M FOUNDATION March 12, 2019 09:00 PM GMT

China's Industrial Evolution Global Semis – How China Will Chip In

We lay out four scenarios for the future evolution of China's semiconductor industry in this report. We also examine what those scenarios could mean for the global semiconductor supply chain.

Morgan Stanley does and seeks to do business with companies covered in Morgan Stanley Research. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of Morgan Stanley Research. Investors should consider Morgan Stanley Research as only a single factor in making their investment decision. For analyst certification and other important disclosures, refer to the Disclosure Section, located at the end of this report. += Analysts employed by non-U.S. affiliates are not registered with FINRA, may not be associated persons of the member and may not be subject to NASD/NYSE restrictions on communications with a subject company, public appearances and trading securities held by a research analyst account. M FOUNDATION Contributors

MORGAN STANLEY TAIWAN LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY & CO. LLC Charlie Chan Shawn Kim Joseph Moore Equity Analyst Equity Analyst Equity Analyst +8862 2730-1725 +8523963-1005 +1212 761-7516 [email protected] [email protected] [email protected]

MORGAN STANLEY & CO. LLC MORGAN STANLEY MUFG SECURITIES CO., LTD.+ MORGAN STANLEY TAIWAN LIMITED+ Craig Hettenbach Kazuo Yoshikawa, CFA Daniel Yen, CFA Equity Analyst Equity Analyst Equity Analyst +1212 761-6435 +813 6836-8408 +8862 2730-2863 [email protected] [email protected] [email protected]

MORGAN STANLEY TAIWAN LIMITED+ MORGAN STANLEY TAIWAN LIMITED+ Eric Chao Jeff Hsu Research Associate Research Associate +8862 2730-1723 +8862 2730-2864 [email protected] [email protected] M FOUNDATION China's Industrial Evolution Global Semis – How China Will Chip In

We lay out four scenarios for the future evolution of China's semiconductor industry in this report. We also examine what those scenarios could mean for the global semiconductor supply chain.

Industry View We believe the direction of China's semiconductor industry will be shaped by two questions: 1) Will the industry aim to advance Greater China Technology Semiconductors - through localization or globalization? and 2) Will the focus be on Cautious high-value design or low-margin manufacturing? These questions inform the four scenarios we outline in this report. the 16% self-sufficiency ratio in 2018 also suggests China's IC design The best-case scenario for China is 'Industry Upgrade', triggered industry has ample room to grow. by technology breakthroughs in both chip design and manufacturing, Investment opportunities in China semis: In this report, we pro- while Chinese companies collaborate more with the global tech vide a preliminary guide to China's local semiconductor supply chain. supply chain. The worst case is 'Budget Substitute', in which Over the long term, we are optimistic on Chinese IC design compa- Chinese companies cannot innovate and the local supply chain is nies. In terms of manufacturing, we are more positive on the scala- used to produce low-value semiconductors. bility of backend foundries than wafer foundries, but the former's Under the 'Global Commoditization' scenario, China breaks into profitability needs to improve. We are Overweight Goodix (IC design) global semiconductor manufacturing through joint ventures and and USI (backend foundry). On memory, we think private company overseas M&A but with limited innovation. Finally, our base-case YMTC could become relevant in the NAND Flash industry by 2020. scenario is that a 'New Breed' of semiconductor company emerges, We think the key challenges facing China's semiconductor industry propelled by technologies like software-driven chip design and are IP acquisition and local industry consolidation. X-stacking NAND. Global implications: The potential emergence of China's semicon- Self-sufficiency is more than just the trade balance: The usual ductor industry could challenge IC designers in the US, Korea, and way to measure China's self-sufficiency in semiconductors is through Taiwan, and be a long-term threat to the global memory industry a simple comparison of imports and exports. However, we think the (particularly NAND and NOR). Growing investment in production resulting deficit – around US$227bn in 2017 – makes the situation would be positive for semi caps, photomasks, and chemicals, but look worse than it is, because China imports most of those semicon- 2019 demand from China needs to be discounted given the slow ductors for use in products that are then exported again. To address progress of Chinese memory fabs. this deficiency, we have developed a 'China Tech Hardware Ecosystem' framework. This measure shows that demand from Chinese system brands (e.g. Lenovo, Xiaomi) is only US$77bn, while M FOUNDATION Contents

5 Four Scenarios for China's Semiconductor Evolution

6 A New Take on China's Semiconductor's Future Development

7 Key charts

8 Executive Summary

14 China's journey so far

17 China's Semi Companies vs. Global Peers in 2018

21 Four scenarios for the future

31 Defining effective self-sufficiency

35 China semi investment perspective

39 Implications for global semis

43 Appendix 1: Potential breakthrough points for China semis

45 Appendix 2: US/China/Taiwan semiconductor policy

48 Appendix 3: Exploring China's semi companies - our tour guide

49 #1 China's wafer foundries

51 #2 China's backend foundries

53 #3 China memory

55 #4 China IC design M FOUNDATION Four Scenarios for China's Semiconductor Evolution

Exhibit 1: Possible outcomes in China's semiconductor industry evolution

Source: Morgan Stanley Research M FOUNDATION A New Take on China's Semiconductor's Future Development

Exhibit 2: Our perspective versus the conventional view of China's semi industry development

Morgan Stanley's perspective Conventional view

Industry strategy Self-sufficiency measure Tech ecosystem Trade balance

Core of industrial policy Products Production

Focus sub-sector IC Design, OSAT Wafer Foundry, Memory

Relevant technology benchmarks System-level integration (e.g. SiP) Moore's Law

IC design expertise Software-driven ASIC design Large-scale integration (LSI) for SoC Corporate strategy Main operation High-value design Low-margin manufacturing

Spending focus R&D Capex

Operational KPI Value creation Market share

Financial KPI ROE Revenue scale

Partnerships with foreign companies Global collaboration JV/M&A, technology transfers

Main founding source Capital market Government funds

Source: Morgan Stanley Research

M FOUNDATION Key charts

Exhibit 3: Exhibit 4: China's semiconductor consumption at different stages of the tech Our estimates of China's self-sufficiency in semiconductor compo- supply chain nents 300 70%

