016 2017 2018 2014 2015 2016 2017 2018 HP Dell ODM Direct China HP Dell ODM Direct China

Source: Company data, IDC, Credit Suisse research Source: Company data, IDC, Credit Suisse research For China vendors, we have included Great Wall, H3C, Hikvision, Huawei, Inspur, Lenovo, Powerleader, Tongfang, Sugon, and ZTE into our analysis. According to IDC, the total unit shipments from these China vendors grew by 19% CAGR from 2015 to 2018, representing closer to 30% of the global units vs 20% in 2015. However, most of their shipments remain in Asia-Pacific and more specifically China, in our view, with only three of the ten vendors, including Huawei, Inspur and Lenovo, shipping to overseas markets. In 2018, only 18% of total server shipments from China vendors were destined for markets outside of the APAC-region, but up from 6-7% range from 2009 to 2013.

Asia Technology Strategy 63

4 November 2019

Figure 78: Only 3 of the 13 China vendors in our analysis have shipments outside of APAC regions and shipments from most of which are still domestically focused % of shipments

100%

80%

60%

40%

20%

17% 21% 18% 16% 18% 6% 7% 6% 6% 0% 1% 5% 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Exposure in APAC Exposure in non-APAC

Source: Company data, Credit Suisse research Among all the shipments made by the China vendors, the majority of them are volume servers with an average selling price of below US$25,000, based on IDC definition. On the blended basis, China vendors have an average selling price at US$5,826, or nearly 20% discount of the global average of US$6,902. In the mid-range segment for pricing in between US$2,500 and US$250,000, the market continues to be led by US vendors, including HP (39%), Dell (15%) and Cisco (28%), and China vendors as a whole represent 11.5% of the total mid-range units. High-end x86 enterprise servers (i.e., ASP above US$250,000) account for ~1% of the overall market, with Huawei and Lenovo the only Chinese vendors competing. Oracle dominates the overall market, representing 72% of the total units, followed by HP/Super Micro for 11%/7% of the total units.

Figure 79: China vendors as a whole with majority of shipment still concentrate in the volume segment Total High-End Enterprise Midrange Enterprise Volume HP 12.8% 11.3% 38.8% 12.4% Dell 19.3% 0.1% 15.1% 19.4% ODM Direct 25.5% 0.0% 0.0% 25.8% China 27.2% 4.1% 11.5% 27.4% Inspur 7.7% 0.0% 2.1% 7.8% Lenovo 6.5% 0.0% 6.6% 6.5% Huawei 6.1% 3.5% 2.7% 6.1% H3C 2.9% 0.2% 0.0% 3.0% Sugon 2.6% 0.0% 0.5% 2.6%

Source: Company data, IDC, Credit Suisse research

Asia Technology Strategy 64

4 November 2019

Figure 80: Global x86 server still selling at a premium of nearly 20% vs China ASP in US$ (L-axis) and % of premium of global vs China

$8,000 160%

$7,000 140%

$6,000 120%

$5,000 100%

$4,000 80%

$3,000 60%

$2,000 40%

$1,000 20%

$0 0% 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19

China Global Global ASP premium vs China vendors

Source: Company data, IDC, Credit Suisse research China vendors also catching up in networking The Global Ethernet switch market has also been on an upward trajectory with total port units China Ethernet switch vendors also catching up with support of home market demand and and sales growing by 7.9% and 5.6% CAGR from 2013 to 2018, respectively, primarily driven more favourable government policy by builds-out of cloud infrastructure, proliferation of cloud services, and increasing wireless end- point devices (we cover this extensively in our networking sector primer: Cloud Network Fabrics Proliferate). In this market, US-based vendor Cisco is the single-largest player, constituting about half of the market total sales, despite only shipping 22-23% of the total ports. Cisco’s ability to have a large revenue share relatively to their port share is predominantly a function of their enterprise business made up of the Fortune 1,000 companies across developed markets. Additionally, Cisco’s dominant channel sales strategy with core value added resellers is a strategic advantage for the company, capable of boxing out competing vendors out of core enterprise deployments in most cases. Comparatively, China vendors as a whole including Huawei, H3C, ZTE, TP-Link and Alcatel-Lucent Enterprise are trailing behind Cisco by a wide margin. But the gap against Cisco has been narrowing from over 55 pp to ~32 pp in 2018/1H19, supported majorly by the home market demand growth along with a more favourable government policy.

Figure 81: Narrowing total sales gap between Cisco and China Figure 82: China total ports shipments already surpassing vendors Cisco in 2016 Market share as % of total sales Market share as % of total ports

80% 40% 70% 35% 60% 30%

50% 25%

40% 20%

30% 15%

20% 10%

10% 5%

0% 0% 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19

Cisco China Cisco China

Source: Company data, IHS, Credit Suisse research Source: Company data, IHS, Credit Suisse research

Asia Technology Strategy 65

4 November 2019

Figure 83: H3C, Huawei and TP-Link largest ethernet switch players in China % of total China sales

100%

80%

60%

40%

20%

0% 2013 2014 2015 2016 2017 2018 1H19

Huawei H3C ZTE TP-Link Alcatel-Lucent Enterprise

Source: Company data, IHS, Credit Suisse research According to IHS, China Ethernet switch vendors total sales increased by 37.8% CAGR for the past five years vs ports units growth of 19.5% CAGR for the same period. Nevertheless, as contrary to the server market, China vendors overall have been more reliant on the home market, with total sales from China growing from ~20% in 2010-11 to over 70% in 2018, driven by the rapid demand growth domestically, with China market sales tripling from 2013 to 2018 vs 30% growth for the same period for the global markets. Specifically, H3C is the only brand deriving all of its sales from China, while the overall reliance of other vendors, including Huawei, ZTE, TP-Link and Alcatel-Lucent Enterprise, on the home market all declined from the 2010 level.

Figure 84: Increasing domestic reliance from the China Figure 85: …however, H3C is only vendor with 100% of sales vendors driven by increasing China market size… from China; other vendors reliance from China declining % of total sales (L-axis) and % of China vs global market % of sales derived from China market

100% 16% 120% 15% 15% 14% 100% 100% 100% 100% 80% 12% 80% 60% 10% 69% 71% 8% 60% 58% 7% 40% 72% 6% 40% 42% 65% 69% 70% 37% 55% 53% 46% 4% 20% 37% 20% 2% 24% 18% 8% 0% 2% 0% 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19 Huawei H3C ZTE TP-Link Alcatel-Lucent Enterprise Exposure in China Exposure in non-China % of China sales vs global

Source: Company data, IHS, Credit Suisse research Source: Company data, IHS, Credit Suisse research Among all the port shipments made by the China vendors, the majority of them focus on the 50GbE and below port speed segment (i.e., 90% of total sales), despite that they are making good progress into the 100GbE port speed, representing over 10% of the total market share in 2018. Comparatively, Cisco continues to dominate across the board despite larger share erosion from the 100GbE port speed segment by the whitebox players and Arista Networks. However, Arista is solely focused on high-speed switching of 100GbE and above currently, with the majority of its share flowing in from major cloud providers.

Asia Technology Strategy 66

4 November 2019

Figure 86: Cisco dominating in ethernet switch; China players catching-up % of market share Total 100ME & 1/2.5GE 10/25/40/50GE 100GE Cisco 50.4% 56.9% 50.0% 31.5% Whitebox 2.8% 0.0% 1.7% 14.3% Juniper 3.4% 1.3% 5.5% 4.8% HPE (Aruba) 6.6% 9.0% 5.9% 0.7% China 15.9% 15.7% 17.9% 11.2% Huawei 9.1% 8.3% 10.5% 7.6% H3C 4.8% 4.8% 5.3% 3.4% ZTE 0.7% 0.2% 1.5% 0.2% TP-Link 0.8% 1.7% 0.0% 0.0% Alcatel-Lucent Enterprise 0.6% 0.8% 0.6% 0.1%

Source: Company data, IHS, Credit Suisse research

Figure 87: China vendors narrowing the pricing gap from global market ASP per port in US$ (L-axis) and pricing gap (R-axis)

$50 60x $45 50x $40 $35 40x $30 $25 30x $20 20x $15 $10 10x $5 $0 0x 2011 2012 2013 2014 2015 2016 2017 2018 1H19

China ASP per port Global ASP per port Global ASP premium vs China

Source: Company data, IHS, Credit Suisse research China's game plan China MIC 2025 laying out concrete targets China officially released the Made-in-China 2025 plan in 2015, aiming to support China China MIC 2025 set forth tangible market shares targets by 2020 and 2025 economic transformation, with key focus on promoting innovation, building internationally competent companies, localising production and data, etc., through the use of state resource and provision of financial policies, funds and subsidies support. Specifically, within the ten major focus industries, China disclosed plans to have global market share for domestically produced high-performance computers/servers to reach 30%, while the domestic market share to reach 60% by 2020, and the shares to reach 40% and 80% by 2025, respectively. It also aimed to China government supporting key domestic leaders to develop high tech projects with have domestically made high-end servers to represent over 50% of the domestic market; and subsidies servers with domestically produced CPU to reach above 30% of the domestic market.

Figure 88: Overview of MIC 2025 targets set forth by China government on high-performance computers and servers Timeline Overall market share targets Other industry specific targets 1) Achieving domestically produced servers for financial services and 1) International market share for domestically produced servers reach 30%; telecommunications industries to reach 75% market share; 2020 2) domestic market share reach 60% 2) Achieve domestically produced servers for financial services and telecommunication industries to reach 90% market share

1) International market share for domestically produced servers reach 40%; 2025 2) domestic market share reach 80%; n.a. 3) domestically made high-end servers share reach 50% in China

Source: U.S. Chamber of Commerce, Credit Suisse research

Asia Technology Strategy 67

4 November 2019

According to IDC, China server vendors total share internationally seems to be on a good track to achieve the targets set forth under MIC 2025 plan with their total shares reaching over 27.2% in 2018 and 26.6% in 1H19 vs less than 20% in the end of 2015, while revenues shares were lower given higher concentration from China vendors in the volume server segment. Comparatively, China server brands total shares picked up more materially by over 12 pp in the domestic market to over 62% in 2018 and 1H19 vs 50% in 2015 by unit shipments, or 17-19 pp to 62-64% in 2018/1H19 vs 45% in 2015 by sales.

Figure 89: China servers vendors global shares by unit Figure 90: China server vendors’ shares increase even more by shipments increase 6-7 pp from 2015 to ~27% now over 10 pp domestically to represent over 60% of total shares China vendors market share % China vendors market share %

30% 70% 27.2% 26.6% 62.5% 62.0% 62.8% 63.5%

25% 22.9% 23.0% 60% 50.3% 19.8% 50% 45.1% 20% 14.8% 40% 15% 30% 10% 20% 5% 10%

0% 0% 2015 2018 1H19 2015 2018 1H19

Units share Revenues share Units share Revenues share

Source: Company data, IDC, Credit Suisse research Source: Company data, IDC, Credit Suisse research China government subsidy supporting growth of domestic leaders In China, subsidy would typically be granted if companies develop high-tech projects that meet the key milestones under the government policy or requirement for government, while the China government tends to support key domestic leaders to achieve the targets. As server/switch are involved with concerns on national security, we see key players have certain government background from the management or major shareholders. For example, despite Huawei is a private company, it is the national champion in the ICT industry. Also, Lenovo is a but its ultimate shareholder, CAS Holdings, owns 29% of Legend which then owns 24% of Lenovo. Sugon’s major shareholder is a subsidiary of Institute of Computing Technology Chinese Academy of Sciences; Inspur is owned by Shandong government (39% stake); and H3C is under Unisplendour which is ultimately controlled by Tsinghua University. Given such strong relationship with government among Chinese server vendors, we could expect their milestone and mission in both the near-term and long-term timeframes are well set, while subsidy serves to be source of sustainable capital to support the business development.

Asia Technology Strategy 68

4 November 2019

Figure 91: Government subsidies increased from Rmb1.26 bn Figure 92: Subsidies received by Huawei exceeded combined in 2017 vs Rmb232 mn in 2011 subsidies for Inspur, Lenovo and Sugon Subsidies in Rmb mn (R-axis) and capex % of subsidies (L-axis) Subsidies income in Rmb mn

$1,400 30% $1,400

$1,200 25% $1,200

$1,000 $1,000 20%

$800 $800 15% $600 $600 10% $400 $400

5% $200 $200

$0 0% $0 2011 2012 2013 2014 2015 2016 2017 2018 2016 2017 2018

Government subsidies (Rmb mn) Government subsidies % of total capex Inspur/Sugon/Lenovo subsidies Huawei subsidies

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse research Note: Total government subsidies by Inspur, Sugon and Lenovo We compile total government subsidies received by the leading China server vendors including Inspur, Sugon and Lenovo, from 2011 to 2018, while we exclude H3C and Huawei from our time series comparison given the lack of historical data. Overall, total government subsidies from the three leading server suppliers saw a significant pick-up along the years to Rmb1.26 bn in 2017, from Rmb232 mn in 2011. However, the total subsidies came down in 2018 due to lower Lenovo contribution (i.e., down nearly 60% to 65% of total subsidies vs 87% in the prior year), despite flattish amounts received by Inspur (12% of total), while Sugon subsidies income increased by 1.1x (23% of total). Comparatively, Huawei’s subsidies income nearly doubled in 2018 from 2016 to Rmb3.0 bn, already 60% higher than the total amounts received by the other three leading China servers supplier. Conclusion Server: China vendors may see more obstacles expanding internationally China is already nearing self-sufficiency in servers from a hardware vendor perspective with Chinese vendors accounting for >80% of the domestic x86 market, per IDC. Given the government’s China First strategy and its intention to break from the US supply chain, we expect China server vendors to continue to expand their dominant position domestically. On the international front, Chinese vendors’ share remains relatively small at ~6%, 70% of which is We expect China vendors to gain more shares Lenovo who purchased IBM’s x86 business in 2015. Inspur has been making good progress in servers especially domestically with policy support; however, they may see more internationally, supported by its JDM business model and alignment to OCP standards. obstacles internationally from recent security However, we believe rising security concerns—especially after a Bloomberg report in mid-2018 and trade tension speculated on a hacking incident allegedly involving China-produced motherboards used in Supermicro’s servers—will slow the progress on Inspur’s expansion at US-based customers in both cloud/enterprise segments. By the same token, Huawei should also see more obstacles expanding internationally, especially for the roll-out of its 5G infrastructure, and we believe it could take longer for Lenovo’s DCG business to break-even. We also see limited room for other Asian vendors to meaningfully displace more traditional market leaders; Fujitsu, which is the largest, only has 2% global x86 revenue share. On the other hand, ODM Direct vendors supplying cloud operators, specifically Quanta and Wiwynn, will continue to take shares from traditional server brands, leveraging their enhanced customisation and hardware design capabilities, on top of lower total cost of ownership.

Asia Technology Strategy 69

4 November 2019

US server vendors adopting different strategies to secure leading shares Dell focusing on expanding leading shares; HPE focusing more on protecting profitability The two major US vendors, are likely to retain the largest market shares outside of China (ex- ODMs) for the foreseeable future, though we’d note each is taking a slightly different strategic approach to the market. Dell has been focused on expanding their leading market share with trailing 12-month server revenue share having risen for the past ten quarters from a trough in 2016, though more recently, the momentum stalled in 2Q19 due to share loss in China where they noted an increasingly competitive pricing environment. Going forward, we think there is scope for Dell to continue to gain share, leveraging the EMC storage footprint to expand deeper into enterprise accounts along with their “family” approach that better aligns innovation and roadmaps across “core” Dell, VMware, and Pivotal. HPE has been more focused on margins than share, working to shift their mix to higher ASP/higher-value areas of the market such as high-powered compute, software defined, composable, etc. Part of this strategy involved exiting the very low-margin Tier 1 (hyperscaler) market, which accounted for roughly half of their share loss over the past two to three years, and is now essentially complete. Within the remaining “Core” enterprise business we expect continued share loss ahead (CSe: ~1 pp annually through 2020) as they continue to focus on optimising mix and due to their outsized (~70%) exposure to Windows based systems that have been ceding share to Linux systems. Supply chain localisation a longer-term target For supply chain localisation, we expect Chinese server vendors will increase their sourcing from local components makers in the next few years; nevertheless, we expect majority of the BOM should still be dominated by the US suppliers. Specifically, we see 40-50% of the total server BOM associated with the chipset (Intel/AMD); 20-30% in memory (Micron, Samsung, SK Hynix, etc.); and the majority of the remaining from storage (WD and Seagate). While we China will take long time to achieve supply chain localization with key components believe China vendors are gradually catching up in their design capabilities and production supplies still dominated by US suppliers know-how as their learning curve scales along with international market share expansion, we do not expect China to have a complete back-up solution, especially on the chipset. However, we do expect sourcing of memory to shift to Japan/Korean players, as they move away from Micron. Domestic sourcing will be encouraged, if CXMT/YMTC are able to deliver a successful product, but our analysts believe that it is unlikely over the next five years (outside of some consumer-level parts). Beyond the CPU, the other major obstacle to true Chinese self- sufficiency in servers is operating systems where Windows and Linux continue to dominate.

Networking: Increasing competition pressure across all port speeds from China vendors internationally For the networking market, we expect to see Chinese vendors to take more share overtime and China vendors will gain more shares long term in Ethernet switch from the US vendors, with in the higher networking speeds of 100 GB+ from the low level today (10% of Chinese vendor also increasing pressure to the higher ports share in this speed) and expect to see more pressure on market share on US-based vendors as speed segment political tensions persist. Where we see additional opportunities for Chinese vendors are European developed markets where Germany recently called out that it will not ban Huawei from its 5G build-outs, giving Chinese vendors more global opportunities for the networking stack. For US overall switching, we expect to see Cisco, Juniper Networks, and HPE to continue dominating the US share, with limited opportunity in China given the political tensions. Finally, given the degree of R&D investments and supply chain ramp taking place in China, we expect Chinese vendors to continue to dominate China, driven by domestic government policy.

Asia Technology Strategy 70

4 November 2019

Jerry Su Display 886 2 2715 6361 [email protected] Current market structure Keon Han 82 2 3707 3740 China’s central government started pushing the construction of new panel fabs in 2009. Since [email protected] then, leading players for large-size display applications like BOE and CSOT have built new Kyna Wong facilities with the support of the central and local governments. Moreover, smaller players like 852 2101 6950 [email protected] CEC-Panda, HKC, and CHOT were also supported by different provincial governments and Sang Uk Kim constructed several new fabs in the past few years. As of 1H19, Chinese panel makers 82 2 3707 3795 accounted for 41% of the area shipment for LCD TV, 36% for NB/tablet, and 30% for monitor. [email protected] Although the industry has been in oversupply, Chinese panel makers are still building new large- Mika Nishimura 81 3 4550 7369 size panel fabs (mostly Gen 10.5) and will begin mass production in the next one to two years. [email protected]

Figure 93: Chinese panel makers grabbed 38% share for large-sized display in 1H19 Large-area LCD Large size area OLED TV China’s market share for LCD panels has LCD TV NB/Tablet Monitor Public Display share (1H19) increased significantly in the past decade Samsung a r a a 13.7% LGD a a a a a 21.1% AUO a a a a 11.7% Innolux a a a r 13.0% HannStar a r 0.1% Sharp a a r r 2.2% BOE a a a a r 18.8% ChinaStar a r r r 9.2% CEC Panda a a a 4.2% HKC a 2.3% CHOT a 3.0% Tianma a a 0.1% Infovision a 0.1% China's area share (1H19) 41% 36% 30% 8% 0% 38%

Source: Company data, IHS, Credit Suisse Adoption of flat-panel displays started from smaller-size handheld devices, such as PDA, computer and other consumer electronics. LCD panel replaced the CRT, becoming the mainstream for monitors in 2005-06, and proliferated into TV since 2008-09. In 2009, Chinese panel makers BOE and CSOT announced the making Gen 8.5 fabs, targeting TV applications. These two fabs entered mass production in 2011-12 and China’s area market share for TV panel increased from 1% in 2010 to 18% in 2015. During the same period, Korean panel makers built two Gen 8.5 fabs in China, and Taiwanese panel makers added some capacity to their existing lines, rather than constructing new fabs. In 2015-18, several new Chinese players started constructing new Gen 8.5/8.6 lines, while BOE/CSOT also announced to build Gen 10.5/11 fabs. For the next few years, we still expect Chinese panel makers to construct multiple new fabs for LCD TV and smartphone OLED panels, although over-capacity continues in the industry.