243 250 60% 50% 200 40% 150 30% 27% (US$bn) (US$bn) 100 91 100 20% 16% 12% 50 10%

- 0% China factory China system brand China end-user China factory self- China system brand self- China end-user self- consumption consumption consumption sufficiency % sufficiency % sufficiency % Source: Gartner, company data, Morgan Stanley Research estimates Source: Gartner, IHS Markit, IDC, company data, Morgan Stanley Research estimates

Exhibit 5: Exhibit 6: China Semiconductor Index market capitalization and stages of devel- Sales of China-based semiconductor companies opment 800 Rmb bn % 50% 700 40%

600 30% 500 20% 400 10% 300 0% 200

100 -10%

0 -20% 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019e Sales (LHS) YoY (RHS)

Source: TrendForce, Morgan Stanley Research

Source: Bloomberg, Morgan Stanley Research

Exhibit 7: Exhibit 8: China semiconductor margins, by sector China's self-sufficiency ratio by end-device 60 Industrial and others 50 Drone Gameconsole 40 Basestation 30 Networking equipment Home Appliances 20 Surveillance 10 Auto

- TV Server (10) PC (20) Feature phone Smartphone (30) Mar-05 Apr-06 May-07 Jun-08 Jul-09 Aug-10 Sep-11 Oct-12 Nov-13 Dec-14 Jan-16 Feb-17 Mar-18 - 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 Foundry IC Design OSAT (US$ mn) China self produced value Produced by other regions Source: Gartner, company data, Morgan Stanley Research estimates Source: Gartner, IHS Markit, IDC, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 7 M FOUNDATION Executive Summary

Why write this report now? …will shape four possible outcomes

We are taking a step back from recent noise about China/US trade Regarding the first question, we think the key to advancing the evolu- tension to assess the structural evolution of China's semiconductor tion of China's semiconductor industry clearly lies in chip design industry. Our aim with this report is to provide long-term perspective rather than pursuing large-scale, low-margin manufacturing. To put to investors who wish to participate in the capitalization of the it another way, Chinese semiconductor firms must prioritize owning industry in China. Global semi investors may also want to know what 'the product' over the means of production. the potential rise of Chinese semi companies could mean for their peers around the world. As for the second question, the current tendency is toward localiza- tion (investing in and depending on China's own technology) over Two critical questions... globalization (leveraging foreign technology through M&A, joint ventures, and licensing). Localization could prove very rewarding but also carries significant risks. We believe China has the potential to increase local production of semiconductors given its large end market and abundance of capital. With all this in mind, we see four potential scenarios for the industry's It also has the motivation given the importance of semiconductors to evolution, ranked from what we view as most favorable to least almost every aspect of modern life. However, compared to technolo- favorable: gies such as solar PV modules and LEDs, the semiconductor industry is much harder to crack given the long supply chain and highly-pro- l 'Industry Upgrade' – China achieves technology breakthroughs tected intellectual property. So far, the Chinese government’s subsi- and high efficiency in chip manufacturing dies and private equity investments have not lifted the country's l 'The New Breed' (our base case) – These are companies that com- semiconductor industry into a position of global leadership, as they bine hardware and software in chip design, while maintaining have for solar and LEDs. selective capabilities in chip manufacturing. l 'Global Commoditization' – China gains a foothold in global semi For China's semiconductor industry to evolve, we believe the funda- manufacturing, with more JVs and overseas M&A but limited inno- mental questions that must be addressed are: vation in chip design. l 'Budget Substitute' – Chinese companies fail to innovate; the 1. What should the main focus be – investing in chip design or local supply chain focuses on production of low-value semicon- increasing production? ductors. 2. Is localization or globalization the best way to achieve effec- tive self-sufficiency?

8 M FOUNDATION The trade imbalance of semis shouldn't be demand from system brands is much lower than what import data would suggest, but still we see an insufficient supply from local IC the core problem design companies with around a 16% self-sufficiency ratio.

The usual way to measure China's self-sufficiency in semiconductors Why 'the product' is more important than is through a simple comparison of imports and exports. However, we think the resulting deficit – around US$227bn in 2017 – makes the production situation look worse than it is, because China imports most of those semiconductors for use in products that are then exported again. China's semiconductor companies account for 3.5% of global semi- Also, there is some confusion about classifying imports from the US, conductor output in value terms. While it is true that semiconductor as described in a report by our US semiconductor analyst Joe Moore manufacturing has the potential to contribute more to China's GDP in Semiconductors: Context on potential China semiconductor pur- and reduce its dependence on imports in the short term, we see long- chase commitments (19 Feb 2019). term risks to an approach that favors production.

We believe it is more effective to evaluate the self-sufficiency of We acknowledge that building fabs is the fastest way to boost China's China's tech supply chain. We measure this by looking at how much global market share in semiconductors. However, chips can be pro- of the demand from China's system brand companies can be supplied duced in any clean room globally, which means that any business out- by domestic IC design companies. The US$77bn semi component sourced to Chinese fabs could easily be moved back overseas.

Exhibit 9: To build a supply chain that will be sustainable over the China's semiconductor imports and exports long term, we believe Chinese semiconductor compa- 350,000 40% nies need to focus more on chip design (the product) 300,000 35% than on building fabs (production). We believe three 30% 250,000 recent success stories illustrate the potential of this 25% 200,000 approach: 20%

(US$ (US$ mn) 150,000 15% 1. Bitmain, a Bitcoin mining chip design company in 100,000 10% China, increased revenue from US$100mn in 50,000 5% 2016 to US$600mn in 2017, reaching over 60% - 0% 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 global market share. China semi export (A; US$mn) China semi import (B; US$mn) Sufficiency Ratio (%) 2. The wireless communication chip design house Source: WIND, Morgan Stanley Research HiSilicon generated US$5bn of revenue in 2018; Exhibit 10: the mainstream 4G smartphone chip vendor, Conventional view of self-sufficiency is based on the Trade Balance method Spreadtrum, earned US$1.5bn of revenue in 2017. 3. The touch panel controller IC design house Goodix has successfully transitioned to new product cycles when disruptive technologies have emerged (e.g, optical fingerprint sensors).

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 9 M FOUNDATION

Exhibit 11: Our Perspective: Self-sufficiency based on the China Tech Hardware Ecosystem method

Source: Morgan Stanley Research

What are the most immediate and gigabit density cost by at least half – paving the way for disruptive trends that may benefit China Chinese memory companies such as YMTC to enter the memory market. Notably, YMTC has invented its own semis? X-stacking technique, which puts NAND cells and I/O arrays on two separate dies, which are then packaged together on a While transistors continue to shrink per Moore's Law, and China single chip. could invest in capacity for leading-edge nodes (e.g., SMIC and 14nm), 3. Heterogeneous chip integration (in contrast to further the returns on such endeavors would likely be poor, as incumbents geometric migration): This is also called “more than Moore”, TSMC and Samsung dominate the space, having invested significant and it integrates semiconductors at the chip packaging level, capex/R&D over the past decade. instead of by using large-scale integration (LSI, the process of integrating or embedding thousands of transistors on a single Rather, as noted, to build a long-term sustainable semi industry, we silicon wafer. In this field, JCET is the leading company in think China should differentiate itself by focusing on chip design, China, although there is still a technology gap vs. global peers system integration, and disruptive technologies. China is already like ASE and Amkor. home to successful IC design companies like HiSilicon, Bitmain, and Goodix, and we believe there will be more to come. For example: Pursuing such developments in the semiconductor industry would require more investment in R&D instead of continuing to pump in 1. ASIC design (software+hardware integration) for AI semis: capex. Please see Appendix 1: Potential breakthrough points for Application-specific integrated circuits (ASICs) are already China semis for more. Besides those new technology trends, we are widespread. The Bitcoin mining ASIC is a recent example also seeing new business models such as "Commune IDM" happening where China had as much as 60% market share globally in in China. 2017. Now, AI applications are becoming the new wave for software+hardware integration chip design. Local Chinese The table below lists Chinese semiconductor companies with poten- companies like Cambricon and Robotic Horizon are early tial, and their closest global and Taiwan industry counterparts. We movers, while internet giants Baidu, Alibaba, and Tencent refer to these companies as 'semi-similars'. For companies that create (BAT) have all started to invest in related chip design. new technologies or business models, and for which we cannot find 2. 3D NAND in memory: This technology has been disrupting many similars outside of China, we call them the 'new breed'. the NAND industry over the past two years by reducing per

10 M FOUNDATION

Exhibit 12: China semi-similars and the 'new breed'

Company Ticker Product Global counterpart Taiwan counterpart Type

A-share Goodix 603160.SS IC design Synaptics Egis New Breed USI 601231.SS Module Murata New Breed JCET 600584.SS Backend foundry Amkor ASE Tongfu 002156.SS Backend foundry Amkor ASE Tianshui Huatian 002185.SS Backend foundry Amkor ASE GigaDevice 603986.SS NOR/MCU Cypress Macronix Sinyang 300236.SZ Chemical Sumco GWC Silan 600460.SS Power IC STMicro GMT NationZ 300077.SS Smartcard IC NXP NAURA 002371.SZ Semi equipment Applied Materials All-Ring WLCSP 603005.SS Testing & packaging Amkor Xintec H-share SMIC 0981.HK Foundry Globalfoundries UMC HHGrace 1347.HK Foundry Globalfoundries Vanguard Huada Semi 0085.HK Smartcard IC NXP US listed Alpha & Omega AOSL.O Power Discrete ON Semi ACM ACMR.O Equipment Applied Materials Taiwan listed Silergy 6415.TW IC design Texas Instrument MediaTek New Breed Focal Tech 3545.TW IC design Synaptics Novatek Private Bitmain IC design Bitfury New Breed Cambricon AI Semi Wave Computing New Breed Robotic Horizon AI Semi Wave Computing New Breed YMTC Memory IDM Micron Nanya Tech New Breed Zhaoxin x86 CPU Intel VIA New Breed SiEn Commune IDM Globalfoundries Powerchip New Breed HiSilicon IC design Qualcomm MediaTek Spreadtrum IC design Qualcomm MediaTek Amlogic IC design Actions Semi Rockchip IC design MediaTek Vimicro IC design Qualcomm MediaTek Actions Semi IC design Qualcomm MediaTek BCD Power IC Maxim Richtek San-An IC GaAs foundry Skyworks WIN Semi AMEC Equipment Applied Materials ChipONE IC design Skyworks Novatek Omnivision CMOS ON Semi Himax GalaxyCore CMOS ON Semi Himax SGIT Smart meter, charger Maxim

Source: Gartner, company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 11 M FOUNDATION How might these outcomes reshape the leading-edge SoC design activity would fuel demand expansion for photomask-related materials and production equipment. global semi industry? Outside of the equipment segment, it could be a zero-sum game. We think the evolution of China's semi industry will generally be posi- China is the world’s largest and fastest-growing market for semicon- tive for the global semi equipment sector. In any of the three sce- ductors, but given that no Chinese chipmaker has broken into the top narios – Industry Upgrade, New Breed, Global Commoditization – we 15 globally in terms of sales, the implications for incumbent players expect China's capex to increase from 2019 (YMTC's 3D-NAND is likely to be meaningful over time. China poses strategic risks to the 64-layer process capacity investment is likely to come first). Under Korean memory segment and Taiwan’s logic and back-end industries. the New Breed scenario, leading-edge SoC design activity by China's This can already be seen in the ramp-up of Chinese purchases of pre- IC designers would likely burgeon over the longer term. The brisker cision machinery for its factories.