Asia Technology Strategy 71

4 November 2019

Figure 94: We see at least 10 fabs under construction/planning and will start production in next five years Beijing Ordos BOE: BOE: G8.5 G5.5 G5

Guan Visionox: G8.6 Zhengzhou Hon Hai: G6 Xianyang HKC: Kunshan Hefei IRICO: G11 AUO: G8.6 BOE: G6 Chengdu G10.5 Visionox/K&D: BOE: G8.5 G5.5 Shanghai G6 G6 IVO: Tianma/AVIC: G4.5 Wuhan Visionox: Chuzhou Mianyang G5 G5.5 Tianma: BOE: HKC: BOE: G6 G4.5 G4.5 G10.5 G8.6 G6 G5 CEC-Panda CSOT: HKC: EDO: G8.6 G6 G8.6 G6 G4.5 G6 Meishan Tianma: Nanjing G6 Truly: Chongqing Panda: Suzhou G4.5 G5 BOE: G8.5 Samsung G6 G8.5 Guangzhou G6 Display: G6 LG Display: G8.5 HKC: G8.5 G8.5 G8.5 Fuzhou BOE: Guizhou Foxconn/ Sharp: G8.5 HH/Century: G10.5 G6 G6 : Truly: Visionox: G4.5 G6 Xiamen Shenzhen Tianma: CSOT: G6 G8.5 G5.5 G8.5 Shanwei G6 G11 Truly: G11 G5 G2 Under production HH/Century: Under construction G5 Planning Royole: G6

Source: Company data, Credit Suisse estimates As a result of aggressive capacity builds, Chinese panel makers have dominated the commodity In 1H19, China accounts for 41% of LCD TV TV segment area share but 78% of unit shipments were for , such as 32”, 39”, 43” and 50”, as their capacity is more economic for producing 50” and below these sizes, as well as smaller sizes have lower technological entry barriers. As of 1H19, China accounted for 41% of the LCD TV area market share, but 78% of the unit shipment comes from 50” and below.

Asia Technology Strategy 72

4 November 2019

Figure 95: Chinese panel makers’ TV panel unit share Figure 96: …but its area share lagged behind Korean peers surpassed Korean peers in 2017 until the crossover in 1H19

TV panel unit shipment share TV panel area shipment share 100% 1% 2% 100% 1% 1% 11% 17% 18% 8% 12% 13% 24% 18% 24% 31% 36% 33% 32% 28% 34% 80% 39% 37% 40% 45% 80% 40% 34% 31% China 60% 45% 43% 48% 48% China 40% 60% 47% 38% Korea 44% 42% Korea 35% 39% 40% 32% 29% 35% 46% 49% Taiwan 40% 51% 54% Taiwan 48% 52% Japan 20% 30% 31% 29% 31% 31% Japan 26% 26% 20% 28% 26% 24% 24% 25% 24% 22% 14% 12% 8% 8% 7% 15% 13% 9% 0% 6% 5% 4% 3% 2% 0% 8% 8% 7% 5% 5% 4% 2% 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19

Source: his Source: his For the IT panel market (NB/tablet and monitor), Chinese panel makers’ share expansion has been slower than the TV in the recent years. We believe this is due to the fact that IT panels, especially tablet and NB, are smaller and requires a thinner design, which is more suitable for producing these panels in earlier-generation fabs (mostly Gen 5 and 6). Given China’s focus on capacity expansion was on higher-generation fabs (Gen 8 and above), Chinese panel makers have only seven Gen 5 and Gen 6 LCD fabs (~20% of global Gen 5/6 capacity). As a result, we think its market share for IT panels would remain stable in the near term. Figure 97: Chinese panel makers catching up on IT panel Figure 98: …but its area share still falls behind Taiwanese shipment with 35% unit share in 1H19… peers

IT panel unit shipment share IT panel area shipment share 100% 7% 8% 100% 11% 11% 15% 7% 8% 12% 11% 14% 17% 25% 16% 30% 33% 23% 29% 30% 80% 35% 80% 32% 41% 38% 35% 42% 40% 38% 50% China 47% China 60% 45% 44% 60% 45% 46% 35% 26% 24% 39% 23% Korea 32% 31% 30% Korea 40% Taiwan 40% Taiwan 53% 51% 54% 38% 38% 37% Japan 51% 52% 50% Japan 20% 38% 39% 38% 39% 20% 41% 40% 37% 37% 37% 37% 36%

0% 1% 0% 2% 1% 1% 2% 2% 5% 5% 5% 0% 0% 0% 1% 1% 1% 1% 1% 3% 2% 3% 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19 2010 2011 2012 2013 2014 2015 2016 2017 2018 1H19

Source: IHS Source: IHS For the small-/medium-size applications, Chinese panel makers have decent market shares in most of the consumer products that use LCD display, such as smartphone (59% area share in 1H19), tablet (53%), and smart watch (60%), especially for the mid- to low-end segments. This is because of the lower entry barriers, as these products can be produced using the legacy Gen 4, Gen 5, and Gen 6 fabs, and do not require the more advanced LTPS substrates. Figure 99: Chinese panel makers have higher share for commodity LCD smartphone/tablet/wearable applications

S/M-sized LCD S/M-sized OLED Smartphone Tablet Automotive Industrial Smart watch Smartphone AR/VR Smart watch BOE a a a a a a a a ChinaStar a r CEC Panda a Tianma a a a a a a r EverDisplay a a Visionox a a Infovision a a a Truly a a a r a Mantix a Royole a China area share (1H19) 59% 53% 25% 0% 60% 11% 0% 45% Source: Company data, IHS, Credit Suisse estimates

Asia Technology Strategy 73

4 November 2019

For the LCD displays used in high-end products, such as iPhone, iPad, automotive, and China’s market share for high-end panels remain low in 1H19 avionic, Chinese panel makers have relatively lower market shares. LCD iPhone displays are supplied by LGD, JDI, and Sharp, while LCD panels for iPad are supplied by LGD and Sharp. BOE entered the iPad supply chain in 2018, though limited to the lower-cost models. On the OLED side, Chinese panel makers’ market shares have been increasing, especially for smaller- size applications such as wearable watches, while OLED displays for smartphones is still dominated by Korean suppliers for both rigid and flexible OLEDs.

Figure 100: Chinese panel makers lead in smartphone Figure 101: …but Koreans still dominate OLED panels shipment…

Smartphone panel area shipment share OLED smartphone panel area shipment share 100% 100% 0% 1% 1% 2% 5% 11% 29% 29% 80% 37% 42% 80% 45% 52% China 60% China 27% 31% 60% Korea 100% 99% 98% 98% Korea 37% 95% 89% 40% 34% 33% Taiwan 40% Taiwan 19% 17% 32% 10% Japan Japan 20% 9% 10% 20% 25% 24% 7% 17% 15% 12% 9% 0% 0% 0% 0% 0% 0% 0% 0% 2014 2015 2016 2017 2018 1H19 2014 2015 2016 2017 2018 1H19

Source: IHS Source: IHS Although Chinese panel makers have established decent market positions in most of the end- Key component and equipment are still sourced from overseas suppliers applications in the past few years, they still heavily rely on overseas companies to supply processing equipment and key materials. For example, processing tools such as stepper for LCD panels are supplied by Nikon and Canon from Japan, while the deposition equipment for making OLED panels are sourced from Canon Tokki. For the key materials, Chinese panel makers still rely on Corning, Asahi, and NEG for glass substrates, although there are some domestic players such as Dongxu and Caihong that are able to supply glass substrates for smaller-generation fabs. Other key components such as polariser, driver IC, and optical films are also in a similar situation, wherein overseas suppliers remain in a dominating position given the higher technology barrier and greater scale. Using polariser as an example, Japanese supplier Nitto Denko mainly supplies polarising film and other high-functional films for high-end models in smartphone. The company currently supplies products to Chinese panel manufacturers for high-definition LCD and OLED smartphone screens, and the share gain by Chinese panel makers will be a positive for Nitto Denko. China game plan Focus shifting from LCD to OLED The decision of constructing new display panel fabs in China is largely related to government Government subsidies play an important role on new fab construction policy and incentives, as well as domestic market demand/self-sufficiency. China has been the largest country for manufacturing TV since the CRT-era and it has accounted for over half of global TV market in the last decade amid the LCD TV take-off. However, China lacked indigenous panel-production capacity during the replacement of CRT TV with LCD TV. As a result, the Chinese government announced its support of building high-generation panel fab in 2010 and implemented various measures to support the industry. BOE and CSOT announced building their Gen 8.5 lines in 2009, and construction started in 2010 and mass production in 2H11. The central and local governments offered initial funding, as well as subsidy, accounting for 30-35% of total investments. In 2010, China’s NDRC decided to issue two more licenses for high-generation fab construction, with Samsung Display (Gen 7.5, subsequently upgrading to Gen 8.5 in Suzhou) and LG Display (Gen 8.5 in Guangzhou) selected, while other companies (BOE, Sharp, Foxconn) applied but didn’t qualify.

Asia Technology Strategy 74

4 November 2019

Later in 2013, the approval power was released from the central government to the provincial Hefei government funded 45% of BOE’s Gen 10.5 fab governments, consequently boosting the take-off of new fab constructions as local governments offer various subsidy and incentives to attract investments for boosting local job employment and tax revenue. For example, BOE announced the construction of 90k substrates/month Gen 10.5 fab in Hefei in April, 2015, with a total investment of Rmb40 bn (US$6.45 bn), where 45% of the investment is funded by Hefei government, 45% from local banks, and only 10% from BOE. Figure 102: Structure for new fab construction

Source: Credit Suisse Research Another example is HKC, a TV set assembler with annual capacity of 22 mn LCD TVs, which Local government funded 50-70% of HKC’s new fabs in Chongqing and Chuzhou built a new Gen 8.6 TFT-LCD fab in Chongqing with capacity of 140k substrates/month and started mass production in 2017. The total investment of the Chongqing Gen 8.6 fab is around Rmb24 bn, where the local government subsidises half of the funding. HKC then announced its second Gen 8.6 line in Chuzhou in 3Q17, with capacity of 120k substrates/month for mass production in 2019. Total investment of the Chuzhou Gen 8.6 fab is around Rmb24 bn, with local government funding 70% of the investment. China government has continued to subsidise TFT-LCD fabs in recent years, but the amount has been declining and it has become more difficult to get funding for new fab constructions due to industry oversupply. Instead, the China government and the industry now aim to break the dominance of OLED panels by Korean panel makers, with the construction of multiple new lines, despite the poor production yield. Focus is now shifting to OLED China display industry is backed by China National High Technology Research and China is investing over Rmb400 bn in OLED panel capacity Development Program, and Made-in-China 2025 policy. China has set a target to produce 300ppi small-to-medium-sized AMOLED panels with curve dimension of <1cm by 2020 and foldable displays by 2025. Despite the Made-in-China 2025 turning low profile since the US- China trade dispute, the progress of moving to OLED and building domestic display ecosystem is still on track. Chinese panel makers started to spend significant capex from 2017 to expand OLED capacities, triggered by Apple's adoption of OLED panel along with a massive subsidy from the government. Several panel makers have announced to accelerate their OLED capacity build-up plans including BOE, Tianma, EverDisplay (privately held), Truly, Visionox, Royole (privately held) and CSOT. We estimate over Rmb400 bn of investment in OLED panel manufacturing.

Asia Technology Strategy 75

4 November 2019

Figure 103: Over Rmb500 bn investment from China panel makers Company Fab Gen Type Production schedule Designed capacity Total investment (K/M) (Rmb bn) BOE Ordos 5.5 LTPS/AMOLED 4Q13 4.00 22.00 Chengdu 6 Flexible AMOLED 3Q17 48.00 46.50 Mianyang 6 Flexible AMOLED 3Q19 48.00 46.50 Chongqing 6 Flexible AMOLED 1H21 48.00 46.50 Fuzhou 6 Flexible AMOLED TBC 48.00 46.50 Visionox Kunshan 5.5 Rigid/Felxible AMOLED 2017 15.00 15.00 Gu'an 6 Rigid/flexible AMOLED 4Q18 30.00 30.00 Hefei 6 Flexible AMOLED 2021 30.00 44.00 Guangzhou 6 Flexible AMOLED TBC 10.00 11.20 EverDisplay Shanghai Quarter 4.5 Rigid AMOLED 4Q14 21.00 7.50 Shanghai 6 Flexible AMOLED 2019 30.00 27.28 Tianma Shanghai 5.5 Rigid AMOLED 4Q15 4.00 1.00 Wuhan 6 Flexible AMOLED 2H21 37.50 26.50 Xiamen 6 Flexible AMOLED 2022 48.00 48.00 Truly Huizhou 4.5 Rigid AMOLED 4Q16 15.00 6.30 Meishan 6 Flexible AMOLED TBC 30.00 27.90 Royole Shenzhen 6 Rigid/flexible AMOLED 2H18 50mn module 11.00 CSOT Wuhan 6 Flexible AMOLED 2020 45.00 35.00 Shenzhen 11 LCD/AMOLED 2021 90.00 42.68 Source: Company data, Credit Suisse estimates China is aggressively moving to OLED, as (1) Chinese panel makers are able to obtain capital resources from the government and the stock market, (2) OLED development is well supported by National high-tech policies, and (3) Chinese players aim to surpass global peers by cutting corners. We see a massive capex roll-out since 2017, mainly driven by BOE, while other OLED makers also ramped up their production. Shanghai’s EverDisplay started mass production of OLED panel for mobile applications in end 2014. Kunshan Visionox started production in end-2015. Tianma also started shipping OLED panels for Chinese smartphone makers in 2016 from its Shanghai OLED fab, and mass production of its Wuhan OLED fab in 2018. Truly announced its OLED production line to start production in 4Q16. BOE started mass production of Chengdu flexible AMOLED panels in 4Q17 which was the first Chinese company to commercialise flexible OLED. We note that the majority of the projects have been set up with government parties and listed companies mainly hold a tens percentage of stake, which could help lessen the financial burden of listed companies. We estimate Rmb25 bn/56 bn/53 bn of capex for OLED fabs during 2017-19 in China and expect Rmb74 bn/93 bn/65 bn in 2020/2021/2022.

Figure 104: Estimated OLED capex in China

100 (Rmb bn)

80

60

40

20

0 2017 2018 2019 2020 2021 2022

Source: Company data, Credit Suisse estimates

Asia Technology Strategy 76

4 November 2019

Among Chinese OLED panel makers, Everdisplay and Tianma were successful in delivering rigid Chinese OLED panel makers rely on overseas supplier for raw material and equipment smartphone OLED panels in 2016. Everdisplay started with non-smartphone applications such as smart watches, industrial and other handheld devices and has taken two years to ramp up. Tianma shipped only a small volume in 720p resolution at the beginning, but now its Wuhan OLED fab can deliver rigid smartphone OLED panels at a reasonable yield rate. BOE started mass production of flexible OLED smartphone panels from 4Q17 but ramping up took a long time without a flagship model. was the first flagship model it successfully shipped in 2H18, and now we see BOE is facing another technical issue on Mate 30 display requirement. We see Chinese OLED panel makers are struggling to improve yield and reduce cost. Longer term, we expect more Chinese OLED panel makers could deliver rigid OLED panels, while BOE remains the leader in flexible OLED panels in the China market.

Figure 105: Chinese panel makers now account for 12% of Figure 106: …with faster share gain for flexible OLED as BOE smartphone OLED panel shipments… ramps up its capacity

OLED smartphone panel unit shipment share Flexible OLED smartphone panel unit shipment share 100% 0% 1% 1% 2% 5% 100% 0% 2% 12% 11% 80% 80% China 60% 60% Korea 100% 99% 99% 98% 94% 100% 98% China 88% 89% 40% Taiwan 40% Korea Japan 20% 20%

0% 0% 0% 0% 0% 0% 0% 0% 2014 2015 2016 2017 2018 1H19 2017 2018 2019E

Source: IHS Source: IHS On top of financial support from the government, the Chinese panel makers also hired R&D and process control talents from Taiwan (Chinese Taipei), Japan, and Korea in the past decades to shorten the learning curve and catch up on premium products. As shown in figure 102, China’s TV panel unit and area shipments have surpassed Taiwan in 2017, but it still lags behind in 4K TV panels. Nevertheless, with the continuous R&D effort, it has now caught up with the 4K TV panel shipment share, having surpassed Taiwan and on a par with Korean peers.

Figure 107: Chinese panel makers also catching up on 4K TV Figure 108: BOE’s R&D expense is now on par vs AUO+INX

US$ mn 4K TV panel area shipment share AUO+INX's R&D BOE's R&D 800 100% 11% 11% 16% 17% 27% 80% 37% 600 48% China 60% 59% 54% 51% 46% Korea 400 39% 40% Taiwan 200 38% Japan 20% 26% 25% 25% 24% 22% 0% 3% 5% 5% 6% 3% 2% 0 2014 2015 2016 2017 2018 1H19 2011 2012 2013 2014 2015 2016 2017 2018 1H19

Source: IHS Source: Company data Government pushing the localisation of key component and raw material Despite the success in taking over leadership on TFT-LCD panels over Taiwanese and Korean Government encouraging the adoption of localised material peers, the majority of the key components and processing material for Chinese panel makers to produce TFT-LCD still relies on foreign suppliers. However, the government has launched subsidies and incentives since 2017 to boost the adoption of key materials from local vendors. Global polariser makers like LG Chem and Samsung have set up production sites in China, while Japanese Nitto Denko SDI is working with several local players by licensing its production Asia Technology Strategy 77

4 November 2019 technology. Based on IHS’ estimate, China’s polariser capacity would reach 40%+ of global capacity by 2021, up from 17% in 2017, although this is likely to lead to a severe oversupply from 2021 onward. For Nitto Denko, Korean and Taiwanese TV manufacturers account for a high share of its business, and Chinese makers currently account for a low share. However, with Chinese panel makers gaining ground, Nitto Denko opened a new plant in Shenzhen, China to manufacture ultra-thin polarizing plate (volume production started 2018). The company is also reinforcing its royalty business, including licensing out roll-to-panel technology to Chinese polarizing plate makers and forming business partnerships in the industry. In 2017, Nitto Denko formed a technology alliance with China’s Jinjang group, which will see Nitto Denko receive up to ¥15 bn in royalties during the five-year agreement period. In terms of the display driver IC, A-share-listed Sino Wealth and H-share-listed Solomon Systech are two IC design houses shipping LCD and OLED driver ICs to domestic panel makers, while BOE has set up a design house ESWIN. Nevertheless, their market shares remain limited and Chinese panel makers still source driver ICs mainly from Taiwanese players such as Novatek, Himax, Raydium, etc. Driver IC fabless design houses are also encouraged to use local foundries for wafer processing and panel makers could also offer higher allocation if the production of the driver ICs are done in China. Novatek has taken advantage of its strong engineering capability and completed the qualification ahead of its peers at Hefei Nexchip’s 12” fab for large-size DDI, which helped it to gain share over its peers at BOE in 2H18-1H19. BOE’s affiliate ESWIN originally focused only on display driver IC design, but it has announced to form a strategic partnership with Chipbond in Dec-2017 and has expanded into driver IC backend packaging by acquiring a majority stake in Chipbond’s China subsidiary Chipmore. BOE and Chipbond have also set up a new JV named ESWIN material, focusing on COF tape manufacturing. Post the transactions, ESWIN now has a COF tape capacity of 50 mn units/months, 8” bumping capacity of 40k wafers/month, and TCP/COF backend capacity of 40 mn/month.

Figure 109: Driver IC maker’s share in 1H19 (LCD+OLED) Figure 110: AMOLED driver IC market share in 2Q19

Fitipower ROHM Solomon Others 1.3% Systech Ilitek 2.6% 3.7% 1.2% 2.6% Synaptics Others Samsung Novatek FocalTech 3.0% 1.0% 2.9% 28.6% 5.5% MagnaChip Raydium 4.0% 6.0% Raydium 5.6% Synaptics 6.0% MagnaChip 24.5% Samsung LSI Sitronix Novatek 60.0% 5.3% Himax 20.4% SiliconWorks 8.0% 7.7%

Source: IHS, Company data Source: IHS, Company data In 2017-18, the China’s MIIT also encouraged display makers to use local processing materials (chemicals, photoresist, gas, substrates, OLED materials, etc.) by offering cash incentives to panel makers if they adopt a certain amount of local materials. The policy was changed in late 2018 to compensating panel makers if the local material/components failed and thereby caused yield-loss during mass production. The new policy could potentially increase the adoption of local materials, as previously, some panel makers would just buy some local materials and store those in the warehouse without using those for production to apply for the subsidy.

Asia Technology Strategy 78

4 November 2019

For OLED, Chinese panel makers still see limited supply from domestic vendors and purchase equipment and raw materials from overseas vendors. We believe the lack of local supply chain will likely prolong the cost reduction progress for panel makers. The quality of equipment and materials also affects the yield of production. We expect Chinese panel makers to continue to rely on overseas suppliers in consideration of the risks of production in terms of mature materials, proven equipment and IP concerns. But in the long run, China aims to build up its ecosystem through cooperation with overseas suppliers, M&A and government subsidy.

Figure 111: China OLED panel supply chain Equipment Materials Driver IC AMOLED Panel 3D glass casing Handset NAURA (002371.SZ) Jilin Optical (private) Sinowealth (300327.SZ) Tianma (000050.SZ) Lens Tech (300433.SZ) OPPO (private) Han's Laser (002008.SZ) RuiYuan (private) Solomon Systech (2878.HK) BOE (000725.SZ) Biel Crystal (private) vivo (private) Dalian Zhiyun (300097.SZ) Puyang Huicheng (300481.SZ) ESWIN (private) Truly (0732.HK) Tongda (0698.HK) Huawei (private) Liande (300545.SZ) Valiant (002643.SZ) Visionox (002387.SZ) AAC (2018.HK) Xiaomi (1810.HK) Selen (002343.SZ) EDO (private) (private) Eternal Material (private) Olightek (private) ZTE (000063.SZ) Kangdexin (002450.SZ) TCL (000100.SZ) Lenovo (0992.HK) Tecno (688036.SS) Source: Company data, Credit Suisse estimates Conclusion: Chinese display makers becoming more meaningful but still lag key tools and raw material

Chinese display panel makers obviously have caught up on producing TFT-LCD panels, especially for the commodity and lower entry barrier products. We expect their market share to further increase as they ramp up new fabs, especially amid Korean panel makers’ strategy change, although there is a concern that some smaller TFT players might not survive without government subsidy. Chinese LCD makers’ share could further Korean LCD makers to exit TFT-LCD manufacture increase as Korean players exit the market We fully expect Korean LCD makers to exit the TFT-LCD panel manufacturing over time. Most of the LCD products are no longer profitable, and Korea's strategy is to move up the technology curve to pure OLED production base while phasing out the existing LCD facilities. Both Chinese and Taiwanese TV LCD panels have attained sufficient quality, where major TV brands such as Samsung and LG Electronics (LGE) have been expanding panel procurement for TVs at lower price points. High-end TV panels will move toward OLED TVs for both brands, with LGE enjoying a head start. For example, While LGE still largely sources TV panels from its subsidiary LG Display (LGD), the proportion has been declining, from 70% in 2015 to about 60% in 2018 and heading toward 50%, with Chinese suppliers gaining the bulk of the increase. This is to improve the cost structure of TV sets. Generally, premium TV LCD panels will still be sourced by LGD, while 100% of WOLED panels will be sourced from LGD as it moves to improve the mix of OLED TVs to improve the TV profits.

Asia Technology Strategy 79

4 November 2019

Figure 112: LGE's TV panel technology hierarchy

LG Signature OLED TV LG OLED TV

LG Super Ultra HD TV Ultra HD LCD TV (Nano Cell) LG Ultra HD TV

LCD TV LG LED TV

Source: Company data, Credit Suisse Research

Figure 113: LGE's TV panel procurement

2015 2018

Other Panel Makers; 30% BOE, Innolux and 40% etc.

LG Display 70% 60%

Source: Company data, Credit Suisse estimates

Figure 114: Samsung's LCD revenue continues to slide

(W bn) 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 1Q08 4Q08 3Q09 2Q10 1Q11 4Q11 3Q12 2Q13 1Q14 4Q14 3Q15 2Q16 1Q17 4Q17 3Q18 2Q19 1Q20E 4Q20E Large Panel TFT-LCD Revenue

Source: Company data, Credit Suisse estimates LTPS LCD production nearly phased out For small TFT-LCD and LTPS LCD display, Samsung began phasing out production in 2014. The company shut down the L4 (2014) and L5 lines (2015) in Cheonan mainly producing displays for low-end smartphones. In November 2016, Samsung proceeded to shutdown L7-1 fab mainly producing LCD panels for laptop PCs and PC monitors, including oxide substrate

Asia Technology Strategy 80

4 November 2019 based LCD. The equipment was sold to Truly of China. Additionally, in 2017, Samsung proceeded to convert its L7-1 LCD line in Tangjung into an OLED facility. Similarly, LG Display has closed its P2, P3 and P4 lines in Paju and Kumi, with most of the AP2 lines used for making LTPS-LCD now being converted for the usage of OLED production. While Samsung had sufficient internal smartphones to sell its LCD products, the strategic shift to move toward OLED for the majority of its smartphones has prompted the move. Today, Samsung produces no smartphone LCD displays internally. Instead, its early shift toward the future of OLED displays enabled the company to enjoy near-monopoly status for more than a decade. The OLED space is marked by fewer competitors and more advanced technology. The near-monopoly status also enables the company to keep a tight control over its equipment and material suppliers. OLED has grown into a K25 bn/year-revenue segment, with a much more stable 10-12% sustainable OPM for the company.

Figure 115: Samsung: OLED has displaced LCDs and display revenue driver

(W bn) 10,000 30% 25% 8,000 20% 6,000 15% 10% 4,000 5% 0% 2,000 -5% 0 -10% 1Q09 4Q09 3Q10 2Q11 1Q12 4Q12 3Q13 2Q14 1Q15 4Q15 3Q16 2Q17 1Q18 4Q18 3Q19E 2Q20E OLED Revenue OLED OPM (RHS)

Source: Company data, Credit Suisse estimates Similarly, LG Display was initially the top LTPS LCD display supplier for Apple's . It also supported its parent, LGE's smartphone initiative as well as supplying to the Chinese brands. While the company still runs a profitable business in the oxide-based LCDs for AAPL's PC monitors and laptops, the mobile LTPS display has also been phasing out as Apple began to adopt more OLED displays in its iPhone models and the smartphone volume of LGE has shrunk considerably. As an LG Group strategy, developing OLED TVs was emphasised first, and then flexible mobile OLED transition following Apple's iPhone shift. Currently, LG Display is the sole supplier of WOLED panel maker enabling many flagship models for global TV brands such as LG, Sony, Panasonic, B&O, etc. Much of the initial WOLED capacity was built by converting its 8G LCD capacity into OLED production facility, and the brand-new fab is just starting production in Guangzhou, China.

Figure 116: LGD's TV OLED capacity build-up

200K

150K P10?

100K

50K

0K 2014 2015 2016 2019E 2020-2021E LGD - OLED TV capacity (K/mon, 8G)

Source: Company data, Credit Suisse estimates

Asia Technology Strategy 81

4 November 2019

Conversion from LCD fabs to OLED capacity Starting with the L7-1 conversion to flexible OLED line, Samsung has begun shutting down More capacity to be shut down by Korean panel makers more Gen 7 and Gen 8 LCD capacities, starting with L8-1. Potential planed capacity elimination could run as high has over 200k 8G substrates, which could represent ~30% of the Samsung's Korea LCD capacity. Much of the capacity loss is expected to be retrofitted to production of QD-OLED, Samsung's version of TV OLED technology. The company has officially unveiled a W13 tn capex plan for the next seven years, with about W10 tn expected to be spent mostly in the OLED area and W3 tn in R&D. While TV OLED technology is largely still in R&D phase, about 20k to 30k in 8G capacity will be built from the converted LCD fab for the initial trial run. Given Samsung's vast experience in more difficulty mobile OLED R&D and production, the company will likely enter successful mass production toward the end of 2021. Large LCD panel business is currently loss-making and no is longer attractive for Samsung given the continual supply ramp in China pressuring global LCD prices. LG Display is in a more dire position as the majority of its revenue is still being generated from large LCD panel sales. Given the current market price of various LCD panels, virtually all of its products are loss-making, except for its oxide-based LTPS LCDs for special laptop and desktop monitor use. While LGE has transitioned to TV OLED early, it has just now reached the B/E point in the Korea operations, while the new China facility is expected to post losses in the near- term due to high ramp-up costs and low initial yields. Additionally, in order to regain the loss of LTPS-LCD sales to iPhones, the company has built three new 6G flexible OLED fabs (two for AAPL) to address the growing OLED usage in the smartphone market and rising demand from the auto industry. However, until the new OLED production sites are on a sound footing, restructuring that will include executive layoffs, voluntary retirement of LCD manufacturing staff and job reassignment are likely to continue. YTD, the company has accepted over 300 voluntary retirement requests from it fab personnel.

Figure 117: Both Samsung and LGD loss-making in LCDs 30%

20%

10%

0%

-10%

-20%

-30%

-40% 1Q08 4Q08 3Q09 2Q10 1Q11 4Q11 3Q12 2Q13 1Q14 4Q14 3Q15 2Q16 1Q17 4Q17 3Q18 2Q19 1Q20E 4Q20E Samsung LGD

Source: Company data, Credit Suisse estimates China to rely on Japan, Korea, and the US for key raw material, especially OLED For the higher-end LCD panels (such as 4K narrow border TV panels, in-cell touch display, large-size TV panels, etc.), we believe Chinese panel makers could become the main suppliers in the next two to three years as they overcome the production yield issues. For the more advanced TFT-LCD panels, such as smartphone panel with full-screen fingerprint sensing, 8K TV panels, miniLED BLM, high-refresh-rate displays, etc., we believe it will take more than three years for Chinese panel makers to catch up with its Taiwanese and Korean peers. In near-to-mid-term (one to three years), China display industry will likely focus on (1) building its capability and scale in OLED panel manufacturing and high-end TV panels (55” or above, 4K and 8K TVs) and (2) incubating domestic display supply chain. BOE will remain the national champion in TFT LCD and flexible OLED, reach international footprint, and gain global recognition. Entering iPhone OLED panel supply chain is a key milestone.

Asia Technology Strategy 82

4 November 2019

We believe Chinese panel makers might use more local material and components for manufacturing in the next few years, although they will still rely on overseas vendors for certain high-end components that requires more fundamental research on material science. We believe material/components such as ultrathin cover glass (Corning’s Gorilla glass or Asahi’s dragontrail), large-size glass substrate (Gen 10.5), OLED polariser (film-type circular or liquid crystal pattern), high-end optical film (quantum dot film), OLED materials, etc., will still be dominated by overseas suppliers, given their know-how of raw material, formulation, and end- customer’s decision. Despite US suppliers like Corning leads in glass substrate/cover glass and leads in optical films, we believe Chinese panel makers could also source these materials from Japan to reduce dependence on US technology. Form a long-term perspective, China would grow its domestic supply chain to get a meaningful contribution from local panel makers and reduce reliance on overseas suppliers, especially US vendors. For OLED materials/components, we believe it will take longer for Chinese panel makers to Samsung’s blackbox strategy for OLED will slowdown China to build its own raw material establish their own supply chain, primarily due to Samsung’s black-box strategy to protect its supply chain core technology. OLED is generally a much more difficult technology to master, as it involves much trial and error due to its analogue manufacturing processes. Furthermore, the pioneer and the leader, Samsung Display, has wanted to strictly control the supply chain, including building up its IP portfolio on production methods, equipment advancement, and upgrading organic materials science. Samsung has even acquired some key OLED materials/equipment supply chain (i.e., Vitex Systems or Novaled) and invested in some critical equipment suppliers with ~5-10% ownership and disallowed them to sell the most advanced OLED materials/equipment to potential Chinese competitors. That said, even if a leading Chinese OLED panel maker such as BOE keeps trying to mass produce flexible OLED panels, it will find it difficult to stabilise yield rates or it will lag behind the technology curve as it will be tough to procure the latest/advanced technology equipment/materials due to patent issues. Particularly for OLED materials, some key suppliers such as Universal Display, Dow Chemical, Idemitsu Kosan, Duksan Neolux, etc., are already the dominant players in its sub-segments where Samsung Display is unlikely to approve of them providing advanced materials supply to Chinese panel makers.

Figure 118: Advanced OLED panel components/materials supply chain (i.e., Galaxy Fold)

Source: Company data, Credit Suisse estimates

Asia Technology Strategy 83

4 November 2019

Kyna Wong Components 852 2101 6950 [email protected] We see China has dominated smartphone manufacturing, making half of the world’s Pauline Chen smartphones. China is also building its ecosystem and strengthening its global presence and 886 2 2715 6323 leadership in component makers. The Chinese smartphone market started with whitebox [email protected] smartphone manufacturing back in 2011-2012 and has passed through several generations Sang Uk Kim 82 2 3707 3795 and brand reshuffles. At that time, the market was driven by operators with central procurement [email protected] on Rmb1,000 phones and the Top 4 domestic smartphone brands were ZTE, Huawei, Coolpad Mika Nishimura and Lenovo. In 2014, operators cut handset subsidy which created open-channel opportunities 81 3 4550 7369 and intensified smartphone competition. Xiaomi successfully created its online business model; [email protected] while Oppo and Vivo are strong in the open channel, down to lower-tier cities in China. Huawei Akinori Kanemoto 81 3 4550 7363 was relatively sustainable in past generations and remains a top-tier smartphone brand. The Top [email protected] 4 Chinese brands (Huawei, Oppo, Vivo, and Xiaomi) accounted for 39%/45% of global Jerry Su smartphone shipments in 2018/1H19. Huawei, Oppo and Xiaomi are also ranked among the 886 2 2715 6361 global Top 5 smartphone makers. [email protected]

Figure 119: Increasing brand concentration (Top 5: 67% in Figure 120: Rising penetration of Chinese brands smartphone 2018) (mn) 1,600 1,600 (mn) 60%

1,400 1,400 50% 1,200 1,200 40% 1,000 1,000 800 800 30% 600 600 20% 400 400 10% 200 200 0 0 0% 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E

Apple Samsung Huawei Xiaomi OPPO vivo Others Chinese brands International brands % of global smartphones

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates China grows its smartphone supply chain by entering Apple supply chain and leveraging the learning experience from making feature phones for and . We have seen Apple shifting its component supply chain from Japan to Taiwan (Chinese Taipei) and then to China in past decade. Domestic smartphone brands like Huawei, Oppo, Vivo and Xiaomi have also grown their domestic supply chain along with increasing manufacturing scale.

Figure 121: China/HK: 21% of total no. of Apple suppliers Figure 122: Almost half of production sites based in China

100% 100%

80% 80%

60% 60%

40% 40% 12% 21% 47% 44% 20% 20%

0% 0% 2018 2013 2018 2013

US China (incl. HK) Taiwan Japan Korea Others US China Taiwan Japan Korea Others

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates

Asia Technology Strategy 84

4 November 2019

We have seen Apple supply chain moving from the US and Japan to China in the past five years. China suppliers accounted for 21% of the total number of Apple suppliers in 2018, from 12% in 2013. 47% of manufacturing locations were based in China in 2018, vs 44% in 2013. As the major EMS is located in China, its supply chain has established production capacity in China in order to shorten the lead time and reduce logistics cost. Chinese components suppliers are well established in certain components such as acoustic (speaker, receiver), haptic, cover glass, wireless charging module, connector/cable, battery module, etc., in the Apple supply chain, with a major share allocation (over 50%). We see domestic supply chain as sufficient to supply key components for Chinese smartphones and we believe domestic substitution is a key trend for components supply chain. We see components still relying on foreign suppliers including CIS, VCM, MEMS chips, HDI/SLP PCB, battery cell and MLCC. We analyse the current market structure, how China would develop its home industry and the near-to-long-term prospects by key sub-sectors—camera components (lens, module, CIS, 3D sensing), PCB/substrate, casing, antenna, acoustic, power supply/battery and MLCC. Display is discussed in a separated session.

Figure 123: Domestic smartphone component suppliers Cover glass Casing Lens, Biel Crystal Lens, Biel Crystal, BYDE, AAC, Everwin, Camera Tongda, CN Lens: Sunny, AAC, Innovations O-film, Star JuYu, LceOptic, Huaxin EMS/ODM Module: O-film, BYDE, Huaqin, Sunny, Q Tech, Wintech, Longcheer, Holitech, Truly, SIM Tech Luxvision, Shine, SZSeasons Fingerprint CIS: Omnivisoin, Module: O-film, Truly, Galaxycore, Q Tech, Holitech Superpix Sensor: Goodix, Silead, VCM: Hozel, ZET, Fortsense Shicoh, JssOpitcal Haptic/vibrator Display AAC, Luxshare, BOE, Tianma, Jinlong CSOT, Visionox, Everdisplay, Truly Battery Acoustic Sunwonda, Desay, AAC, Goertek, ATL Luxshare Connector Antenna Luxshare, Sunway Sunway, Speed, AAC, Luxshare, MLCC JESONcom, Deman Fenghua, Three Circle, Torch, PCB/substrate Hongyuan, SYE, WUS, SCC, Sinocera, Sunlord, Avary Hongda

Source: Company data, Credit Suisse estimates Acoustic Current market structure For miniature acoustic products, including receiver, speaker and MEMS microphones, we see the Apple supply chain relatively concentrated, and the Android market relatively fragmented. A few major products are driving growth and upgrade trends in the acoustics area.

Asia Technology Strategy 85

4 November 2019

We expect an upgrade trend for Android acoustics, e.g., AAC’s SLS or Goertek’s SBS penetrated mid-tiered models, expecting uptrend in both shipment/ASP. AAC also released Classic SLS speaker for high-end smartphones, with increased vibration area, new metal housing design, offering 40% improvement in acoustic performance and better noise cancellation and waterproofing abilities. High-end Android speaker/receiver ASP is ~US$1-2, or about half the ASP of similar components of iPhone. For low-end, ASP is only ~US$0.5. We estimate the Android acoustic market by our Android shipment units, and the market share of AAC acoustic components among Android supply chain is ~17%, and Goertek ~5% in 2018, and expect AAC’s share to expand to 30-40% and Goertek’s market share to 9-12% in 2020/21E. We see another strong growth driven by TWS earphones in the acoustic area. Counterpoint forecasted TWS shipment to reach 27 mn units globally in 2Q19, up 54% vs 17.5 mn in 1Q19. 2019 shipment is expected to reach ~120 mn units, with YoY growth of 160%+. For Apple AirPod, we forecasted global shipment to reach 40-50 mn/55-60 mn units in 2019/20E. For Android TWS, Samsung (Galaxy Bud) is the leader in the Android camp and Counterpoint estimated it having ~8% market share. With the release of Huawei TWS FreeBuds 3, together with the release of new Android smartphone models, from Huawei and Xiaomi, we expect the market share from Android TWS to increase. For global smart speakers, Canalys estimates continuous strong growth in 2Q19, with global shipment reaching 26.1 mn units, up 55% YoY, with Echo still the market leader (25.4% share), followed by Baidu’s 17.3% market share. Within the Top 5 players, three Chinese players accounted for 44% share in 2Q19 (up from 31% in 2Q18). We believe that smart speaker growth will mostly benefit MEMS microphone and speaker module suppliers, as smart speakers usually use MEMS microphone arrays (usually from 2- 12 MEMS microphones). MEMS microphone arrays provide better far-field voice recognition, noise reduction, reverberation elimination, echo reduction, etc. Goertek mentioned it has a large order pipeline and targets mass production in 2020. MEMS microphones are seeing a growth trend, driven by wide adoption in smartphones (e.g., iPhones), and strong growth of smart speakers and smart home appliances, as their voice recognition function relied on MEMS speaker arrays, which also helped realise the noise reduction function. We estimate the MEMS microphone market size to be US$1.2 bn-1.7 bn in 2019, with expected CAGR in the low-teen level over 2019 -2025. MEMS microphone has a smaller package size, lower impedance, and more scalable with better sound quality than traditional (ECM) microphones, and is widely adopted in smartphones, laptops/PCs, cameras, and automobiles. Competitive landscape Acoustic product GM varies in the range of 25-40% in recent years, and we see AAC’s acoustic components’ average GM higher than Goertek’s. MEMS components GM is in the 25%+ level in 2018. However, the Luxshare/Merry joining the Apple acoustics supply chain has driven down the acoustics GM, especially for AAC which lowered from ~41% in 2017 to ~37% in 2018 and ~27% in 1H19, and we expect it to stay in the 30%+ level up to 2021. In the Apple supply chain, we see AAC, Goertek and Merry/Luxshare as the Top 3 suppliers for acoustics products, including receivers and speakers, for different products such as iPhone, iPad and Apple Watch. In the speaker market, Goertek is still the market leader, with 40% or higher market share, but is likely to reduce to below 40% from 2020E—Merry and Luxshare have quickly gained market share (more from AAC) since late 2017 by entering the Apple supply chain to supply for iPhones. In the receiver market, we expect Merry and Luxshare to quickly gain share from 2018 and rise to the No. 1 position from 2019, with ~40% market share. Overall, Merry and Luxshare were the biggest share gainers since 2018 in the Apple supply chain.