Exhibit 13: Global implications of our four scenarios 1 2 3 4 Scenario Industry Upgrade The New Breed Budget Substitute Commoditization Semi industry's focus area Design / R&D Design / R&D Manufacturing / capex Manufacturing / capex Semi industry's overall direction Globalization Localization Localization Globalization

Implications for China semis IC Design, Foundry, OSAT, High-end IC Design, NAND, Low-end IC Design, OSAT, Semi Winners OSAT, Semi Cap, Chemical Memory OSAT Cap Losers Semi Cap Semi Cap IC Design, Memory High-end IC Design, Foundry Implications for global semis Japan Semi Cap, Design Service, Overseas IC Design, Memory, Winners Global Semi Cap US and Japan Semi Cap IP Analog IDM Global IC Design, NAND, Analog Second-tier Foundry, Memory, Losers US IC Design, NAND Second-tier Foundry, OSAT IDM, Discretes Analog IDM Impact on overseas peers Logic chip design Wafer foundry OSAT DRAM NAND Analog/RF/Power discrete Semi equipment/Raw wafers Design service/Photomask/IP

Impact on domestic Chinese companaies Logic chip design Wafer foundry OSAT DRAM NAND Analog/RF/Power discrete Semi equipment/Raw wafers Design service/Photomask

Source: Morgan Stanley Research

12 M FOUNDATION

Exhibit 14: Identifying potential winners and losers in our four scenarios

Scenario Winners Losers China Overseas China Overseas IC Design, Foundry, OSAT, Global IC Design, NAND, Global Semi Cap Semi Cap Memory Analog IDM, Discretes Industry Upgrade SMIC, Silergy, GigaDevice, USI, NVIDIA, MediaTek, TI, Micron, AMAT, LAM, TEL JCET Samsung, Cypress, ON Semi

China Overseas China Overseas High-end IC Design, NAND, Japan Semi Cap, Design Semi Cap US IC Design, NAND OSAT Service, IP The New Breed Micron, SK Hynix, Samsung, Goodix, USI Hoya, NuFlare, Global Unichip NVIDIA, AMD, Xilinx

China Overseas China Overseas Overseas IC Design, Memory, OSAT, Semi Cap, Chemical IC Design, Memory Second-tier Foundry, OSAT Analog IDM Budget Substitute Samsung, SK Hynix, Micron, JCET Goodix, GigaDevice Vanguard, UMC, ASE NVIDIA, Intel, TI, ADI, MTK

China Overseas China Overseas Low-end IC Design, OSAT, Semi Second-tier Foundry, Memory, US and Japan Semi Cap High-end IC Design, Foundry Cap Analog IDM Commoditization Samsung, SK Hynix, Micron, JCET AMAT, LAM Reasrch, TEL Goodix, SMIC Vanguard, UMC, TI

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 13 M FOUNDATION China's journey so far

Will semiconductors follow the path of Government subsidies have played a key solar, LEDs, and panels? role

An interesting approach is to look at industries China has success- In contrast with some third-party research that bases its analysis on fully entered in the past decade, namely solar, LEDs, and panels. As semi companies' revenues, we track China's local semiconductor China obtained technological know-how in different industries, its demand via our proprietary China semi market cap index. We believe main focus was often on capturing market share through aggressive that this approach better captures the value rather than just the capacity expansion and cost cutting. As shown in the chart below, scale of the industry. We divide the development of China's semicon- once China ramps up its global market share, there is usually a drastic ductor industry into five stages, each propelled or hampered by dif- decline in gross and operating margins for the industry. In the short ferent key events. term, it could be beneficial for system brands that can adopt the latest technology at a cheaper price point, speeding up the adoption Exhibit 16: phase. In the long run, however, it may cause many international com- Stages of China's semiconductor capitalization index panies, who demand higher margins, to invest less in R&D than previ- ously.

It will be an important trend to watch. It is unknown whether China will adopt a similar strategy for semiconductors once it obtains crit- ical IP and makes important breakthroughs. However, commoditiza- tion is a definite possibility. So far, other than some mild impact on gross margins on the backend, the overall semiconductor industry still has entry barriers that have fended off commoditization in recent years. Source: Bloomberg, Morgan Stanley Research

Exhibit 15: Comparison of semiconductor, solar, LED, and panel industries

Semiconductor Solar modules LED TFT-LCD Panel 2018 Industry size (US$bn) 463.4 38.9 20.5 57.2 China market share in 2010 1% 50% 8% 15% China market share in 2018 4% 70% 25% 55% Leading company Intel JinkoSolar Cree LG Display GM in 2010 65% 29% 47% 15% GM in 2018 62% 13% 27% 12% Diff. -3.6% -16.0% -19.8% -2.6% OPM in 2010 36% 21% 23% 7% OPM in 2018 33% 2% -22% 0% Diff. -2.8% -19.4% -44.8% -6.9%

Source: Gartner, company data, Morgan Stanley Research

14 M FOUNDATION Stage 1 (2004): Infancy Exhibit 17: SMIC R&D vs. government support 400 The first major China semi company capitalization was SMIC’s H-share listing in 2004. At that time, SMIC’s goal was to address 200 China’s local semiconductor demand. Before 2004, the industry was - in its infancy and almost 100% dependent on semiconductor imports. (200)

(400)

Stage 2 (2005-2012): US-trained talent returns, (600) NASDAQ listings 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 R&D dollar Government Subsidy from financials (US$mn)

Starting in 2005, Chinese semiconductor experts who had been Source: Company data, Morgan Stanley Research working in the US started to return to China to launch their own com- panies, such as Vimicro, Spreadtrum, and RDA. These companies pur- Exhibit 18: sued NASDAQ listings to gain exposure to global investors. In the JCET R&D vs. government support 400 same period, there was also a slew of A-share listings, including 31e HKD to SMIC 200 300mn USD to JCET TFME, RDA, and NationZ. - (200) Stage 3 (2012-2014): Capitalization from China A-/ (400) H-shares, government subsidies (600) (800) China's capital markets yielded high valuations, attracting more local (1,000) (1,200) companies to list, such as NationZ and HHGrace Semi. Meanwhile, 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 the government expanded subsidies to local Chinese companies in R&D dollar (Rmb$mn) Government Subsidy (Rmb$mn) terms of R&D funding and tax breaks via the country's 12th and 13th Source: Company data, Morgan Stanley Research Five-Year Plans. Exhibit 19: Big Fund's original investment breakdown by segment, through 2017 Changing direct government subsidies to 7% private equity fund investments 10%