Asia Technology Strategy 86

4 November 2019

Figure 124: Speaker market share and estimates Figure 125: Receiver market share and estimates

60% 50%

50% 40% 40% 30% 30% 20% 20%

10% 10%

0% 0% 2016 2017 2018 2019E 2020E 2021E 2016 2017 2018 2019E 2020E 2021E

AAC Goertek Merry/Luxshare AAC Goertek Merry/Luxshare

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates For Android acoustic (speakers and receivers) supply chain, we estimate market leaders AAC and Goertek accounting for ~17%/5% in 2018; other players include Knowles (KN.O), Merry (2439.TT), Hosiden (6804.TYO). The market for Android micro speakers and receivers is still more fragmented than Apple acoustics. For TWS, we believe Luxshare and Goertek will continue to gain market share from Inventec and become the dominant suppliers for AirPods from 2020. Goertek sees strong growth in its hearable business in the next two to three years, driven by AirPod and other brands. The company’s investment in common equipment for AirPod with process and efficiency reached the industry level, and it claims that its TWS’ yield and efficiency have exceeded customer request.

Figure 126: AirPod shipment estimate by supplier

60 (mn) 50 40 30 20 10 0 2017 2018 2019E 2020E

Luxshare Goertek

Source: Company data, Credit Suisse estimates MEMS microphone: The Top 3 suppliers of MEMS microphones are Knowles, AAC and Goertek. In terms of global market share in MEMS microphone, we estimate Knowles/Goertek/AAC to be ~50%/15%/9%. Other players include ST Microelectronics, BSE, NeoMEMS, Hosiden, Sanico Electronics, Bosch, MEMSensing, and InvenSense. The core competitiveness for MEMS microphone is the MEMS IC. For a long time, the MEMS IC supply was largely dominated by Knowles. But we saw AAC invested in Vesper Tech to produce MEMS IC, ~10% MEMC IC is self-designed, and we expect this share to continue to go up. We believe this technology advantage could help increase AAC’s market share in smart speakers. For speaker/receivers, as the technology has been more mature, we see the growth is more tempered. Luxshare focuses on earphones for entertaining-use and high-power speakers. Luxshare’s product offering focuses on high-quality speakers, capturing the strong demand from smart speaker growth. The entertainment business sector will also benefit from the smart home-related demand. Luxshare’s JV leverages the client base and resources from parent companies. Asia Technology Strategy 87

4 November 2019

Divergence in development directions for Chinese acoustic leaders: Goertek and AAC have different development directions besides their major products in acoustics—Goertek focuses on expanding its footprint in wearables, AR/VR products, integrating capabilities for components, accessories and overall solution assembly; while AAC is focusing on expanding its components offering to optics, such as WLG lens and precision components for antenna uses for 5G smartphones. The Chinese game plan Similar to other components sectors such as casing and camera modules, government grants/subsidies have been in the single-digit % of NI for AAC and Luxshare in recent years; but for Goertek, it was much higher, at double-digit %, reaching 30% in 2018. Also, the government grants/subsidies % of NI for Luxshare are relatively low and in a decreasing trend, while that for AAC and Goertek were in an increasing trend for the past few years, implying a lower reliance on government subsidies for Luxshare and stronger execution vs its peers. The majority of Goertek’s government subsidies was from the Weifang City or Shandong Province—for example, 81% in 1H19 subsidies was from Weifang City’s Corporate Innovation Development Fund, and 57% in FY18 subsidies was from Shandong Province’s special fund to support the growth of super-large-scale companies in the province. The subsidies usually came in the form of awards or grants, and mainly support R&D, equipment upgrade/expansion, commercialisation of new technologies, talent attraction, etc. Similar to other high-technology companies, acoustics players also enjoy VAT refund and preferential tax treatment from the Chinese government. For Luxshare, the government subsidies include support for technology revamp/upgrade, and some refunds regarding land use. For AAC, the government grants are related to engagement in high-tech business, employment of expatriates, and technologically advanced staff, as well as for acquisition of PP&E. To reduce reliance on overseas IC supplies, Chinese players also have developed/developing their own chipset—for example, Goertek has built a factory to produce MEMS chips. However, the relatively lower reliance on government funding indicates Chinese players’ have developed in-house technology organically, which helped them achieve global leadership positions in the acoustics market. Such in-house capabilities will likely support them to further enhance new technology capabilities. We see a high R&D expense share % to revenue from the Chinese acoustics providers: AAC reached 11%, while Luxshare reached 7% and Goertek ~5% by 1H19 results. Continuous R&D provides a strong support for the technology advancement of Chinese acoustics players.

Figure 127: Government subsidies as % of NI for leading Figure 128: R&D expense as % of revenue of leading Chinese Chinese acoustics companies acoustics companies

35.0% 12.0%

30.0% 10.0%

25.0% 8.0% 20.0% 6.0% 15.0%

4.0% 10.0%

5.0% 2.0%

0.0% 0.0% 2015 2016 2017 2018 1H19 2015 2016 2017 2018 1H19

AAC Goertek Luxshare AAC Goertek Luxshare

Source: Company data Source: Company data

Asia Technology Strategy 88

4 November 2019

View on near-term and mid-to-long-term development We see low overseas reliance on acoustic components, but continuing reliance on overseas codec chip and MEMS chip in mid-to-high-end smartphones. We expect Chinese players to continue to gain shares in key areas and see that they have already achieved dominant market shares in the Apple acoustics supply chain, for speakers and receivers. AirPod production is expected to be also largely divided between Luxshare and Goertek. In the MEMS microphone market, Chinese players are quickly catching up with Knowles. Goertek and AAC are producing MEMS IC and looking to increase the self-supply ratio of MEMS IC. We expect the Chinese players to take longer to realise domestic substitution for MEMS IC—more of a medium-term effort (three to six years). In the smart speakers market, we are already seeing an increase in market share from both the end-product perspective—with Baidu, Alibaba and Xiaomi entering the Top 5 global brands—and the supply chain perspective—with increasing share gains by Chinese players such as AAC and Goertek. Overall, we think the opportunity for China to realise domestic substitution in acoustic components is high, with the majority to be realised in the short term (within three years), except for key components such as MEMS IC which may take longer, likely in the medium term (three to six years).

Antenna Current market structure We focus on antenna for mobile devices in this section. The major driver for mobile antenna is still from smartphones, with trends of a rising number of antennae and of modularisation (integrated packaging with other components). We forecast a 10% CAGR for the market (shipment demand) over 2018-21, driven by 5G upgrade. The top vendors in the antenna market are Amphenol, Pulse, Molex, Skycross, Galtronics, Sunway, Speed, JESONcom, Auden, Deman, Ethertronics, Sky-wave, 3gtx, and Southstar.

Figure 129: Global antenna shipment to grow at 10% CAGR for 2018- 21E

10,000 (mn) 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E

Handset antenna Notebook antenna Tablet antenna

Source: Credit Suisse estimates FPC antenna is a common type of antenna used in mobile devices. Polyimide (PI) is the major type of materials used for flexible antenna, given its cost advantage, high resistance to humidity, and good heat dissipation. In the value chain of PI or modified PI, we see the US, Taiwan (Chinese Taipei) and Japan companies dominating in material/film supply space, and the US and Taiwan (Chinese Taipei) companies dominating in the flexible copper clad laminate (FCCL) space. China and Taiwan (Chinese Taipei) companies are mature in flexible board manufacturing for PI type antenna.

Asia Technology Strategy 89

4 November 2019

Figure 130: PI/MPI antenna value chain Material/film FCCL board Du Pont (US) Du Pont (US) Avary (TW) Taimide (TW) Shengyi (CN) MFLEX (CN) Mortech (TW) Taiflex (TW) Flexium (TW) UBE (JP) Thinflex (TW) Career (TW) Mitsubishi Gas (JP) AEM (TW) Source: Company data, Credit Suisse estimates Apple adopted liquid crystal polymer (LCP) antenna from 2H17 iPhone and then moved one LCP to MPI in 2H19 for LAT (Low Antenna) on cost concerns, while UAT (Upper Antenna) remains using LCP materials. In iPhone antenna supply chain, we see as Murata dominant in upstream supply and involved in FPC and module assembly. Amphenol and Luxshare are another two LCP module suppliers. In the general LCP antenna value chain, Japan companies are dominant in LCP resin/film supply. Chinese companies are able to provide downstream processes such as FPCB and module assembly. Sunway could get 30-40% share allocation in Macbook and iPad FPC antenna, while Luxshare could get 40- 50% share allocation in iPhone LCP antenna module, in our view.

Figure 131: LCP antenna value chain LCP resin LCP film LCP FCCL LCP flex board LCP module Sumitomo (JP) Sumitomo (JP) Sumitomo (JP) Sumitomo (JP) Sumitomo (JP) Rinkuru (JP) Murata (JP) Murata (JP) Murata (JP) Amphenol (US) Toray (JP) Kuraray (JP) Rogers (US) Rogers (US) Murata (JP) Polyplastics (JP) Du Pont (US) Shengyi (CN) MFLEX (CN) Luxshare (CN) UENO (JP) Gore-Tex (JP) Thinflex (TW) Kinwong (CN) Sunway (CN) Mitsubishi (JP) WOTE (CN) Career (TW) ECT (CN) Celanese (US) Pret (CN) Speed (CN) Solvay (EU) AAC (CN) WOTE (CN) Forewin (CN) Pret (CN) Source: Company data, Credit Suisse estimates Chinese game plan We think FPC antenna is a component that China learns from and grows with its customers. Back to 2000, or the Nokia and Motorola generation, Laird, the largest antenna supplier, entered the China market and set up production plants in China. Sunway acquired Laird (Beijing) in 2012 and entered the Apple supply chain later. Chinese companies entered the market by mergers and acquisitions in the early stage. Under the Made-In-China 2025 policy, we do not see a specific roadmap and goal setting for FPC antenna, but it identifies the key technology, including large-scale array antenna for communication infrastructure. Given the target China has set to build up its capability for 5G network, antenna design is also a key technology to develop its in-house capability. The upstream supply chain is still dominated by overseas vendors, while China has been building its capability downstream, including flexible boards and module assembly. Shengyi is the domestic vendor that is able to supply FCCL. We see Sunway, Speed, Luxshare, JESONcom, Deman as the major domestic players, and AAC as an emerging player. We do not see a massive subsidy supporting this industry development—it depends on the local government’s investment promotion and industry development plan. For the 5G antenna development project, local companies can apply for R&D grant and subsidy. We also believe the Chinese government would support the upstream development on materials/film as well as CCL.

Asia Technology Strategy 90

4 November 2019

View on near-term and medium-to-long-term development We believe China will strengthen its domestic value chain and ecosystem by moving upstream in the FPC antenna industry, somewhat overlapping with the PCB industry. We think industry/enterprise investments will move from FPCB and antenna module assembly to CCL midstream in next one to three years. Besides, local government or institute will continue to invest in R&D for 5G antenna design, materials/resin development, etc. In the long term, China would like to catch up with Japanese or US chemical companies in resin and film manufacturing. 5G smartphone antenna upgrade and local substitution are two opportunities for local FPC antenna suppliers. Challenges remain in terms of insufficient domestic materials supply, but M&A or equity investments in overseas suppliers would be a way to build up a connection in the upstream supply chain. Casing Current market structure

In the casing industry, current major materials include plastic, metal, glass, glastic and ceramic. Plastic is commonly used for mid-to-low-end smartphones, while metal casings are used in mid-end models, glass casings are used more in mid-to-high-end models. Glastic casing offers a cheaper solution than glass but assembles some qualities similar to glass, which has been adopted in mid-to-low-end models. We see 3D glass casing in the uptrend, driven by (1) improving yield, (2) decreasing costs (cheaper than ceramic), and (3) better suited for 5G smartphones as glass casing will not block RF signals as metal casings do. Ceramic casing still faces the challenge in scale production and yield, leading to higher ASP, so it is adopted more by flagship models. Glass casing is also suitable for wireless charging as it does not block electromagnetic waves. Within glass casing, 3D glass is gaining popularity along with rising adoption of flexible OLED panels. Glass casings are also adopted in smartphones, tablets, smart TVs and wearable products, due to their pricing and quality advantages. Heat bending machine is no longer the bottleneck for 3D glass manufacturing, with a lower share in the overall cost structure on technology maturity in the past two years. In global smartphone back-cover shipment, penetration of plastic casing for smartphone back-cover decreased from 89% in 2011 to 35% in 2018 and will likely rebound to 40% due to glastic technology. Metal casing replaced plastic along with the launch of unibody metal casing in iPhone. The adoption increased to 45% in 2017, but decreased in 2018 and will further decrease in 2019-21. The trend of metal cover will become the mid-frame mechanical structure, supporting glass-casing smartphones. Glass casing’s share reached 34% in 2018, surpassing metal casing’s 30.2%, and it is expected to further expand to 39% in 2019E, mainly driven by 3D glass casing, whose share is expected to reach ~20%, almost on a par with 2D/2.5D glass casing’s share. We forecast ceramic casing reaching 6% share by 2021.

Asia Technology Strategy 91

4 November 2019

Figure 132: Global smartphone back cover share – increasing Figure 133: Glass to achieve 40%+ penetration on back-casing share of glass casing mainly driven by 3D in China in 2021

100% 0% 0% 0% 0% 0% 0% 0% 3% 9% 3% 0%3% 4% 6% 4% 6% 100% 2% 2% 3% 4% 10% 6% 6% 14% 6% 3% 5% 6% 6% 90% 3% 11% 15% 10% 20% 90% 13% 19% 10% 19% 24% 11% 25% 80% 24% 26% 80% 28% 20% 37% 70% 39% 70% 11% 20% 51% 15% 60% 45% 22% 60% 39% 15% 20% 25% 50% 100% 100% 98% 12% 50% 30% 18% 92% 9% 89% 9% 7% 40% 83% 40% 80% 80% 76% 68% 30% 57% 30% 51% 20% 41% 39% 42% 44% 42% 20% 42% 41% 39% 35% 40% 10% 10% 0% 0% 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E Plastic Metal casing Glass 2/2.5D Glass 3D Ceramic Others Plastic Metal casing Glass 2/2.5D Glass 3D Ceramic Others

Source: Credit Suisse estimates Source: Credit Suisse estimates Android glass casing: We see the majority of HOVX flagship and high-end models (roughly US$500+ models) adopting 3D glass casing in 1H19, and shipment could add up to ~27 mn units. We think this trend will continue next year. We think competition in 3D glass is still manageable, with Biel Crystal (private), Lens Tech and BYDE sharing the Android market. However, we have seen slow growth with flattish market share and intensified competition in the US$300-500 segment, which could limit the adoption of 3D- glass unless ASP can be significantly reduced. For Apple glass casing, our check suggests a more complicated process and potential composite change in 2H20 iPhone glass casing. In terms of ASP change, our check suggests 40%+ ASP increase in back glass, and next year we expect another double-digit growth on composite change. Glastic casing: An important direction is the development of glastic casing, which offers a lighter weight but look and feel close to glass cases. Leading Chinese players in this field include CN Innovation and Tongda. Tongda’s IMT (In-Mold Transfer) technology could improve anti-fingerprint, wear resistance qualities. Tongda also expects rising ‘glastic’ adoption and its share gains at Samsung and Vivo (a new customer from 2019). Tongda will also unveil a next-generation 3D ‘glastic’ casing, ICM (injection compression moulding) in 2H19. Despite average glastic casing pricing likely to decline ~US$0.5 per year, as for the current 2.5D glastic, the new 3D ICM glastic, with 20%+ higher pricing, should help stabilise the corporate ASP.

Figure 134: Global smartphone glass demand (front + back) Figure 135: China smartphone metal-casing peaked in 2017 mn units 2,500 600 mn units

500 2,000

400 1,500

300 1,000 200

500 100

0 0 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E

2D 2.5D 3D Metal casing Metal mid-frame

Source: Credit Suisse estimates Source: Credit Suisse estimates

Asia Technology Strategy 92

4 November 2019

Ceramic casing: Besides the cost/yield issue, we see another challenge—the electromagnetic shield of ceramic casing is stronger than glass but weaker than metal, yet it is to be tested for the applicability for 5G smartphones. We expect the penetration rate to be in single-digit % by 2021E. Currently, Lens Tech and Three Circle (300409.SZ, not covered) have implemented production lines for ceramic casing.

Figure 136: ASP trend down across multiple casing materials Figure 137: Metal mid-frames growing fast globally/in China

20.00 800.0 250% 18.00 700.0 200% 16.00 600.0 14.00 150%

12.00 500.0 100% 10.00 400.0 8.00 50% 300.0 6.00 0% 200.0 4.00 -50% 2.00 100.0

0.00 0.0 -100% 2018 2019E 2020E 2021E 2011 2012 2013 2014 2015 2016 2017 2018E 2019E 2020E 2021E

Plastic Metal casing (front + back) Metal midframe 2.5D 3D Ceramic Global shipment China shipment Globa YoY % China YoY %

Source: Credit Suisse estimates Source: Credit Suisse estimates Competitive Landscape

Overall, the casing market is fragmented, and we see Taiwan suppliers (Foxconn/Catcher/Casetek) mainly focusing on metal casing for Apple products, including Macbook, iPad and iPhone. Taiwanese suppliers hold over 90% share (in terms of value) in the MacBook business, with Everwin supplying the bottom casing for Macbook. Our channel check indicates Apple watch casing could be provided by Chinese suppliers with plants in Huadong region, Taiwan suppliers have less exposure, and Casetek getting into in 2H19 and 2020, probably starting with small shares. We see Everwin wanting to enter the Macbook casing chain but not qualified yet on lagging in metal casing technology. Thus we see the two groups of players competing in different fields. In Android, FIH, BYDE, Tongda, Everwin and AAC are the major players. For Apple products glass casing, we think Lens and Biel remain the two key players, with difficulty continuing to increase. BYDE and Foxconn are still lagging much behind and are finding it hard to get in as they do not meet the quality standard and yields are very low. In the Android supply chain, Lens, Biel and BYDE are the major players for Android phone casing. Lens Tech is the leader in iPhone/iPad cover glass, with 55%+/45% share expected through 2019-21E, while it’s also increasing its market share in Chinese smartphone/tablet cover glasses.