Stage 4 (2014-2017): National IC Fund investment, JV, Foundry M&A IC design 20% Packaging and testing The China Integrated Circuit Industry Investment Fund, or 'The Big Equipment materials 63% Fund', was founded in September 2014. It was the first fund of its kind to utilize the philosophy of the capital markets to replace traditional subsidies and tax exemptions to nurture an industry. Through the end of 2017, it had invested in 62 projects with investment outflows of Source: Trendforce, Morgan Stanley Research Rmb79.4bn. The invested projects are spread across many segments of the semiconductor industry, including foundry, IC design, pack- M&A, although these have sometimes faced challenges from the aging and testing, and equipment materials. It is important to monitor acquiree candidate's local government. Given this, China often turns the fund's activities to determine the focus of China's semiconductor to the JV option – for example, Unigroup's JV with ChipMOS, a efforts. Taiwan-based driver IC-focused packaging and testing company. For foreign companies wishing to operate in China, it is common for China Running JVs and pursuing M&A are other major avenues by which to require them to establish a JV with Chinese investors who have at China is trying to participate in the global semiconductor supply least a 50% share. Examples include JLQ Technology, the JV entity chain. In the past few years the media has reported on several created in China by Qualcomm, JAC Capital, Datang Telecom, instances of Chinese semi companies being interested in pursuing Leadcore, and Wise Road Capital.

MORGAN STANLEY RESEARCH 15 M FOUNDATION

Exhibit 20: Recent investment activity by China's Big IC Fund Date Target Chinese Buyer Activity Technology Big IC Fund, Zhangjiang Torch Dec-18 VeriSilicon Strategic investment Design service Venture, Xingcheng Capital Big IC Fund, Summit View Nov-18 Rockchip Capital, Fortune VC, YiHe Strategic investment SoC Capital, Shanghai STVC Aug-18 Suzhou Guoxin Big IC Fund D Round Multimedia Jun-18 Cengol Big IC Fund, TFTR Investment C Round Power discretes Jun-18 YDME Big IC Fund, E-Town Capital Strategic investment Power discretes Apr-18 SMIC Big IC Fund B Round Foundry Jan-18 SMIC Big IC Fund Strategic investment Foundry Jan-18 HuaHong Big IC Fund Strategic investment Foundry Dec-18 WLCSP Big IC Fund Strategic investment Packaging Sep-17 Semtech Big IC Fund Strategic investment Power discretes Dec-16 AMEC Big IC Fund A Round Equipment Oct-16 Naura Big IC Fund Strategic investment Equipment Oct-16 Darbond Big IC Fund A Round Chemicals Sep-16 Centec Big IC Fund Strategic investment Networking Aug-16 Jiangsu Xinhua Big IC Fund A Round Materials Jul-16 Anji Microelectronics Big IC Fund B Round Materials Jun-16 SMIC Big IC Fund A Round Foundry Nov-15 ZTE Big IC Fund Strategic investment Telco equipment

Source: Rhino data, Morgan Stanley Research

Exhibit 21: China semiconductor JVs

Date JV Entity Overseas Partner Chinese Partner % Owned by Chinese Partner Technology Mar-18 SAIC Infineon (SIAMPM) Infineon SAIC 51% Power discretes May-18 JLQ Technology Qualcomm Datang 75.9% Smartphone chipset Jan-18 Suzhou Lite-On Storage Lite-On Unigroup 55.0% SSD May-17 Arm mini China ARM Bank of China + Baidu 51% IP Nov-16 United Semiconductor (Xiamen) UMC Xiamen Government 66.7% Foundry Nov-16 ChipMOS Technologies (Shanghai) ChipMOS Unigroup 55.0% TDDI Packaging Jan-16 Unisoc Intel Unigroup 80% Logic

Source: Morgan Stanley Research

Exhibit 22: China semiconductor M&A Date Target Main Chinese Buyer % Owned Technology Dec-15 ISSI GigaDevice Terminated SRAM, DRAM Aug-15 STATSChipPAC JCET 100% Advanced packaging Dec-15 Mattson Technology E-Town Dragon 100% Equipment Jun-16 Exar iML ChipOne 100% Analog, PMIC Jun-16 Lfoundry SMIC 70% Foundry Oct-15 AMD (2 packaging units) TFME 100% Advanced packaging May-16 Imagination Unigroup 3% IP Apr-16 Lattice Unigroup 6% FPGA and logic Jul-16 Omnivision Ingenic 100% CMOS

Source: Morgan Stanley Research

16 M FOUNDATION China's Semi Companies vs. Global Peers in 2018

Specialization in the global semi supply chain There are peripheral industries to support core semi production, such as semi IP/EDA, equipment, photomasks, and chemicals. The core of semiconductor industry is chip design and manufac- Component distributors also help to handle logistics and customer turing. Around 30 years ago, most semis were designed and fabri- support for OEM/EMS companies. cated by IDMs (integrated device manufacturers) which run a vertically-integrated business model. Wafer foundry TSMC com- menced an outsourced manufacturing business model, and chip pro- duction can be specialized by after wafer foundry, OSAT, and IC design.

Exhibit 23: Global semiconductor supply chain flow chart

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 17 M FOUNDATION Snapshot of China's global semi If we look at China's semiconductor share by subsegment ( Exhibit 7 ), there are only three segments within the supply chain in which positioning Chinese players had over 5% revenue market share as of end-2017: OSAT (24%), IC design (14%), and wafer foundry (7%). The key differ- Looking at the following two charts on China's 2017 semi market entiator for these leading segments in terms of market share is the share side by side, one can easily spot a current weakness in China's need for expertise beyond purchasing equipment from foreign ven- semi supply chain. Examining core segments (e.g., IC design, foundry, dors. Taking OSAT as an example, 40% of the overall OSAT market and OSAT), it is clear that Chinese companies can leverage existing is still dominated by the most traditional wire bonding business. With infrastructure, equipment, and IP pools if they obtain the capital – as less advanced expertise needed after purchasing wire bonders, new evidenced by the higher market shares in these segments. However, entrants could adjust the parameters of the equipment and start pro- the difficult link for China to break through is in its supporting semi duction in a shorter period of time. This explains why OSAT became value chain (e.g., IP/EDA tools, equipment, raw wafers). In these seg- one of the most dominant (but margin-poor) semiconductor-related ments it is more difficult to buy and accumulate knowledge by simply segments in China, and the earliest such segment that the country contributing government money; rather, it tends to take years of penetrated. basic research before becoming a significant player in the global semi value chain.