Asia Technology Strategy 93

4 November 2019

Figure 138: Lens Tech’s share in iPhone cover glass of 55%+, Figure 139: Lens Tech’s high market share for iPad and and increasing for Chinese smartphone cover glass Chinese tablet cover glass

120% 60%

100% 50%

80% 40%

60% 30%

40% 20%

20% 10%

0% 0% 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E

iPhone cover glass Samsung smartphone cover glass Chinese smartphone cover glass iPad cover glass Samsung tablet cover glass Chinese tablet cover glass

Source: Credit Suisse estimates Source: Credit Suisse estimates BYDE is a well-rounded casing material provider, covering all major casing materials. It’s a market leader in metal/plastic/glass casing for Android smartphones. As per the company, its metal casing market share increased from 30-35% in 2017 to 37-28% in 2018 and 40%+ in 1H19 (despite revenue decreasing), and it’s also the No. 1 provider of plastic casing. The company expects its market share in glass casing to be near 30% for FY19, and was ranked in the Top 3 in terms of capacity and shipment in the past three years for glass casing. BYDE also expects glass casing and ceramic casing to report strong double- digit % YoY growth, and ceramic materials to expand their use from external/decoration materials to functional components, such as in RF modules for 5G. Tongda is one of the major players in glastic casing. Its management maintains a 140 mn/80 mn-200 mn smartphone unit estimate for 2019/2020, in which ‘glastic’ casing is expected to account for 70%/75-80% of its shipment, implying 50% unit growth. Chinese game plan

Chinese government subsidies for smartphone casing companies were in double-digit % of NI in 2018, with that for Lens Tech and Everwin much higher than peers. The subsidy % of NI generally increased during the past few years—for example, BYDE’s subsidy % of NI increased from 2% in 2016 to 13% in 2018. The dependency on foreign suppliers is more in the equipment-end—CNC machines can be internally developed by BYDE, but other machines such as AF/PET coating and PVD (physical vapor deposition) coating machines are mostly provided by foreign suppliers.

Figure 140: Government subsidies/grants as % of revenue Figure 141: Government subsidies/grants as % of NI

3.5% 250.0%

3.0% 200.0%

2.5% 150.0%

2.0% 100.0%

50.0% 1.5%

0.0% 1.0% 2015 2016 2017 2018 1H19 (50.0%) 0.5% (100.0%) 0.0% 2015 2016 2017 2018 1H19 (150.0%)

BYDE Lens Tongda Everwin BYDE Lens Tongda Everwin

Source: Company data Source: Company data

Asia Technology Strategy 94

4 November 2019

Government grants and subsidies come more from local governments in the provincial and city levels (particularly where their headquarters or factories locate), and in the form of grants or awards, to the company itself or its local subsidiaries. The government grants and awards come by projects, and are mostly related to key product R&D, automation, equipment upgrade and other projects, and each grant/award varies by size. For example, BYDE’s total government grants and subsidies for FY18 was of Rmb287 mn and it was divided between assets and income. The total is an aggregate of multiple subsidiaries in Huizhou, Shantou, Xi’an, etc., and of other entities including the Bureau of Human Resources and Social Security, for the purpose of technical transformation, R&D, employee stability, logistics and opex. View on near-term and medium-to-long-term development In the casing market, Chinese players are already the market leaders in terms of market share and technologies across different casing materials for Android smartphones and tablets. For the Android supply chain, BYDE leads in metal, plastic, and glass casings, and is aggressively pricing and has capacity expansion strategy. In terms of quality and segment, FIH and AAC target high-end mid-frame/metal casing, BYDE focuses on mass market and Tongda focuses on mid-low-end segments. Lens Tech is expanding its share in Android glass casing, with new products of rising precision, strength and more colour options. Tongda is leading in glastic casing and innovating with ICM and 3D glastic casing materials. For Apple glass cover/casing, Chinese companies (Lens Tech and Biel Crystal) are also the market leaders, with advanced new products with rising precision from unibody design and higher strength/more colour options. However, Chinese players are trying to get into Apple metal casing but still lag behind their peers in, and it’s unlikely for Chinese players to get into Apple metal casing supply chain in the short term as it is lagging in technology advancement and precision manufacturing. In the Android supply chain, we see low reliance on overseas suppliers. Domestic vendors dominate in plastic casing, metal casing, glass casing and ceramic casing. But reliance on overseas equipment makers is still high as the casing manufacturing process involves multiple steps and require custom machines with specific/high-entry barrier technology, some of which can be sourced only from foreign suppliers so far, e.g., for a certain coating, photolithography, and silk printing machines. However, BYDE has developed and adopted some in-house equipment, including CNC machines, and it is also adopting more general equipment that can be used for multiple materials/product lines. The current geopolitical situation, including the US-China trade dispute, will likely accelerate domestic substitution, but many equipment can be procured in Japan and Korea.

Camera module Current market structure The CCM (Compact Camera Module) market is expected to grow from US$27.1 bn in 2018 to US$45.7 bn by 2024, with a five-year CAGR of 9.1%, according to Yole. CCM shipment growth is entering a slower and stabilising growth phase, yet the upside mainly comes from dual/triple/quad cams. We have observed several trends in the smartphone lens industry: Increasing multi-cam adoption with more complex designs: TSR forecasts multi- cam (including dual cam) adoption to increase from 5.3% in 2016 to 85.9% in 2021, with 52.2% of dual cam and 33.7% triple cam or above, representing a 75% five-year CAGR of multi-cam smartphone unit shipment for the period. Chinese CCM makers benefit from rapid multi-cam adoption among Chinese smartphone makers. Sunny, Luxshare, O-film and Q Tech are capable of dual-cam/triple-cam production. Globally, LG Innotek and SEMCO are the leading in high-end CCM products, while Sunny and Luxshare are leading in the China market.

Asia Technology Strategy 95

4 November 2019

Figure 142: Global smartphone CCM shipments (by single Figure 143: Multi-cam penetration increasing in smartphone CCM) shipment

5.0 4.7 0.18 (mn units) 1,600 70% 4.5 0.16 4.0 16% 16% 1,400 60% 60% 4.0 0.14 3.5 3.5 3.2 3.3 1,200 0.12 50% 3.0 1,000 0.1 40% 40% 2.5 800 0.08 2.0 30% 0.06 600 1.5 6% 20% 400 1.0 0.04 17% 2% 10% 0.5 0.02 200 5% 0.0 0 0 0% 0% 2015 2016 2017 2018 2019E 2015 2016 2017 2018 1H19

Global CCM shipment (mn units) YoY % Single or below Dual Triple or above % of dual or above

Source: TSR Source: IDC Module assembly technology trends: Within the current CSP (chip size package), COB (chip on board) and FC (flip chip) technologies, with a decreasing module size, thickness and increasing light transmittance. The cost of assembly line for FC is also 30-50% higher than that of COB assembly line, and the yield for FC is lower due to the higher technical difficulty. COB is still the mainstream technology among Android smartphone cameras. With the penetration of full-screens comes higher demand for miniature and thinner CCMs and with better cost. So far, mainly iPhones have adopted FC in its cameras, and suppliers include LGI, Sharp, O-film (Sony) and Cowell. O-film acquired Sony’s factory in South China and gained the FC technology. Sunny developed its MOB (Molding on Board) and MOC (Molding on Chip) technologies for full-screen front cameras, while O-film developed CMP (Chip Molding Package). Dual and triple cameras require higher precision and algorithms: Dual/triple cameras have higher algorithm requirement and packaging precision for image sensors, lenses and VCM, which can be realised by active alignment (AA). AA equipment are mainly provided by in-house tooling, ASM Pacific, Pioneer and HyVision. Sunny Optical has developed its own AA equipment. It claims that its in-house AA equipment is used for high- end models, while the ASM Pacific equipment is used in mass models. New technology innovations increasing: ToF penetration is increasing in high-end smartphone models, driving the growth of WLO (wafer level optics) and VCSEL technology development. Periscope-style products are also gaining popularity. WLO techniques is mostly adopted by ams, Himax, and AAC, driven by 3D camera growth. As per a Yole report, WLO is expected to grow at a CAGR of 55% in the next five years and reach US$1.6 bn, primarily driven by smartphone growth demand. We see limited supply chain in the domestic market, including VCSEL wafer fabrication, DOE, and sensor for 3D sensing measurement. However, Vertilite has designed its own VCSEL chip and outsourced to VPEC and Win Semi. It has passed Huawei's qualification and started volume production in 3Q19. Other components for 3D depth sensing, such as CMOS IR camera, have lower entry barriers and can be provided by domestic module suppliers like Sunny, Q Tech or O- film. CIS increasing share of higher resolution: Based on a TSR report, for rear cameras, we see 13MP+ CIS sensors share increasing, in particular for 40-48MP CIS, which is expected to reach ~23% by 2023E; while 10-13MP CIS is still mainstream, its share is expected to decrease from 50%+ in 2018 to ~29% in 2023E. In front cameras, although 5MP and 7-8MP CIS are still mainstream, the share of 10MP+ CIS is increasing quickly and is expected to exceed 7-8MP and take the largest share of ~32% in 2020E. The trend of rising resolution is benefiting CIS, lens, and module suppliers.

Asia Technology Strategy 96

4 November 2019

Figure 144: 10-16MP still mainstream but 32MP+ share Figure 145: 5MP and 7-8MP still mainstream, but 10MP+ increasing quickly (rear camera) shares increasing quickly (front camera)

100% 0% 0% 1% 100.0% 3.9% 10% 11.8% 90% 17% 22% 90.0% 16.9% 27% 26.1% 33% 31.9% 35.6% 80% 80.0% 41.5% 48.7% 70% 70.0%

60% 60.0%

50% 50.0%

40% 40.0%

30% 30.0%

20% 20.0%

10% 10.0%

0% 0.0% 2016 2017 2018 2019E 2020E 2021E 2022E 2023E 2016 2017 2018 2019E 2020E 2021E 2022E 2023E

<10MP 10-16MP 20-32MP >32MP <5MP 5MP 7-8MP 10-13MP 13MP+

Source: TSR Source: TSR Apple CCM makers: In high-end module, LGI has the majority of share (over 70%) in triple-cam and 3D sensing supply. In dual-cam and single-cam, Sharp could take over 50% share, while O-film is the new entrant. However, Apple is concerned about O-film’s capability and financial risk and is looking for an alternative. Android CCM makers. O-film has the largest CCM capacity, while Sunny and Q Tech are expanding. We see the three major players are key suppliers to HOVX CCM. As per TSR forecast, for CCM counted by single camera, O-film and Sunny will have the biggest market shares of 17% and 13% in 2019, respectively. Sunny supplies more on the high- end, while O-film and Q Tech have higher share of mid-/low-tier CCM supplies. In Huawei, O-film is the major supplier for mid-to-low-end phones, and Luxvision and Sunny are for high-end models. In OPPO/Vivo, Truly and Q Tech are for mass models, while Sunny is for high-end models. O-film is a major CCM supplier for , while Xiaomi also uses SEMCO, Luxvision and Sunny for the Xiaomi series.

Figure 146: CCM market share of 2019E Figure 147: LGI/Sharp dominated iPhone CCM supply (2020E)

100% 0% 10% O-film 15% Others 23% 30% 18% 17% 80% Patron 3% 60% Sharp 65% 58% 3% Sunny Foxconn 13% 40% 78% 3% 70% SEMCO 3% 20% 25% 28% Luxvision/ Lite On Q Tech 3% 9% 0% Cammsys Holitech Single camera Dual camera Triple camera 3D sensing 3% Truly LGI 6% 5% MC Nex PowerLogics 6% 4% 4% LGI Sharp O-film

Source: Techno Systems Research, Credit Suisse estimates Source: Techno Systems Research, Credit Suisse estimates Increasing concentration by top players: Market concentration is expected for the top players in the CCM segment, and TSR forecasts the Top 3 players (O-film, Sunny, Q-Tech) to take 39% market share in 2019E from 19% in 2016. This is expected to be driven by the higher technology requirement and barrier from dual/triple and even quad cameras, in terms of higher precision and better algorithm and automation, while maintaining a relatively high yield and lower costs, which require R&D investment and technology accumulation. The small companies are disadvantaged in this competition.

Asia Technology Strategy 97

4 November 2019

Top Chinese players leading and gaining shares: When classifying the CCM by single cameras, we also see Chinese players leading in market share, with O-film and Sunny leading with 17%/13% share in 2019E respectively, followed by Q Tech/Holitech/LGI to reach 8.6%/6.0%5.8% share respectively, and other players in single digit % share according to TSR forecasts. This is driven by more system makers purchasing single cameras and assemble them into multi-cameras. forecast, We also see shares of many Japanese/Korean players (e.g., Sony, Cowell and CMK) have been shrinking, while top Chinese players are the biggest share gainers, including O-film, Sunny, Q Tech, and Holitech, comparing 2018 vs 2016. Sunny and O-film lead in multi-cameras: Sunny, O-film and LGI lead in multi-camera CCM shipment, while Q Tech and Truly have a higher share % of single-camera CCMs. A higher share % of multi-camera will also help increase the ASP of overall CCM.

Figure 148: Sunny, O-film and LGI with highest multi-camera Figure 149: Top Chinese players gaining shares while most shipment expected for 2019 other players are losing shares (2018 vs 2016)

(mn units) 60% 250

50% 200 40%

150 30%

100 20%

50 10%

0% 0 O-film Sunny Q Tech Holitech LGI Truly MC Nex Luxvision/ Foxconn Sharp Others Sunny O-film LGI PowerLogics SEMCO Sharp Luxvision Primax Truly Darling Q Tech Lite On

2019E 2016A 2018A

Source: TSR Source: TSR Chinese game plan Similar to camera lens, we have observed that most of the Chinese players have developed from lower-spec products such as lower-resolution single camera, with a lower entry barrier, to gradually enhancing their capacity and expanding into more complex structures and higher-spec products, such as dual/triple cameras. Higher risks for players with higher reliance on government funding: Government grants/subsidies are mainly related to technology enhancement of production lines and R&D. For the Top 4 Chinese CCM suppliers (O-film, Sunny, Q Tech and Holitech), government grants/subsidies % of NI has a diverse range, and we see O-film and Holitech having relatively high reliance on government subsidies—O-film with an average of ~20% of profit from 2015-17, and spiked to 150%+ in 1H19, and Holitech with average ~22% from 2015 to 1H19. Sunny and Q Tech have a relatively lower grants/subsidies % of NI, indicating a self-reliance of development path for both companies with higher competitiveness. Sunny’s government subsidies % of NI were 2.8%/8.0% in 2017/2018, and Q Tech’s were 3.8%/721.6% in 2017/2018 (2018 was a challenging year for Q Tech, with a very low earnings base), but down to 3.6% in 1H19 on earnings recovery. For Q Tech, the government grants were received from several local government authorities as a recognition of the company’s contribution towards the local economic development. Due to the large revenue scale of CCM players, government subsidies % of revenue was on average 1-2%, with Holitech much higher than its peers. As the CCM market has entered a slowing growth stage due to sluggish smartphone demand, and decreasing dividend from the industry with GM pressure in the upcoming years, even some large players such as O- film and Holitech are facing financing risk in adverse market conditions, impacting the performance of these companies. We see O-film has received significant

Asia Technology Strategy 98

4 November 2019

investments/financings from the governments (e.g., Nanchang government) and credit lines from local banks in 2019 to support its cash position.

Figure 150: Government subsidies was on average 1-2% of Figure 151: O-film and Holitech with highest government revenue across major Chinese CCM suppliers subsidies as % to NI among major Chinese CCM suppliers

3.0% 800.0%

700.0% 2.5% 600.0% 2.0% 500.0%

1.5% 400.0% 300.0% 1.0% 200.0%

0.5% 100.0%

0.0% 0.0% 2015 2016 2017 2018 1H19 2015 2016 2017 2018 1H19 (100.0%)

Sunny O-film Q Tech Holitech Sunny O-film Q Tech Holitech

Source: Company data, Credit Suisse estimates Source: Company data M&A was another method of expanding business footprint: Luxshare acquired the CCM business from Lite-On in June 2018 and formed Luxvision, expanding into the CCM market. Lite-On was one of key suppliers to Chinese Android brands such as Huawei and Xiaomi. As per TSR, Luxvision’s market share was ~3-4% from 2016 to 2018, and is expected to reach 3.4% in 2019E. Preferential tax treatment for tech companies: Most Chinese technology companies that qualified as National High-Tech firms, will receive a preferred taxation treatment (rather than the usual 25% corporate income tax rate). For Sunny Optical, its effective corporate income tax rate was in the range of 11-12%, and for other CCM peers’ in the low-teen % in general, with preferential tax treatment.

Figure 152: CCM value chain CCM 3D sensing module CIS Lens VCM CCM Assembly VCSEL DOE NIR sensor Filter Module Assembly Sony (JP) Largan (TW) Mitsumi (JP) O-film (CN) Lumentum (US) TSMC (TW) STM (Switzerland) Viavi (US) O-film (CN) Samsung (KR) Sunny (CN) SEMCO (KR) Sunny (CN) Ams (AT) Ams (AT) Sony (JP) PTOT (TW) Q Tech (CN) Omnivision (US) Genius (TW) JAHWA (KR) Q Tech (CN) Win Semi (TW) Himax (TW) Omnivision (US) Truly (CN) ON Semi (US) Sekonix (KR) TDK (JP) Holitech (CN) IIVI (US) Himax (TW) Jabil (US) Galaxycore (CN) Kantatsu( JP) Alps (JP) LG Innotek (KR) Finisar (US) Sunny (CN) SK Hynix (KR) AAC (CN) Shicoh (JP) Truly (CN) Vertilite (US) Luxvision (CN) G2Hysonic (KR) SEMCO (KR) LG Innotek (KR) LG Innotek (KR) Patron (KR) Sharp (JP) Source: Company data, Credit Suisse estimates View on near-term and medium-to-long-term development Chinese players have well established their capability in CCM (multi-cam) and handset lens. We think, in the near term (within three years), they will likely establish their leadership in 3D sensing and advanced optical solutions, such as WLO and periscope-style products. However, in the CCM value chain, key components such as CIS and VCM are still largely dominated by Japanese and Korean players despite some Chinese players making progress in these products, for example, Win Semiconductor for CIS and ZET, JSSOptical Technology and Shanghai B.L. Electronics for VCM. For handset lens, Sunny Optical is an important supplier to HOVX smartphones. We believe that in the long term we will see Chinese players establishing their ecosystem and building a domestic supply chain for optical components, including CIS, high- end lens and VCM, and gradually increasing their share and technology leadership.

Asia Technology Strategy 99

4 November 2019

Hence, Chinese local camera modules makers such as Sunny Optical and Q Tech are already taking a major share at local Chinese handset makers (Huawei/Oppo/Vivo/Xiaomi), not just for mid-to-low-end phones, but for high-end flagship phones as well. Meanwhile, a growing Chinese camera modules industry as well as its improving technology curve is offering better growth opportunities out of China by penetrating into the Samsung handset supply chain, for instance. Considering a rapidly rising multi-cam penetration at SEC, even for mid-to-low-end phones (i.e., Galaxy A series), relevant camera modules makers (both Korea or China camera modules suppliers) are all enjoying the total addressable market growth through shipment volume as well as blended ASP-hikes.

Figure 153: SEC handset camera suppliers have enjoyed blended ASP-hikes on rapidly rising multi-cam adoption.

Triple-cam (high-end, rear) Dual-cam (high-end, rear) Single-cam (high-end, rear) Quad-cam (mid/low-end, rear) Triple-cam (mid/low-end, rear) Dual-cam (mid/low-end, rear) Dual-cam (front) Single-cam w/ iris scanner (front) Single-cam (front) $0 $10 $20 $30 $40 $50 $60

Source: Company data, Credit Suisse estimates However, we believe that in the medium-to-long-term, the existing Korea handset supply chain will likely face more headwinds amid intensifying competition at the major client (SEC), either through market shares loss or ASP-cut pressures. As of recent, Samsung is even testing the ODM outsourcing strategy to Chinese makers. While we believe that the total smartphone volume outsourced will remain controlled (~20 mn in 2019E and up to 40 mn in 2020E) and only applicable to the very low-end A series which has little chance of ever becoming profitable, it seems there still is a downside risk to existing Korea camera modules supply chain (mainly for mid-to-low-end) such as Partron, Mcnex, and Powerlogics. In contrast, the strategy will likely provide better growth opportunities for local Chinese camera modules makers, in our view. For high-end flagship phones at Samsung, SEMCO may remain a key supply chain for longer, considering the needs for key optics technology protection. However, more ASP-cut pressure seems inevitable given the emerging alternative options from Chinese module makers with better technology readiness.

Figure 154: Samsung – Galaxy A series-mix by models (production units base)

Source: Company data, Credit Suisse estimates

Asia Technology Strategy 100

4 November 2019

CMOS Image sensor (CIS) Current market status According to data from Gartner, Sony was the leader of the CMOS image sensor (CIS) market in CY2018, with a 47% value share, followed by Samsung Electronics in the second and Omnivision in third positions. Sony commands a technical lead in high-resolution sensors for smartphones, supplying products fitted to many of the leading handsets on the market. Demand is expected to remain strong in the near term for CMOS used in smartphones. The trend toward equipping handsets with multiple and more advanced cameras is not only driving CIS volume growth, but also allowing suppliers to charge higher unit prices as the technology shifts to larger formats. Besides smartphone applications, demand is also expected to grow for sensors used in automotive applications, industrial machinery, and surveillance cameras. ON Semiconductor is the leader in automotive CIS.