Exhibit 24: Core semi value chain market share, 2017

Source: Gartner, Morgan Stanley Research

Exhibit 25: Supporting semi value chain share, 2017

Source: Gartner, Morgan Stanley Research

18 M FOUNDATION China's local semiconductor revenue and Exhibit 26: Sales of China-based semi companies (Rmb bn) growth by subsegment 300 250

According to the China Semiconductor Industry Association (CSIA), 200

China's local semiconductor industry size exceeded Rmb446bn in 150 revenue for 9M18, up 22.4% from 9M17. We believe the key drivers 100 behind the expansion include: 1) faster substitution by local providers 50 of semi components that have lower barriers to entry (e.g., low - power discretes), and 2) stronger local support from the government IC Design Foundry OSAT 2015 sales 2016 sales 2017 sales 2018 sales in view of increasing trade tensions in 1H18. Source: TrendForce, Morgan Stanley Research l Wafer foundry – 9M18 revenue of Rmb115bn, +27.6% YoY Exhibit 27: l Backend foundry – 9M18 revenue of Rmb152bn, +19.1% YoY Sales of China-based semi companies l IC design – 9M18 revenue of Rmb179bn, +22.1% YoY 800 Rmb bn % 50% 700 40%

600 However, assessing 3Q18 alone, total semiconductor revenue of 30% 500 20% Rmb173bn represented a 6.1% YoY decline, suggesting that trade ten- 400 10% sions may have affected demand – a factor that we expect will persist 300 0% into 2019. We believe that without support from foreign companies, 200 the development of China's domestic semiconductor industry would 100 -10% 0 -20% decelerate, as it would face steeper challenges operating on its own. 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019e Sales (LHS) YoY (RHS)

Source: TrendForce, Morgan Stanley Research Among subsegments, China has built up its local OSAT supply the fastest, led by JCET, despite poor profitability. However, China has also developed several strong IC design companies, such as HiSilicon focus purely on its capabilities. For memory IDM, it could still be diffi- (smartphone processors, system-on-a-chip [SoC] products), Bitmain cult for China to be a substantial player in the field of DRAM, but, with (the industry leader in cryptocurrency mining ASICs), and Goodix YMTC's potential breakthrough in 3D NAND, it is likely to have a cap (fingerprint sensors), all of which are among the top players in their of 20% market share (absent any consideration given to the supply industries. and demand dynamics of the global supply chain).

Where can Chinese companies be more Industry segments that we expect to experience commoditization by 2025 include: 1) NAND flash, 2) raw wafers, 3) RF semis, 4) IC design, globally relevant in the long term? and 5) the backend. Segments that we believe present high barriers to entry include: 1) power semis, 2) DRAM, 3) semi equipment, and 4) Another interesting framework we introduce is the 'technology cap' IP. for each segment. This shows our estimated maximum market share for the specific segment if China is to spend money to buy more By business model segment, through 2025 we expect China to make equipment in that particular field. For example, SMIC, with its 28nm significant market share gains (with some exceptions) in IP/EDA tools technology, would likely obtain, per our assessment, a theoretical (5% by 2025), chemicals (5%), and memory IDM (3%). global market share of 70%, if we ignore the competition factor and

MORGAN STANLEY RESEARCH 19 M FOUNDATION

Exhibit 28: China's market share by semiconductor product and its maximum potential in a world without trade or policy limitations

Others (LED, Laser, etc.) Other sensors (Fingerprint, Magnetic, etc.) Image sensors Wireless/Wired connectivity and RF Discretes ( Transistors, Diodes) Other analogs Analog - power IC Other memory NAND DRAM Other logics (DSP, Smartcard IC, etc.) DDIC ASIC MCU FPGA / PLD GPU ARM-based processors x86 CPU

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

China market share in 2017 China market share in 2020 China market share in 2025 Max Other Global Vendors Source: Gartner, company data, Morgan Stanley Research estimates

Exhibit 29: Global semi value chain TAM and market share summary

China China China China semi market market market Technology US$mn Global market size Global leading companies China local companies Taiwan companies value share in share in share in cap 2017 2020 2025 IP/EDA tools 11,000 ARM, Synopsys Imagination, Empyrean, Semitronix, XpeedicAndes, M31, eMemory, SpringSoft 185 2% 3% 5% 5% Photomask 4,000 DNP, Shin-Etsu Zhongwei Mask, CR Micro Taiwan Photomask 120 3% 3% 10% 20% Chemicals 12,800 Shin-Etsu Sinyang GlobalWafers 80 1% 2% 5% 10% Equipment 35,000 Applied Materials, Lam Research, ASMASM Pacific, AMEC, ACM, Sever Starts, Naura 350 1% 2% 10% 10% Wafer foundry 64,000 GlobalFoundries, Samsung SMIC, HHGrace TSMC, UMC, Vanguard 4,290 7% 9% 11% 70% OSAT 29,000 Amkor JCET, Nantong, Tianshui ASE, Chipbond, Powertech, KYEC 7,018 24% 28% 32% 90% IC Design 270,000 NVIDIA, Qualcomm, Broadcom, Intel,Goodix, T HiSilicon, Omnivision, Spreadtrum MediaTek, Novatek, Realtek 13,860 5% 7% 10% 70% Memory IDM 130,000 Samsung, Hynix, Micron YMTC, GigaDevice Nanya Tech, Macronix, Winbond 400 0% 1% 3% 20% Supply chain total market size 555,800 26,303 5% 6% 9% End products market size 420,393

Source: Gartner, Morgan Stanley Research estimates

For 2025 market share, we have different approaches based on the below 10% by 2025. Similarly, in terms of memory IDM, other than dynamics of each area of the semi supply chain. The rationales are NOR by GigaDevice and some 3D NAND by YMTC, China still lacks described below. For IP/EDA tools, chemicals, and equipment, we crucial memory IP for DRAM and mass-produced NAND to have a find it difficult for China to have meaningful, organic (without M&A) substantial share by 2025 (we forecast it to be 3% by then). For OSAT, breakthroughs as these segments require years of R&D efforts and given the strategy of competitive pricing to gain share, it is likely that the overall semiconductor industry in China has been so reliant on the three China OSATs will outgrow their Taiwan-based and global overseas support in these segments. Because of this, the market peers to achieve a higher market share by 2025. shares for these segments will grow but with limited magnitude at or

20 M FOUNDATION Four scenarios for the future

Our base case is that a 'new breed' of globally com- terms of critical semiconductor technology, China lacked CPU core, petitive semi companies will emerge in China DRAM IP, and 300mn raw wafer for chip production. China's pres- ence in power semis (e.g., IGBT) was very small. As we've said, Chinese semi companies need to focus more on chip design than on building fabs for production. Design requires a greater Evaluating the outlook for 2020 and 2025 level of commitment to investing in R&D, and it demands a significant amount of time. However, we have already seen some top-notch IC Given the emergence of trade tensions, we highlight the 'trade fric- design companies, including HiSilicon, Bitmain, and Goodix, and we tion vs. globalization' dynamic to assess the likely outcome of China's believe more will emerge. efforts to develop its semiconductor industry.

The industry's status in 2018 We also provide some key metrics to monitor in Chinese companies' 1) design capability, 2) semi production, and 3) critical technology Chinese semiconductor companies had 3% global market share in development by 2020 and 2025. For semi production, we break 2018 (by revenue), with an industry gross margin of around 30%. For down the field to wafer foundry, backend foundry, and memory IDM. semiconductor production, China's wafer foundry segment accounted for 9% of global capacity, while the country's backend Lastly, we try to identify the potentially advantaged and disadvan- foundry segment accounted for 24% of global production. taged players in the global semi supply chain under the four sce- Own-brand memory production (IDM) was almost zero in 2018. In narios.

Exhibit 30: Four scenarios for China semiconductors: Industry evolution hinges on technology sourcing (local vs. global) and investment (capex vs. R&D)

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 21 M FOUNDATION

Exhibit 31: Key metrics for China's semiconductor industry, by scenario Subject 1 2 3 4 Scenarios As of 2017 Industry Upgrade New Breed Budget Substitute Commoditization Key industry focus areas Design / R&D Design / R&D Manufacture / Capex Manufacture / Capex Semi Industry's Overall Direction Globalization Localization Localization Globalization 1. Design capability Global market share 3% 15% 8% 5% 8% Industry margin (local vender) 30% 40% 30% 20% 20% 2. Semi production Wafer foundry market share 9% 15% 11% 11% 20% capacity local vendor % 80% 70% 40% 50% most leading edge 5nm 7nm 14nm 7nm

Backend foundry market share 24% 50% 32% 20% 50% local vendor % 80% 70% 60% 70%

Memory market share 0% 13% 5% 5% 13% 3D NAND capacity by 2020 200k 100k 100k 200k DRAM capacity by 2025 100k 0 0 100k 3 .Critical technology development CPU Core x86 CPU RISC-V ecosystem lack lack Memory IP M-RAM Xstacking 3D NAND lack lack Equipment Export Local supply minor minor Raw wafer 300mn qualified 200mn qualified JV JV Power semi IGBT/SiC low-end MOSFET JV JV

Implications to China semis IC Design, Foundry, OSAT, High-end IC Desing, NAND, Low-end IC Design, OSAT, OSAT, Semi Cap, Chemical Winners Memory OSAT Semi Cap

Semi Cap Semi Cap IC Design, Memory High-end IC Design, Foundry Losers

Implications to global semis Japan Semi Cap, Design Overseas IC Design, Memory, Global Semi Cap US and Japan Semi Cap Winners Service, IP Analog IDM Global IC Design, NAND, Second-tier Foundry, US IC Design, NAND Second-tier Foundry, OSAT Losers Analog IDM, Discretes Memory, Analog IDM

Source: Gartner, IHS Markit, company data, Morgan Stanley Research estimates

22 M FOUNDATION 1. 'Industry Upgrade' scenario Exhibit 32: Scenario 1: Industry Upgrade

In this scenario, China is able to export its self-developed semiconductor technology to the world at decent mar- gins. Outcomes include: 1) increased productivity for China and global industrials, and 2) successful capitaliza- tion of Chinese semi companies. We view this as the most favorable scenario for China, with the country becoming a new force in global semis on the back of technology breakthroughs and the proven efficiency of its semi chip manufacturing. a. Design capability

This scenario envisions China growing its chip design Source: Morgan Stanley Research scenarios (products) market share to 15% globally. With stronger R&D efforts translating into capabilities in chip design, China should d. Potentially advantaged players be able to gain market share on all fronts. This scenario sees the overall China semiconductor supply chain ben- b. Semi production efitting, given that the clustering effect in China will enable the whole industry to enjoy an industry upgrade. Semi production in China increases significantly by 2025, on the back of its own development in manufacturing capabilities along with e. Potentially disadvantaged players China IC design companies' share gains. In this scenario, China SMIC successfully mass produces 5nm by 2025, increasing China's foundry Overseas semiconductor companies that lose market share to China, production to 15% of global capacity. Backend foundries' market such as memory players, analog IDMs, and logic chip design houses. share increases to 50% by 2025 from 24% in 2018. NAND flash capacity first reaches 200k by 2020, or 10% of global supply, while DRAM capacity reaches 100k by 2025. c. New technology

In this scenario, China enjoys a breakthrough in designing its own x86 CPU; X-stacking achieves good yields by 2020, while M-RAM is a suc- cess. Self-designed semiconductor equipment is competitive and is able to export overseas. 300mn raw wafer passes the qualification of global leading fabs. The development of new power semi technol- ogies, such as SiC, proceeds.