Figure 155: Market share (value, all application) Figure 156: Market share (volume, >13MP, only for mobile phone)

Others Others 9% SK Hynlx 0% SK hynix OmniVision 1% 2% 5% Sony Galaxycore 47% 3% ON Semiconductor 6%

Omnivision 12%

Sony Samsung Samsung 56% Electronics 38% 21%

Source: Gartner, Credit Suisse Research Source: Techno Systems Research, Credit Suisse Research

Figure 157: Diffusion rate by camera resolution in global Figure 158: Multi camera diffusion rate in global smartphone smartphone market market

120% 70% 20MP< 16MP< 12MP< 8MP< % of multi total % of triple or more 60% 100%

50% 80% 40% 60% 30% 40% 20%

20% 10%

0% 0%

2011Q1 2013Q3 2015Q3 2018Q1 2010Q1 2010Q3 2011Q3 2012Q1 2012Q3 2013Q1 2014Q1 2014Q3 2015Q1 2016Q1 2016Q3 2017Q1 2017Q3 2018Q3 2019Q1

2017Q2 2017Q3 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 2016Q4 2017Q1 2017Q4 2018Q1 2018Q2 2018Q3 2018Q4 2019Q1 2019Q2 2015Q1

Source: IDC, Credit Suisse Research Source: IDC, Credit Suisse Research

Asia Technology Strategy 101

4 November 2019

Chinese game plan Huawei and other Chinese smartphone manufacturers are speeding up the adoption of multiple cameras per handset. More of their flagship models include a high-resolution sensor of at least 40MP. Along with Apple’s new iPhone, this is likely to drive demand in the CIS market. Going forward, we think demand will also increase for time-of-flight (ToF) sensors. This year, Sony has already begun supplying ToF sensors to Huawei, Oppo and other Chinese handset manufacturers for high-end models, and we think that Apple could introduce ToF functionality to the iPhone range in 2020. A rear ToF sensor could become a standard feature of high-end smartphones, potentially giving a further boost to demand in 2021 and beyond. Given that the Chinese CIS vendors are focusing their business development on 2–5MP products (for models such as the Galaxy Core), we think the market share status quo in the high-resolution CIS segment is unlikely to shift substantially going forward.

Figure 159: % of multi camera (triple and more) in major Figure 160: Estimates for multi camera diffusion rate Chinese smartphone makers

40% 80% Huawei Xiaomi vivo OPPO % of Multi Camera total % of Triple Camera

35% 70%

30% 60%

25% 50%

20% 40%

15% 30%

10% 20%

5% 10%

0% 0% 2018Q1 2018Q2 2018Q3 2018Q4 2019Q1 2019Q2 2015 2016 2017 2018 2019E 2020E

Source: IDC, Credit Suisse estimates Source: IDC, Credit Suisse estimates View on near-term and medium-to-long-term development In the near term, we expect CIS suppliers to continue expanding production capacity to cater to the growing demand. Sony plans to increase monthly capacity to 130,000 units by March 2021 (up from 105,000 in Apr–Jun-2019, based on 300 mm wafer). Over the medium and longer term, we see Sony focusing on the functional fusion of sensors for automotive applications (to integrate data taken from cameras, LIDAR and millimeter-wave radar). Through its alliance with Microsoft, the company is also targeting a business model that generates recurring income by channelling development resources into CMOS image sensors with embedded edge AI processing capabilities. CIS is one of the very few semiconductor areas where China has a top-tier player. We think OmniVision’s CIS technology could be one to two years behind Sony, but at a similar level to Samsung. In the automotive segment, OmniVision is ahead of Sony and Samsung, and only behind On Semi. Driven by the geopolitical tension, OmniVision has been taking in the mid-to- high-end-spec smartphone segment from Samsung since mid-2019 and also growing its share in China automotive market. OmniVision’s gap vs Sony should continue for one to two years as most OEMs customise CIS chips for flagship models, so it will take time for OmniVision to regain position in the premium segment. In five years, we expect OmniVision’s technology to be slightly behind, but close to Sony’s and its market share to grow from 12% in 2018 (per Gartner data) to the high-teens or 20s percentage in 2023.

Asia Technology Strategy 102

4 November 2019

Lens Current market structure

Strong volume growth from accelerated lens upgrade: According to TSR, Smartphone lens saw 6% YoY shipment growth in 2018, and Chinese lens suppliers saw the strongest growth of 56% YoY, driven by the faster multi-camera adoption, as triple/quad camera has become a mainstream feature for flagship smartphones, and triple/dual camera further penetrated into mid-end smartphones. We think the proliferation of other lens technology, including TOF lens and under-display optical lens, also contributed largely to the high shipment growth of Chinese lens suppliers. TSR forecasts that total smartphone shipment lens will grow 16% YoY in 2019, leaded by growth in 6P+ (which is estimated to grow 40% YoY), driven by accelerated lens upgrade including 7P launch and continued MP migration. 5P and 3P shipment are also expected to grow double-digit YoY in 2019, driven by triple camera set-up and depth sensing demand. Largan remains the largest player and dominates 5-7P lens market: Largan remained the largest smartphone lens supplier, with 32.8% market share, followed by Sunny optical (23.6%) and Kantatsu (5.9%), in 2018. The Top 3 suppliers’ market share is forecasted to decline slightly to 61.0% (from 62.3%) in 2019, but market share for the Top 3 in 5-7P lens market will stay at 78%. Largan is the largest vendor in 5-7P lens, with 48% market share in 2019, while Sunny dominates the 4P lens, with 44% market share. Takes time to bridge the technology gap; but good growth opportunities from new 5G features: We believe the launch of 5G smartphone, with new features such as video applications and VR/AR, could further accelerate lens spec upgrade. We expect 7P proliferation in high-end flagship smartphone in 2020 and 8P lens to start emerge, as we see lens element crucial to enhance aperture in the continued pursuit of image quality and increasing demand for video applications. Other new features that also require higher technology input include periscope lens (with increasing need to add more lens elements or glass lens to solve heat problem) and under-display lens (the light and size problem). Sunny started some shipment of 7P in late-2018; however, we don’t expect meaningful contribution in 2019, but a good pick-up in 2020 when 7P demand is growing. AAC’s shipment-mix, compared to other Chinese peers, sees higher exposure in 5-7P lens (61%) driven by its 13MP/5P lens to Redmi smartphones, in our view. In the last upgrade cycle— the migration to 6P lens—Sunny/Genius took an additional one to two years, compared to Largan, to enter mass production, and other Chinese suppliers today still focus on the 5P/4P and lower market. We see the gap of about two years between Largan and Sunny to achieve mass production yield for the new specs, but the gap will be gradually narrowing along with increasing focus on camera and spec upgrade acceleration from mobile networks. While Sunny and other advanced suppliers are going after high-end, other Chinese suppliers will also need to improve their technology to fill in for 5P/6P in order to support TSR’s forecasts of 16MP segment making up 30-50% of 2021-2023E smartphones. Overall, we think it would take a longer for Chinese smartphone vendors to rely on domestic supply.

Asia Technology Strategy 103

4 November 2019

Figure 161: Lens market share (2018) Figure 162: Lens element migration

(mn pcs) 5P & below 6P & above 6P+ adoption Largan

Others (TW) 6,000 30% 27% 33% 5,000 25%

Kantatsu 4,000 20% (JP) 6% 3,000 15% 2,000 10% Sekonix Sunny (KR) Genius Optical 1,000 5% (CH) 5% (TW) 0 0% 5% 24% 2016 2017 2018 2019F 2020F

Source: TSR (1H19) Source: TSR (1H19) Chinese players quickly taking up market shares: As per TSR estimate, comparing market share of 2018 vs 2016, Sunny increased by 12.7 pp, AAC increased by 3.5 pp, Huaxin Optical increased by 4.7 pp and O-film increased by 0.7 pp, while majority of other players see a decreased of their market share during this period. AAC’s optical revenue grew by 73%+ YoY in 1H19 to Rmb417 mn. Sunny’s handset lens shipment increased by 38% YoY in 1H19 in which 20MP+ accounted for 13.7%, optical components revenue grew 42% YoY and takes up 22% of the total sales. Sunny’s handset lens shipment was Global No. 2 and vehicle lens shipment is global No. 1 in 2018, with high-end dual/triple/quad cam shipment global No. 1. On the other hand, we see Japanese and Korean competitors’ shares are decreasing.

Figure 163: Lens industry leaders are deploying more R&D Figure 164: Chinese players are gaining market share while resources YoY others decreasing

12% 35.0%

10% 30.0%

25.0% 8%

20.0% 6% 15.0% 4% 10.0%

2% 5.0%

0% 0.0% 2017 2018 1H19 Largan Sunny Huaxin Optical AAC O-film Sekonix Kantatsu CHT/Xinhjuyu Genius Xingbang Others

Sunny Largan AAC 2016 2018

Source: Company data, Credit Suisse estimates Source: Techno Systems Research Chinese game plan Despite the general tax benefit such as VAT treatment for qualified high-tech companies and R&D subsidy from local governments, the reliance on government overall is relatively small, and most companies grew through their own organic expansion and R&D. This is partly due to a low entry barrier for camera lens, and the product complexity only increases with more complex lens structure and new technology innovation. We also see diversity among companies, e.g., O-film was quite relied on Nanchang government financing support. Relatively small reliance on government funding: Within the component segment, we didn’t see a lot of subsidy and special support from the government. For example, for leading Chinese lens company such as Sunny Optical, which received funding and awards from Yuyao local government, and its government grants has been increasing gradually every year, and totalled to ~Rmb400 mn in the past five years, with a CAGR of 78% (2014-2018). The grants are mainly supported by the government, and such subsidies are related to technology enhancement of production lines and R&D of tech projects.

Asia Technology Strategy 104

4 November 2019

Government grants/subsidies % of NI has a diverse range, and we see O-film has a relatively high reliance on government subsidies—an average of ~20% from 2015-17, and spiked to 150%+ in 1H19, but Sunny and AAC have a relatively lower grants/subsidies % of NI, indicating a self-reliance of development path for both companies. For AAC, the government subsidy was mainly to support constructing manufacturing plants and acquiring machines. AAC’s government subsidies % of NI were 0.2%/1.2% in 2017/2018, and Sunny’s government subsidies % of NI were 2.8%/8.0% in 2017/2018, both in much better positions than O-film’s 17.3%/-57.5% in 2017/2018. Government subsidies % of revenue was on average <1%, a relatively smaller impact.

Figure 165: Government subsidy was on average <1% of Figure 166: O-film has the highest government subsidy % as revenue across Chinese lens suppliers of NI among major Chinese lens suppliers

1.4% 200.0%

1.2% 150.0%

1.0% 100.0% 0.8%

50.0% 0.6%

0.4% 0.0% 2015 2016 2017 2018 1H19 0.2% (50.0%)

0.0% 2015 2016 2017 2018 1H19 (100.0%)

Sunny AAC O-film Sunny AAC O-film

Source: Company data Source: Company data M&A was another method of expanding business footprint: O-film acquired Sony’s largest factories of CCM in 2018, and entered the Apple supply chain for handset module. O-film also acquired Fujifilm’s Tianjin factory which are mainly vehicle lens production and consists of certain lens patents. Sunny also supplied to Samsung, including for flagship models such as Galaxy S9, and Sunny entered Apple supply chain in 2018. Leica handset lens was manufactured by Sunny Optical. Overall among Chinese lens players, Sunny is the global leader with increasing market share, while, and will have GM upside in 2020. AAC’s optical business is still small-scale (4% in 2018 and 6% in 1H19); while O-film is at a distant lagging position, with <1% global market share, and currently manages to self-supply for low-end smartphones camera modules. Government support for autonomous driving: In China, the government also declared that it would speed up the development and promulgation of the relevant laws and regulations in respect of autonomous driving. Driven by policy, cross-border competition, consumers’ needs and other factors, the penetration rate of ADAS increased rapidly. In addition, with the further development of the driverless vehicle market, the application demands for LIDAR, smart headlights, night vision cameras and other related products will be increased continuously. Sunny, as a leading vehicle lens supplier (~30% market share), continued to leverage its pioneer advantages in the vehicle field to further enhance its technical capabilities, explore application demands for its products and make micro- innovations in its design, material selection, processing technologies, testing and management processes, so as to further improve its market share and consolidate its No. 1 position in the global market of vehicle lens. Preferential tax treatment for tech companies: Most Chinese technology companies qualified as National High-Tech firms, will receive a preferred taxation treatment (rather than the usual 25% corporate income tax rate). For Sunny Optical, its effective corporate income tax rate was in the range of 11-12%.

Asia Technology Strategy 105

4 November 2019

View on near-term and medium-to-long-term development As the overall reliance on government is small, and leaders such as Sunny can become No. 2 in terms of market share in the smartphone lens market, with continuous R&D investment. The biggest competitors include Largan, Genius, which entered Apple supply chain (Sunny not yet). We think throughout development history, Chinese players have been gradually picking up the capabilities of lens and investing in R&D for high-end models for it. The probability is high to realise domestic substitution in the near term (one to three years), based on the following reasons: (1) technology wise, Sunny is already the market leader, mastering the supplies for high-end lens suppliers to major Chinese smartphone OEMs; (2) Chinese players are gaining shares with shipment penetrating into Chinese and non-Chinese OEMs; (3) continued investment and R&D in the new business areas and technology advancement; (4) scale and ASP advantage which helps Chinese players further expand market share and client base.

MLCC Current market structure The global MLCC (Multi-Layer Ceramic Capacitor) market size is expected to reach ~US$13 bn in FY20E. Considering the majority of MLCC usage is still highly linked to IT products (i.e., handsets, PC, AV and appliances), China consumes the largest part of global MLCC demand, in our view (more than 50%).

Figure 17: Global MLCC market trend

Source: TTI, Company Data, Credit Suisse Research In particular, MLCC demand in China has rapidly grown up in the past five years in tandem with rising smartphone penetration in the region. In addition to smartphone unit shipment growth in China, MLCC contents value per unit has also contributed to incremental MLCC value creation.

Asia Technology Strategy 106

4 November 2019

Figure 18: Smartphone shipment in China vs. as % of global Figure 19: MLCC application-mix in China (as of FY19E)

40% 500 Military, 6% 35% Others, 400 30% 9% 25% 300 Automotive, 20% 10% 15% 200 10% Industrials, 100 10% Consumer 5% Electronics 0% 0 (inc. Handset), 2010 2011 2012 2013 2014 2015 2016 2017 2018 65% China smartphone shipment (mn, RHS) China smartphone as % of global

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates China MLCC demand growth is expected to accelerate from FY20E once 5G roll-out starts to positively impact smartphone replacement rates. We assume 83 mn/167 mn 5G smartphone shipment in China in FY20E/FY21E, which occupies 47%/40% of global 5G demand. Furthermore, 5G smartphones or relevant networks investment (i.e., base station) will also likely to lead to incremental MLCC contents growth per unit. For instance, a Taiwan MLCC maker, Yageo anticipates an increase in 5G smartphone-related MLCC demand of 10–20% for sub-6 GHz and 20-30% for millimetre wave networks, while Korean makers (i.e. SEMCO) project an increase of around 200 units of MLCCs per base station for sub-6 GHz services. Demand growth from sub-6 GHz base stations is expected to be chiefly RF-related, while millimetre wave demand growth is likely to be driven by PMIC peripherals as well as RF. Despite the fact that China is one of the largest MLCC consuming market, the self-sufficient rates appears relatively low. As of current, MLCC supply is majorly dominated by Japan (Murata/TDK/Taiyo Yuden/Kyocera), Korea (SEMCO/Samwha Capacitor) or Taiwan manufacturers (Yageo/Walsin Tech). A growing number of Chinese players are comprised of Fenghua Advanced Technology (000636.SZ), Tianli Holdings Group Limited (0117.HK), Chaozhou Three-circle (300408.SZ), and Fujian Torch Electron Technology (603678.CN). Chinese MLCC manufacturers are rapidly growing in recent years but we would believe global market shares are still only ~5% in aggregate. Also, a local Chinese MLCC maker is highly likely to address MLCC demand in consumer electronics, mainly for the mid-to-low-end segment (i.e., larger size/low capacitance) where leading JP or KR MLCC makers are withdrawing for production optimisation (i.e., A/V, PC, appliances or low-end handsets).

Figure 20: MLCC global market shares (1H19E based on sales) Figure 21: MLCC global capacity trend (Top 5)

(bn umits/month) 400.0 25% 7% 4% 350.0 20% 8% 300.0

Murata 250.0 15% SEMCO 44% 200.0 Taiyo Yuden 10% 14% TDK 150.0 Yageo 100.0 Others 5% 50.0

0.0 0% 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 20182019E 23% Murata Mfg Taiyo Yuden TDK SEMCO Yageo yoy % chg.

Source: Company data, Credit Suisse estimates Source: Company data, Credit Suisse estimates

Asia Technology Strategy 107

4 November 2019

Chinese game plan and view on near-term and medium-to- long-term development In the MLCC arena, while Japanese-owned MLCC makers and SEMCO are working to lock in technology by manufacturing materials and equipment in-house, Taiwanese and Chinese-owned makers are dependent on Japanese materials and equipment suppliers for the majority of their raw materials and production equipment. This makes it highly unlikely that Chinese MLCC makers will close the technology gap with the Japanese and Korean frontrunners anytime soon. Our focus in the near term is on how they can shrink the gap between themselves and Taiwanese makers. PCB/substrate

China has larger share in conventional PCB, multi-layer and HDI: While China controls half of the global PCB production value, Chinese PCB players have relatively stronger presence in commodity, multi-layer, and HDI (high-density interconnection) segments. In 2018/2019, Chinese players oversaw strong growth in the multilayer PCB segment, thanks to strong network build out (mainly from 4G LTE). Among the players, Shennan and WUS PCB are the major beneficiaries of the booming growth in multilayer PCB, both having exhibited strong double-digit sales growth in 1H19 sales, with WUS PCB noting growth in communication was highest at 37% YoY. Chinese PCB players, however, are generally weaker in FPC (Flexible PCB) and substrates, with less than 50% market share in these two areas. Chinese players have been aggressively building related technology through acquisition. DongShan Precision (DSBJ) acquired US top FPC supplier, MFLEX, in 2016, and has then grown its FPC sales from less than 20% of Avary’s revenue to 40% of Avary’s revenue in 2018. We see high growth in the substrate market, with double-digit sales growth seen through 2018/2019 among leading suppliers including Shennan and Fastprint. We also see growing consolidation in the substrate market (with Suntak acquiring PRV China) accelerating technology advancement. Nevertheless, China's largest domestic substrate supplier, Shennan Circuit, held only 1.7% global market share in IC substrates in 2017, vs global Top 10 suppliers' aggregate market share of 80-85% for the same period. Shennan’s IC substrate business is mainly on smartphone application (MEMS Mic and RF), but Shennan and Fastprint (second largest) both substrate business have exposure in memory. Chinese customers (networking customer) have been asking Taiwan substrate suppliers to add production in China, but given rising tension with data security, especially on a national level, it is unlikely that significant amount of capacity would be added in the medium term. Government support has been a big help: China State Council stated in its 13th Five- Year National Plan that it included PCB as one of the key technologies in strengthening its stance in the IT industry in 2016, and China National Development and Reform Commission incorporated HDI, FPC and specialty PCB board as the main focus in the electronic products industry, and encouraged foreign investments in HDI, multilayer PCB, FPC and IC substrate. China PCB suppliers have therefore enjoyed faster earnings growth, compared to Taiwan peers, benefiting from government help, with subsidy to ease their cost burden, as well as lower tax rate. We note that the top Chinese players, including DSBJ, Shennan, WUS PCB, and Kinwong, received subsidy that aggregately adds up to 14% of their pretax profit.

Asia Technology Strategy 108

4 November 2019

Figure 167: China PCB by technology Figure 168: China PCB global share by technology

China PCB China PCB global market share by technology

70%

60% FPC, 17% Commodity China WW market share , 20% Substrates, 50% 3% 40%

HDI, 17% 30%

20% Multi-layer, 44% 10%

0% Commodity Multi-layer HDI Substrates FPC

Source: Company data Source: Company data

Asia Technology Strategy 109

4 November 2019

4November 2019 4November 2019 4November 2019 4November 2019

Companies Mentioned (Price as of 01-Nov-2019) AAC Technologies Holdings Inc (2018.HK, HK$51.5) ACM Rsrch (ACMR.OQ, $12.59) AMEC (688012.SS, Rmb70.09) ASE Industrial Holdings (3711.TW, NT$80.0) ASM Pacific Technology Ltd (0522.HK, HK$103.0)

ASML Holding N.V. (ASML.AS, €239.75)

AU Optronics (2409.TW, NT$7.72) AVX (AVX.N, $15.32) Acer Group (2353.TW, NT$17.95) , Inc. (AMD.OQ, $33.93) Advanced Wireless Semiconductor Company (8086.TWO, NT$114.5) Advantest (6857.T, ¥5,130) Alibaba (BABA.K, $176.67) All Winner Tech (300458.SZ, Rmb24.77) Alps Alpine (6770.T, ¥2,334) Amkor Technology Inc. (AMKR.OQ, $12.43) Amphenol Corporation (APH.N, $100.33) Analog Devices Inc. (ADI.OQ, $106.63) Anji (688019.SS, Rmb112.63) Anritsu (6754.T, ¥2,110) Apple Inc (AAPL.OQ, $248.76) Applied Materials Inc. (AMAT.OQ, $54.26) Artivision (ARTT.SI, S$0.004) Asustek (2357.TW, NT$208.5) Auden (3138.TWO, NT$42.53) Avary Holding (002938.SZ, Rmb45.74) BYD Electronic (International) Company Limited (0285.HK, HK$13.4) BenQ Materials (8215.TW, NT$18.55) Bluecom (033560.KQ, W3,040) Bosch (BOSH.NS, Rs15327.05) CCTC (300408.SZ, Rmb18.57) CGT Group (000066.SZ, Rmb15.08) CMMT Co (4960.TW, NT$7.6) Cadence Design System (CDNS.OQ, $65.35) CammSys (050110.KQ, W2,980) Canon (7751.T, ¥2,961) Career Tech. (6153.TW, NT$38.1) Casetek Holdings Limited (5264.TW, NT$55.7) Catcher Technology (2474.TW, NT$265.0) Changelight (300102.SZ, Rmb4.38) ChipMOS Technologies Inc. (8150.TW, NT$30.55) Chipbond (6147.TWO, NT$61.1) Chroma (2360.TW, NT$143.0) Chunghwa Precision (6510.TWO, NT$960.0) Cirrus Logic (CRUS.OQ, $67.96) (CSCO.OQ, $47.51) (2324.TW, NT$18.2) Coolpad Group Limited (2369.HK, HK$0.231) Corning, Inc. (GLW.N, $29.63) Cree (CREE.OQ, $47.73) Crucialtec (114120.KQ, W959) DISCO (6146.T, ¥23,650) DNF (092070.KQ, W9,140) Daeduk Elec (008060.KS, W10,600) Dai-ichi Seiko (6640.T, ¥2,626) Datang Telecom (600198.SS, Rmb9.09) Delloyd Ventures (DELL.KL^G15) (2308.TW, NT$133.5) Dongjin Semichem (005290.KQ, W16,500) DreamTech (192650.KS, W6,580) Duk San Neolux (213420.KQ, W22,050) EM-Tech (091120.KQ, W8,310) EPI (Holdings) (0689.HK, HK$0.1) EWPT (300115.SZ, Rmb15.1)

Egis Technology Inc. (6462.TWO, NT$263.5)

Elan Microelectronics Corp (2458.TW, NT$93.4) Elcomtec (037950.KQ, W1,490) Epistar Corporation (2448.TW, NT$30.0) Espressif (688018.SS, Rmb133.24) Eternal (1717.TW, NT$25.85) Eugene Tech (084370.KQ, W15,250) Extreme Networks (EXTR.OQ, $6.44) F-GIS (6456.TW, NT$117.5) FIT Hon Teng Limited (6088.HK, HK$3.18) FST (3532.TW, NT$118.0) Fenghua (000636.SZ, Rmb12.28) Ferrotec (6890.T, ¥1,074) Fitipower (4961.TWO, NT$40.51) Flexium (6269.TW, NT$113.5) FocalTech Corporation, Ltd. (3545.TW, NT$24.1) Foxconn Industrial Internet (601138.SS, Rmb15.59) Fudan Microelect (1385.HK, HK$5.91) Fujikura (5803.T, ¥473) Fujimi (5384.T, ¥2,818) Fujitsu (6702.T, ¥9,518) Fullhan (300613.SZ, Rmb165.22) GSEO (3406.TW, NT$447.0) Asia Technology Strategy 110

4 November 2019

4November 2019 4November 2019 4No 4November 2019

GeoVision (3356.TW, NT$34.0) vember2019 Gigadevice (603986.SS, Rmb159.84) Globalwafers (6488.TWO, NT$370.0) GoerTek Inc. (002241.SZ, Rmb18.63) Goke (300672.SZ, Rmb41.4) Griam (600206.SS, Rmb10.26) HC SemiTek (300323.SZ, Rmb4.81)

HP Inc. (HPQ.N, $17.37)

Haier Electronics Grp (1169.HK, HK$23.35) Haier Smart Home (600690.SS, Rmb17.36) Han's Laser Technology Co., Ltd (002008.SZ, Rmb37.89) Hangzhou Hikvision Digital Technology Co., Ltd. (002415.SZ, Rmb32.9) Hannstar Display (6116.TW, NT$6.5) Hewlett Packard Enterprise (HPE.N, $16.41) Himax Technologies, Inc. (HIMX.OQ, $2.36) Hirata (6258.T, ¥7,620) Hirose Electric (6806.T, ¥13,730) Hitachi High-Tec (8036.T, ¥6,750) Holitech (002217.SZ, Rmb5.32) Hon Hai Precision (2317.TW, NT$84.8) Hongda Elec (300726.SZ, Rmb25.47) Hongyuan Elec (603267.SS, Rmb47.6) Hua Hong Semiconductor Limited (1347.HK, HK$15.62) Huada Automotive (603358.SS, Rmb13.01) Huayi Elec (600290.SS, Rmb3.89) Huicheng (300481.SZ, Rmb14.92) IBIDEN (4062.T, ¥2,496) II-VI (IIVI.OQ, $33.15) Idemitsu Kosan (5019.T, ¥3,170) Inari Amertron (INAR.KL, RM1.99) Infineon Technologies AG (IFXGn.DE, €17.796) Ingenic (300223.SZ, Rmb53.85) Innolux Corporation (3481.TW, NT$6.68) Intel Corp. (INTC.OQ, $56.53) Interflex (051370.KQ, W17,700) International Business Machines (IBM.N, $133.73) Inventec Co Ltd (2356.TW, NT$22.05) Irico Display (600707.SS, Rmb4.32) Iriso Electronic (6908.T, ¥5,320) JCET (600584.SS, Rmb18.75) JDS Uniphase Corp (VIAV.OQ, $15.96) JSR (4185.T, ¥2,044) Jabil Circuit Inc. (JBL.N, $36.82) Jahwa Electronic (033240.KS, W10,550) Japan Aviation (6807.T, ¥2,080) Japan Display (6740.T, ¥69) Jingjia Micro (300474.SZ, Rmb54.33) KCTech (281820.KS, W19,650) KDX (002450.SZ, Rmb3.52) KEMET (KEM.N, $21.74) KFMI (300666.SZ, Rmb39.7) KLA-Tencor Corp. (KLAC.OQ, $169.04) KYEC (2449.TW, NT$37.8) Kaifa (000021.SZ, Rmb10.24) Kingpak Technology (6238.TWO, NT$149.0) Knowles (KN.N, $21.58) Kolen (078650.KQ, W1,750) Korea Circuit (007810.KS, W10,100) Kyocera (6971.T, ¥6,945) LB Semicon (061970.KQ, W7,770) LCT Holdings (LCTH.SI, S$0.5) LG Chem Ltd. (051910.KS, W309,000) LG Display Co Ltd. (034220.KS, W13,300) LG Electronics Inc (066570.KS, W67,200) LG Innotek (011070.KS, W121,500) Lam Research Corp. (LRCX.OQ, $271.04)

Largan Precision (3008.TW, NT$4460.0)

Lattice Us (LSCC.OQ, $19.59) Lenovo Group Ltd (0992.HK, HK$5.67) Lens Technology Co., Ltd (300433.SZ, Rmb13.6) Liande (300545.SZ, Rmb24.16) Lite-On Technology (2301.TW, NT$49.95) Lumentum Hldg (LITE.OQ, $62.66) Luxshare Precision Industry Co., Ltd (002475.SZ, Rmb32.81) MCNEX (097520.KQ, W23,800) Macronix (2337.TW, NT$32.0) Maxim Integrated Products (MXIM.OQ, $58.66) Maxis Berhad (MXSC.KL, RM5.38) MediaTek Inc. (2454.TW, NT$405.0) Melexis (MLXS.BR, €62.3) Merck & Co., Inc. (MRK.N, $86.66) Merry Electronics Co. Ltd (2439.TW, NT$150.0) Micro-Star International Co., Ltd (2377.TW, NT$90.4) Microchip Technology Inc. (MCHP.OQ, $94.29) Micron Technology Inc. (MU.OQ, $47.55) Microsoft (MSFT.OQ, $143.37) MinebeaMitsumi (6479.T, ¥2,078) Mitsubishi Corp (8058.T, ¥2,732)

Asia Technology Strategy 111

4 November 2019

4November 2019 4November 2019 4November 2019 4November 2019

Montage Tech (688008.SS, Rmb59.25) Murata Manufacturing (6981.T, ¥6,053) NAURA (002371.SZ, Rmb70.95) NOK Corp (7240.T, ¥1,663) NVIDIA Corporation (NVDA.OQ, $201.02) NXP Semiconductors N.V. (NXPI.OQ, $113.68) Nan Ya Printed Circuit Board (8046.TW, NT$54.2)

Nanya Technology (2408.TW, NT$70.6)

Nations Tech (300077.SZ, Rmb6.83) Nichias (5393.T, ¥2,285) Nikon (7731.T, ¥1,377) Nippon Steel & Sumitomo Metal (5401.T, ¥1,592) Nitto Denko (6988.T, ¥5,930) Novatek Microelectronics Corp Ltd (3034.TW, NT$202.0) ON Semiconductor Corp. (ON.OQ, $20.4) On-Bright Elec (4947.TWO, NT$181.0) Orbita (300053.SZ, Rmb10.16) Osram (OSRn.DE, €40.1) PSK Hldg (031980.KQ, W8,150) PXPL (087600.KQ, W4,715) Panasonic (6752.T, ¥981) Pegatron (4938.TW, NT$60.2) Photronics (PLAB.OQ, $11.8) Power Logics (047310.KQ, W9,800) PowerLeader (8236.HK, HK$1.9) Powertech Technology (6239.TW, NT$95.0) Q Technology Group Co Ltd (1478.HK, HK$10.52) QUALCOMM Inc. (QCOM.OQ, $80.44) Qorvo (QRVO.O, $80.86) (2382.TW, NT$59.9) ROHM (6963.T, ¥8,400) RS Technologies (3445.T, ¥4,495) Rambus Incorporated (RMBS.OQ, $13.845) Raydium (3592.TWO, NT$92.29) Raytron Tech (688002.SS, Rmb36.69) Realtek Semiconductor (2379.TW, NT$229.0) Renesas Electronics (6723.T, ¥725) Richwave (4968.TW, NT$234.0) SCREEN (7735.T, ¥7,700) SELEN (002341.SZ, Rmb5.46) SFA (056190.KQ, W42,000) SG Micro (300661.SZ, Rmb211.5) SHIBAURA MECHATR (6590.T, ¥3,550) SK Hynix Inc. (000660.KS, W83,100) SK Materials (036490.KQ, W185,000) SKC Kolon PI (178920.KQ, W34,000) STMicroelectron (STM.MI, €20.75) SUMCO (3436.T, ¥1,818) Samsung Electro-Mechanics (009150.KS, W114,000) Samsung Electronics (005930.KS, W51,200) Samsung SDI (006400.KS, W231,000) Sanan Optoelectronics Co. Ltd (600703.SS, Rmb15.44) Semiconductor Manufacturing International Corp. (0981.HK, HK$10.1) Seoul Semiconductor Co Ltd (046890.KQ, W12,950) Shanghai Sinyang (300236.SZ, Rmb23.53) Sharp (6753.T, ¥1,287) Shennan Circuits (002916.SZ, Rmb148.46) Shenzhen Goodix (603160.SS, Rmb183.0) Shenzhen O-film Tech Co., Ltd (002456.SZ, Rmb11.25) Shenzhen Sunway Communication Co., Ltd (300136.SZ, Rmb41.15) Shin-Etsu Chemical (4063.T, ¥12,030) Silan (600460.SS, Rmb14.79) Silergy (6415.TW, NT$844.0) Silicon Works (108320.KQ, W33,200) Sino Wealth (300327.SZ, Rmb24.55) Sinocera (300285.SZ, Rmb21.0)

Skyworks Solutns (SWKS.O, $91.06)

Solomon Systech (2878.HK, HK$0.16) Sony (6758.T, ¥6,619) Soulbrain (036830.KQ, W77,800) Speed (300322.SZ, Rmb21.22) Sugon (603019.SS, Rmb33.5) Sumitomo Chemical (4005.T, ¥501) Sumitomo Corp (8053.T, ¥1,664) Sun.King (0580.HK, HK$1.07) Sunlord (002138.SZ, Rmb21.18) Sunny Optical Technology Group Co.Limited (2382.HK, HK$129.4) Suzhou Good-Ark (002079.SZ, Rmb7.4) Synaptics (SYNA.OQ, $42.11) Synopsys Inc. (SNPS.OQ, $135.75) Sytech (600183.SS, Rmb23.35) TCL Corporation (000100.SZ, Rmb3.39) TDK (6762.T, ¥10,700) TE Connectivity (TEL.N, $89.5) TJ Semi (002129.SZ, Rmb11.61) TTM Technologies (TTMI.O, $11.71) TXC Corp. (3042.TW, NT$37.85) Taiflex (8039.TW, NT$44.2)

Asia Technology Strategy 112

4 November 2019

4November 2019 4November 2019 4November 2019 4November 2019

Taimide Tech (3645.TW, NT$51.3) Taiwan Semiconductor Manufacturing (2330.TW, NT$299.0) Taiyo Yuden (6976.T, ¥2,861) Techwing (089030.KQ, W10,750) Tes Co Ltd (095610.KQ, W20,450) Texas Instruments Inc. (TXN.OQ, $117.99) ThinFlex (3144.TWO, NT$30.75)

Tianli Holdings (0117.HK, HK$0.48)

Tianma Microelectronics Co. Ltd (000050.SZ, Rmb13.86) Tianshui Huatian Technology Co., Ltd (002185.SZ, Rmb5.5) Tokyo Electron (8035.T, ¥22,245) Tokyo Ohka Kogyo (4186.T, ¥4,240) TongFu Microelectronics Co.,Ltd. (002156.SZ, Rmb11.63) Torch Electron (603678.SS, Rmb22.18) Toshiba (6502.T, ¥3,640) Toyoda Gosei (7282.T, ¥2,500) Truly International (0732.HK, HK$1.02) Tunghsu Optoelec (000413.SZ, Rmb4.83) Ube Industries (4208.T, ¥2,348) Ulvac (6728.T, ¥4,780) UniStrong (002383.SZ, Rmb10.71) UniTest (086390.KQ, W14,700) Unimicron Technology Corp (3037.TW, NT$47.2) United Microelectronics (2303.TW, NT$14.5) Universal Scientific Industrial (Shanghai) (601231.SS, Rmb14.74) UniversalDisplay (OLED.OQ, $200.18) Valiant (002643.SZ, Rmb12.38) Vanguard International Semiconductor (5347.TWO, NT$65.2) Viatron Technolo (141000.KQ, W9,350) Victory Giant (300476.SZ, Rmb15.32) Vishay Intertech (VSH.N, $20.15) Visionox (002387.SZ, Rmb15.22) Visual Photonics Epitaxy Co., Ltd (2455.TW, NT$123.0) WITS (4953.TWO, NT$111.5) WLCSP (603005.SS, Rmb22.58) WTC (2492.TW, NT$181.5) WUS (002463.SZ, Rmb21.85) Wafer Works Corp (6182.TWO, NT$35.7) Western Digital (WDC.OQ, $51.65) WiSoL (122990.KQ, W15,000) Will Semi (603501.SS, Rmb107.28) Win Semiconductors Corp (3105.TWO, NT$324.5) Winbond (2344.TW, NT$17.05) Wonik IPS (240810.KQ, W32,550) Wonik Materials (104830.KQ, W27,900) XC TECH (300139.SZ, Rmb8.75) Xiaomi Corporation (1810.HK, HK$8.96) Xilinx (XLNX.OQ, $90.74) Yageo (2327.TW, NT$316.0) Yangjie Elctrnc (300373.SZ, Rmb15.49) Yoke Technology (002409.SZ, Rmb18.42) ZHIYUNAUTOMATION (300097.SZ, Rmb8.5) ZJ Ruiyuan Tech (8249.HK, HK$0.169) ZTE Corporation (0763.HK, HK$22.7) ZTE Corporation (000063.SZ, Rmb34.0) Co., Ltd (002236.SZ, Rmb16.38)

Disclosure Appendix Analyst Certification Manish Nigam and Clive Cheung each certify, with respect to the companies or securities that the individual analyzes, that (1) the views expressed in this report accurately reflect his or her personal views about all of the subject companies and securities and (2) no part of his or her

compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed in this report.

As of December 10, 2012 Analysts’ stock rating are defined as follows: Outperform (O) : The stock’s total return is expected to outperform the relevant benchmark* over the next 12 months. Neutral (N) : The stock’s total return is expected to be in line with the relevant benchmark* over the next 12 months. Underperform (U) : The stock’s total return is expected to underperform the relevant benchmark* over the next 12 months. *Relevant benchmark by region: As of 10th December 2012, Japanese ratings are based on a stock’s total return relative to the analyst's coverage universe which consists of all companies covered by the analyst within the relevant sector, with Outperforms representing the most attractive, Neutrals the less attractive, and Underperforms the least attractive investment opportunities. As of 2nd October 2012, U.S. and Canadian as well as Europea n ratings are based on a stock’s total return relative to the analyst's coverage universe which consists of all companies covered by the analyst within the re levant sector, with Outperforms representing the most attractive, Neutrals the less attractive, and Underperforms the least attractive investment opportunities. For Latin American and Asia stocks (excluding Japan and Australia), ratings are based on a stock’s total return relative to the average total return of the rele vant country or regional benchmark (India - S&P BSE Sensex Index); prior to 2nd October 2012 U.S. and Canadian ratings were based on (1) a stock’s absolute total return pote ntial to its current share price and (2) the relative attractiveness of a stock’s total return potential within an analyst’s coverage universe. For Australian and New Zealand stocks, the expected total return (ETR) calculation includes 12-month rolling dividend yield. An Outperform rating is assigned where an ETR is greater than or equal to 7.5%; Underperform where an ETR less than or equal to 5%. A Neutral may be assigned where the ETR is between -5% and 15%. The overlapping rating range allows analysts to assign a rating that puts ETR in the context of associated risks. Prior to 18 May 2015, ETR ranges for Outperform and Underperform ratings did not overlap with Neutral thresholds between 15% and 7.5%, which was in operation from 7 July 2011.

Asia Technology Strategy 113

4 November 2019

4November 2019 4November 2019 4November 2019 4November 2019 Restricted (R) : In certain circumstances, Credit Suisse policy and/or applicable law and regulations preclude certain types of communications, including an investment recommendation, during the course of Credit Suisse's engagement in an investment banking transaction and in certain other circumstances. Not Rated (NR) : Credit Suisse Equity Research does not have an investment rating or view on the stock or any other securities related to the

company at this time.

Not Covered (NC) : Credit Suisse Equity Research does not provide ongoing coverage of the company or offer an investment rating or investment view on the equity security of the company or related products. Volatility Indicator [V] : A stock is defined as volatile if the stock price has moved up or down by 20% or more in a month in at least 8 of the past 24 months or the analyst expects significant volatility going forward. Analysts’ sector weightings are distinct from analysts’ stock ratings and are based on the analyst’s expectations for the fundamentals and/or valuation of the sector* relative to the group’s historic fundamentals and/or valuation: Overweight : The analyst’s expectation for the sector’s fundamentals and/or valuation is favorable over the next 12 months. Market Weight : The analyst’s expectation for the sector’s fundamentals and/or valuation is neutral over the next 12 months. Underweight : The analyst’s expectation for the sector’s fundamentals and/or valuation is cautious over the next 12 months. *An analyst’s coverage sector consists of all companies covered by the analyst within the relevant sector. An analyst may cov er multiple sectors. Credit Suisse's distribution of stock ratings (and banking clients) is:

Global Ratings Distribution Rating Versus universe (%) Of which banking clients (%) Outperform/Buy* 47% (32% banking clients) Neutral/Hold* 38% (26% banking clients) Underperform/Sell* 13% (23% banking clients) Restricted 2% *For purposes of the NYSE and FINRA ratings distribution disclosure requirements, our stock ratings of Outperform, Neutral, a nd Underperform most closely correspond to Buy, Hold, and Sell, respectively; however, the meanings are not the same, as our stock ratings are determined on a relative basis. (Please refer to definitions above.) An investor's decision to buy or sell a security should be based on investment objectives, current holdin gs, and other individual factors. Important Global Disclosures Credit Suisse’s research reports are made available to clients through our proprietary research portal on CS PLUS. Credit Suisse research products may also be made available through third-party vendors or alternate electronic means as a convenience. Certain research products are only made available through CS PLUS. The services provided by Credit Suisse’s analysts to clients may depend on a specific client’s preferences regarding the frequency and manner of receiving communications, the client’s risk profile and investment, the size and scope of the overall client relationship with the Firm, as well as legal and regulatory constraints. To access all of Credit Suisse’s research that you are entitled to receive in the most timely manner, please contact your sales representative or go to https://plus.credit-suisse.com . Credit Suisse’s policy is to update research reports as it deems appropriate, based on developments with the subject company, the sector or the market that may have a material impact on the research views or opinions stated herein. Credit Suisse's policy is only to publish investment research that is impartial, independent, clear, fair and not misleading. For more detail please refer to Credit Suisse's Policies for Managing Conflicts of Interest in connection with Investment Research: https://www.credit- suisse.com/sites/disclaimers-ib/en/managing-conflicts.html . Any information relating to the tax status of financial instruments discussed herein is not intended to provide tax advice or to be used by anyone to provide tax advice. Investors are urged to seek tax advice based on their particular circumstances from an independent tax professional. Credit Suisse has decided not to enter into business relationships with companies that Credit Suisse has determined to be involved in the development, manufacture, or acquisition of anti-personnel mines and cluster munitions. For Credit Suisse's position on the issue, please see https://www.credit-suisse.com/media/assets/corporate/docs/about-us/responsibility/banking/policy-summaries-en.pdf . The analyst(s) responsible for preparing this research report received compensation that is based upon various factors including Credit Suisse's total revenues, a portion of which are generated by Credit Suisse's investment banking activities

For date and time of production, dissemination and history of recommendation for the subject company(ies) featured in this report, disseminated

within the past 12 months, please refer to the link: https://rave.credit- suisse.com/disclosures/view/report?i=469018&v=40xavti50t9h803dy6d6qdj9h . Important Regional Disclosures Singapore recipients should contact Credit Suisse AG, Singapore Branch for any matters arising from this research report. The analyst(s) involved in the preparation of this report may participate in events hosted by the subject company, including site visits. Credit Suisse does not accept or permit analysts to accept payment or reimbursement for travel expenses associated with these events. For Credit Suisse Securities (Canada), Inc.'s policies and procedures regarding the dissemination of equity research, please visit https://www.credit-suisse.com/sites/disclaimers-ib/en/canada-research-policy.html. Investors should note that income from such securities and other financial instruments, if any, may fluctuate and that price or value of such securities and instruments may rise or fall and, in some cases, investors may lose their entire principal investment. Taiwanese Disclosures: This research report is for reference only. Investors should carefully consider their own investment risk and note they may be subject to the applicable rules and regulations in Taiwan. Investment results are the responsibility of the investor. Reports written by Taiwan based analysts on non-Taiwan listed companies are not considered as recommendations to buy or sell securities. Reports may not be reproduced without the permission of Credit Suisse. Pursuant to the ‘Taiwan Stock

Asia Technology Strategy 114

4 November 2019

4November 2019 4November 2019 4November 2019 4November 2019 Exchange Regulations Governing Securities Firms Recommending Trades in Securities to Customers’ and the ‘ Rules Governing Securities Firms Recommending Trades in Securities to Customers’, in order for a non-client of Credit Suisse AG, Taipei Securities Branch to receive this research report, no provision by such non-client of the content of the report to a third party, nor any conflict of interest, is permitted. By receiving this research report, any such non-client is deemed to acknowledge

and accept our terms and disclaimers included herein.

This research report is authored by: Credit Suisse () Limited ...... Manish Nigam ; Clive Cheung ; Kyna Wong ; Chaolien Tseng Credit Suisse Securities (Japan) Limited ...... Hideyuki Maekawa ; Yoshiyasu Takemura ; Mika Nishimura ; Akinori Kanemoto Credit Suisse Securities (Europe) Limited, Seoul Branch ...... Keon Han ; Sang Uk Kim Credit Suisse AG, Taipei Securities Branch ...... Randy Abrams, CFA ; Jerry Su ; Haas Liu ; Pauline Chen Credit Suisse Securities (USA) LLC ...... Matthew Cabral ; Sami Badri To the extent this is a report authored in whole or in part by a non-U.S. analyst and is made available in the U.S., the following are important disclosures regarding any non-U.S. analyst contributors: The non-U.S. research analysts listed below (if any) are not registered/qualified as research analysts with FINRA. The non-U.S. research analysts listed below may not be associated persons of CSSU and therefore may not be subject to the FINRA 2241 and NYSE Rule 472 restrictions on communications with a subject company, public appearances and trading securities held by a research analyst account. Credit Suisse (Hong Kong) Limited ...... Manish Nigam ; Clive Cheung ; Kyna Wong ; Chaolien Tseng Credit Suisse Securities (Japan) Limited ...... Hideyuki Maekawa ; Yoshiyasu Takemura ; Mika Nishimura ; Akinori Kanemoto Credit Suisse Securities (Europe) Limited, Seoul Branch ...... Keon Han ; Sang Uk Kim Credit Suisse AG, Taipei Securities Branch ...... Randy Abrams, CFA ; Jerry Su ; Haas Liu ; Pauline Chen Important disclosures regarding companies that are the subject of this report are available by calling +1 (877) 291-2683. The same important disclosures, with the exception of valuation methodology and risk discussions, are also available on Credit Suisse’s disclosure website at https://rave.credit-suisse.com/disclosures . For valuation methodology and risks associated with any recommendation, price target, or rating referenced in this report, please refer to the disclosures section of the most recent report regarding the subject company.

Asia Technology Strategy 115

4 November 2019

This report is produced by subsidiaries and affiliates of Credit Suisse operating under its Global Markets Division. For more information on our structure, please use the following link: https://www.credit-suisse.com/who-we-are This report may contain material that is not directed to, or intended for distribution to or use by, any person or entity who is a citizen or resident of or located in any locality, state, country or other jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation or which would subject Credit Suisse or its affiliates ("CS") to any registration or licensing requirement within such jurisdiction. All material presented in this report, unless specifically indicated otherwise, is under copyright to CS. None of the material, nor its content, nor any copy of it, may be altered in any way, transmitted to, copied or distributed to any other party, without the prior express written permission of CS. All trademarks, service marks and logos used in this report are trademarks or service marks or registered trademarks or service marks of CS or its affiliates.The information, tools and material presented in this report are provided to you for information purposes only and are not to be used or considered as an offer or the solicitation of an offer to sell or to buy or subscribe for securities or other financial instruments. CS may not have taken any steps to ensure that the securities referred to in this report are suitable for any particular investor. CS will not treat recipients of this report as its customers by virtue of their receiving this report. The investments and services contained or referred to in this report may not be suitable for you and it is recommended that you consult an independent investment advisor if you are in doubt about such investments or investment services. Nothing in this report constitutes investment, legal, accounting or tax advice, or a representation that any investment or strategy is suitable or appropriate to your individual circumstances, or otherwise constitutes a personal recommendation to you. Please note in particular that the bases and levels of taxation may change. Information and opinions presented in this report have been obtained or derived from sources believed by CS to be reliable, but CS makes no representation as to their accuracy or completeness. CS accepts no liability for loss arising from the use of the material presented in this report, except that this exclusion of liability does not apply to the extent that such liability arises under specific statutes or regulations applicable to CS. This report is not to be relied upon in substitution for the exercise of independent judgment. CS may have issued, and may in the future issue, other communications that are inconsistent with, and reach different conclusions from, the information presented in this report. Those communications reflect the different assumptions, views and analytical methods of the analysts who prepared them and CS is under no obligation to ensure that such other communications are brought to the attention of any recipient of this report. Some investments referred to in this report will be offered solely by a single entity and in the case of some investments solely by CS, or an associate of CS or CS may be the only market maker in such investments. Past performance should not be taken as an indication or guarantee of future performance, and no representation or warranty, express or implied, is made regarding future performance. Information, opinions and estimates contained in this report reflect a judgment at its original date of publication by CS and are subject to change without notice. The price, value of and income from any of the securities or financial instruments mentioned in this report can fall as well as rise. The value of securities and financial instruments is subject to exchange rate fluctuation that may have a positive or adverse effect on the price or income of such securities or financial instruments. Investors in securities such as ADR's, the values of which are influenced by currency volatility, effectively assume this risk. Structured securities are complex instruments, typically involve a high degree of risk and are intended for sale only to sophisticated investors who are capable of understanding and assuming the risks involved. The market value of any structured security may be affected by changes in economic, financial and political factors (including, but not limited to, spot and forward interest and exchange rates), time to maturity, market conditions and volatility, and the credit quality of any issuer or reference issuer. Any investor interested in purchasing a structured product should conduct their own investigation and analysis of the product and consult with their own professional advisers as to the risks involved in making such a purchase. Some investments discussed in this report may have a high level of volatility. High volatility investments may experience sudden and large falls in their value causing losses when that investment is realised. Those losses may equal your original investment. Indeed, in the case of some investments the potential losses may exceed the amount of initial investment and, in such circumstances, you may be required to pay more money to support those losses. Income yields from investments may fluctuate and, in consequence, initial capital paid to make the investment may be used as part of that income yield. Some investments may not be readily realisable and it may be difficult to sell or realise those investments, similarly it may prove difficult for you to obtain reliable information about the value, or risks, to which such an investment is exposed. This report may provide the addresses of, or contain hyperlinks to, websites. Except to the extent to which the report refers to website material of CS, CS has not reviewed any such site and takes no responsibility for the content contained therein. Such address or hyperlink (including addresses or hyperlinks to CS's own website material) is provided solely for your convenience and information and the content of any such website does not in any way form part of this document. Accessing such website or following such link through this report or CS's website shall be at your own risk. This report is issued and distributed in (except Switzerland): by Credit Suisse Securities (Europe) Limited, One Cabot Square, London E14 4QJ, England, which is authorised by the Prudential Regulation Authority and regulated by the Financial Conduct Authority and the Prudential Regulation Authority. Germany: Credit Suisse (Deutschland) Aktiengesellschaft regulated by the Bundesanstalt fuer Finanzdienstleistungsaufsicht ("BaFin"). United States and Canada: Credit Suisse Securities (USA) LLC; Switzerland: Credit Suisse AG; Brazil: Banco de Investimentos Credit Suisse (Brasil) S.A or its affiliates; Mexico: Banco Credit Suisse (México), S.A., Institución de Banca Múltiple, Grupo Financiero Credit Suisse (México) and Casa de Bolsa Credit Suisse (México), S.A. de C.V., Grupo Financiero Credit Suisse (México) ("Credit Suisse Mexico"). This document has been prepared for information purposes only and is exclusively distributed in Mexico to Institutional Investors. Credit Suisse Mexico is not responsible for any onward distribution of this report to non-institutional investors by any third party. The authors of this report have not received payment or compensation from any entity or company other than from the relevant Credit Suisse Group company employing them; Japan: by Credit Suisse Securities (Japan) Limited, Financial Instruments Firm, Director-General of Kanto Local Finance Bureau ( Kinsho) No. 66, a member of Japan Securities Dealers Association, The Financial Futures Association of Japan, Japan Investment Advisers Association, Type II Financial Instruments Firms Association; Hong Kong: Credit Suisse (Hong Kong) Limited; Australia: Credit Suisse Equities (Australia) Limited; Thailand: Credit Suisse Securities (Thailand) Limited, regulated by the Office of the Securities and Exchange Commission, Thailand, having registered address at 990 Abdulrahim Place, 27th Floor, Unit 2701, Rama IV Road, Silom, Bangrak, Bangkok10500, Thailand, Tel. +66 2614 6000; Malaysia: Credit Suisse Securities (Malaysia) Sdn Bhd; Singapore: Credit Suisse AG, Singapore Branch; India: Credit Suisse Securities (India) Private Limited (CIN no.U67120MH1996PTC104392) regulated by the Securities and Exchange Board of India as Research Analyst (registration no. INH 000001030) and as Stock Broker (registration no. INZ000248233), having registered address at 9th Floor, Ceejay House, Dr.A.B. Road, Worli, Mumbai - 18, India, T- +91-22 6777 3777; South Korea: Credit Suisse Securities (Europe) Limited, Seoul Branch; Taiwan: Credit Suisse AG Taipei Securities Branch; Indonesia: PT Credit Suisse Sekuritas Indonesia; Philippines:Credit Suisse Securities (Philippines ) Inc., and elsewhere in the world by the relevant authorised affiliate of the above. Additional Regional Disclaimers Australia: Credit Suisse Securities (Europe) Limited ("CSSEL") and Credit Suisse International ("CSI") are authorised by the Prudential Regulation Authority and regulated by the Financial Conduct Authority ("FCA") and the Prudential Regulation Authority under UK laws, which differ from Australian Laws. CSSEL and CSI do not hold an Australian Financial Services Licence ("AFSL") and are exempt from the requirement to hold an AFSL under the Corporations Act (Cth) 2001 ("Corporations Act") in respect of the financial services provided to Australian wholesale clients (within the meaning of section 761G of the Corporations Act) (hereinafter referred to as “Financial Services”). This material is not for distribution to retail clients and is directed exclusively at Credit Suisse's professional clients and eligible counterparties as defined by the FCA, and wholesale clients as defined under section 761G of the Corporations Act. Credit Suisse (Hong Kong) Limited ("CSHK") is licensed and regulated by the Securities and Futures Commission of Hong Kong under the laws of Hong Kong, which differ from Australian laws. CSHKL does not hold an AFSL and is exempt from the requirement to hold an AFSL under the Corporations Act in respect of providing Financial Services. Investment banking services in the United States are provided by Credit Suisse Securities (USA) LLC, an affiliate of Credit Suisse Group. CSSU is regulated by the United States Securities and Exchange Commission under United States laws, which differ from Australian laws. CSSU does not hold an AFSL and is exempt from the requirement to hold an AFSL under the Corporations Act in respect of providing Financial Services. Credit Suisse Asset Management LLC (CSAM) is authorised by the Securities and Exchange Commission under US laws, which differ from Australian laws. CSAM does not hold an AFSL and is exempt from the requirement to hold an AFSL under the Corporations Act in respect of providing Financial Services. This material is provided solely to Institutional Accounts (as defined in the FINRA rules) who are Eligible Contract Participants (as defined in the US Commodity Exchange Act). Credit Suisse Equities (Australia) Limited (ABN 35 068 232 708) ("CSEAL") is an AFSL holder in Australia (AFSL 237237). Malaysia: Research provided to residents of Malaysia is authorised by the Head of Research for Credit Suisse Securities (Malaysia) Sdn Bhd, to whom they should direct any queries on +603 2723 2020. Singapore: This report has been prepared and issued for distribution in Singapore to institutional investors, accredited investors and expert investors (each as defined under the Financial Advisers Regulations) only, and is also distributed by Credit Suisse AG, Singapore Branch to overseas investors (as defined under the Financial Advisers Regulations). Credit Suisse AG, Singapore Branch may distribute reports produced by its foreign entities or affiliates pursuant to an arrangement under Regulation 32C of the Financial Advisers Regulations. Singapore recipients should contact Credit Suisse AG, Singapore Branch at +65-6212-2000 for matters arising from, or in connection with, this report. By virtue of your status as an institutional investor, accredited investor, expert investor or overseas investor, Credit Suisse AG, Singapore Branch is exempted from complying with certain compliance requirements under the Financial Advisers Act, Chapter 110 of Singapore (the “FAA”), the Financial Advisers Regulations and the relevant Notices and Guidelines issued thereunder, in respect of any financial advisory service which Credit Suisse AG, Singapore Branch may provide to you. EU: This report has been produced by subsidiaries and affiliates of Credit Suisse operating under its Global Markets Division In jurisdictions where CS is not already registered or licensed to trade in securities, transactions will only be effected in accordance with applicable securities legislation, which will vary from jurisdiction to jurisdiction and may require that the trade be made in accordance with applicable exemptions from registration or licensing requirements. This material is issued and distributed in the U.S. by CSSU, a member of NYSE, FINRA, SIPC and the NFA, and CSSU accepts responsibility for its contents. Clients should contact analysts and execute transactions through a Credit Suisse subsidiary or affiliate in their home jurisdiction unless governing law permits otherwise. Please note that this research was originally prepared and issued by CS for distribution to their market professional and institutional investor customers. Recipients who are not market professional or institutional investor customers of CS should seek the advice of their independent financial advisor prior to taking any investment decision based on this report or for any necessary explanation of its contents. CS may provide various services to US municipal entities or obligated persons ("municipalities"), including suggesting individual transactions or trades and entering into such transactions. Any services CS provides to municipalities are not viewed as "advice" within the meaning of Section 975 of the Dodd-Frank Wall Street Reform and Consumer Protection Act. CS is providing any such services and related information solely on an arm's length basis and not as an advisor or fiduciary to the municipality. In connection with the provision of the any such services, there is no agreement, direct or indirect, between any municipality (including the officials,management, employees or agents thereof) and CS for CS to provide advice to the municipality. Municipalities should consult with their financial, accounting and legal advisors regarding any such services provided by CS. In addition, CS is not acting for direct or indirect compensation to solicit the municipality on behalf of an unaffiliated broker, dealer, municipal securities dealer, municipal advisor, or investment adviser for the purpose of obtaining or retaining an engagement by the municipality for or in connection with Municipal Financial Products, the issuance of municipal securities, or of an investment adviser to provide investment advisory services to or on behalf of the municipality. If this report is being distributed by a financial institution other than Credit Suisse AG, or its affiliates, that financial institution is solely responsible for distribution. Clients of that institution should contact that institution to effect a transaction in the securities mentioned in this report or require further information. This report does not constitute investment advice by Credit Suisse to the clients of the distributing financial institution, and neither Credit Suisse AG, its affiliates, and their respective officers, directors and employees accept any liability whatsoever for any direct or consequential loss arising from their use of this report or its content. No information or communication provided herein or otherwise is intended to be, or should be construed as, a recommendation within the meaning of the US Department of Labor’s final regulation defining "investment advice" for purposes of the Employee Retirement Income Security Act of 1974, as amended and Section 4975 of the Internal Revenue Code of 1986, as amended, and the information provided herein is intended to be general information, and should not be construed as, providing investment advice (impartial or otherwise). Copyright © 2019 CREDIT SUISSE AG and/or its affiliates. All rights reserved. When you purchase non-listed Japanese fixed income securities (Japanese government bonds, Japanese municipal bonds, Japanese government guaranteed bonds, Japanese corporate bonds) from CS as a seller, you will be requested to pay the purchase price only.

Asia Technology Strategy 116