MORGAN STANLEY RESEARCH 23 M FOUNDATION

2. The 'New Breed' scenario Exhibit 33: Scenario 2: The New Breed Our base case scenario assumes that China's semicon- ductor industry integrates semi chip and software design, and also strategically pursues specific types of semi chip manufacturing. Local innovation from system brands and internet companies creates new types of IC design com- panies. Technology breakthroughs are achieved in certain semiconductor supply chains (e.g., 3D NAND, raw wafers). Higher self-sufficiency is achieved for local China system brands. a. Design capability

This scenario assumes China can increase its chip design Source: Morgan Stanley Research (products) market share to 8% globally, by replacing for- eign chip providers. d. Potentially advantaged players b. Semi production More success stories may emerge from China local IC design houses such as Goodix. Design service and EDA tool demand increases, Semi production in China increases by 2025, given China IC design yielding strong demand for tailor-made chips from companies in the companies' share gains. China SMIC successfully mass produces 7nm China supply chain. by 2025, increasing China's foundry production to 11% of global capacity. Backend foundry market share increases to 32% by 2025 e. Potentially disadvantaged players from 24% in 2018. NAND flash capacity first reaches 100k by 2020, or 5% of global supply. Global NAND flash companies could suffer pricing pressure from China's volume production, and US IC Design houses. c. New technology

China adopts open source CPU architecture to develop AI and IoT chips. X-stacking reaches a good yield by 2020.

24 M FOUNDATION 3. 'Budget Substitute' scenario Exhibit 34: Scenario 3: Budget Substitute

Our most unfavorable scenario would be if China fails to develop its own design/IP, while the US restricts exports of advanced semis such as CPUs and GPUs. China limits foreign semi companies’ business development to pro- tect local companies. Semiconductor investment yields low ROI, and the industry is dependent on local policy support and is hard to capitalize. a. Design capabilities

In this scenario, it is likely that many IC design companies that depend heavily on foreign IP and EDA tools would need to start from scratch in terms of accumulating crit- Source: Morgan Stanley Research ical know-how and in-house IP. Such a development could be positive in the very long term, as it would force China to c. New technology develop not only core technologies but also to focus on peripheral semiconductor segments such as IP and EDA tools. However, under In this scenario, China lacks the new technology required to achieve this scenario we assume that the market share of design companies a meaningful breakthrough in its semiconductor standing interna- in China falls to 5% by 2025, while their average gross margin drops tionally, as it would have to build from the ground up. It would not to 20%. have x86 CPU or critical memory IP, equipment supply would be lim- ited, and the rest of the core technologies would be dependent on JVs b. Semi production with global peers.

Semi production market share might stabilize at 11% as companies d. Potentially advantaged players lose some international orders but gain local demand. However, these products would focus mainly on legacy nodes – 14nm, at most, IC design companies could gain under this scenario as China seeks by 2025, with lower margins as compared with international substi- cost-effective substitutes to international products, while China's tutes. As local IC design companies take a hit, China OSATs could market share could eventually trend up. Chinese OSAT could also come under pressure, as most of these companies are primarily benefit in the long run as the total addressable market would be domestically driven, while some important international projects bigger. might be lost. e. Potentially disadvantaged players

Chinese IC design companies would suffer most under this scenario, as they would likely lose share as international IP becomes more diffi- cult to license. China system brands start to rely more on Taiwan partners to satisfy their global and domestic demand.

MORGAN STANLEY RESEARCH 25 M FOUNDATION 4. 'Global Commoditization' Exhibit 35: Scenario 4: Global Commoditization scenario

Under this scenario, if restrictions on overseas tech- nology were removed yet China still focuses on manufac- turing, global semis may face value disruption. China acquires overseas trailing edge semi technology. The focus here is on production through more JVs, foundry outsourcing, or even IP licenses (with limited usage). Full self-sufficiency. Faster commoditization of global semis. a. Design capability

With some overseas IP or semi company acquisitions,

China gains market share in mature products and Source: Morgan Stanley Research scenarios increases its market share to 8%. That said, industry mar- gins shrinks to 20% given the focus on low-end products. b. Semi production d. Potentially advantaged players

Overseas companies build more fabs in China or use JVs with Chinese Semi equipment vendors benefit from strong orders from China. companies to serve the China market. China's foundry production Overseas semi companies are willing to leverage China's cheap pro- doubles to 20% of global capacity by 2025, similar to the trend in duction to lower costs. China's backend foundries. With potential foundry outsourcing, or JV memory fabs increasing NAND capacity to 200k by 2020, DRAM e. Potentially disadvantaged players capacity to 100k from scratch in 2018. Memory companies suffer from oversupply in NAND and/or DRAM. c. New technology Overseas foundry/backend fabs lose market share.

No major technology breakthrough, but larger scale of manufac- turing.

26 M FOUNDATION Key Pivot #1: Investment direction: 'Design' the innovation leaders economically.” The report said Chinese indus- trial policy can be divided into two categories – subsidies and first or 'Manufacturing' first? zero-sum tactics – and recommends six principles to respond to chal- lenges from China. The market measures China's self-sufficiency in semis from the finan- cial deficit in customs data – i.e., imports minus exports. However, we In November 2018, the US Department of Commerce's Bureau of think it should instead be measured by whether demand from Industry and Security published a document addressing its review of Chinese system brands can be met by local vendors. For example, controls for certain emerging technologies, including AI and machine 100% of Apple iPhones are produced in China, which causes a high learning, microprocessor technologies, and cloud technologies. dollar value of semi components to be imported because local Please see more on US semiconductor policy in Appendix 2: US/ Chinese semi companies don't 'win the design'. In 2018, China was the China/Taiwan semiconductor policy . largest exporter of the Bitcoin mining chip, and the chip design com- pany, Bitmain, could decide the manufacturing location. Therefore by The UMC/Jinhua event is an instance of the US 2025, we think China could focus more on chips 'Designed in China' becoming more diligent about protecting and exe- than chips 'Produced in China'. cuting its tech protection regulations

Key Pivot #2: Technology sourcing: On November 1, a federal grand jury indicted a PRC state-owned com- pany (Jinhua), a Taiwan company (UMC), and three individuals, and Inwards (localization) or outwards charged them with crimes related to conspiracy to steal, convey, and (globalization)? possess trade secrets of an American semiconductor company. A press release indicated that, if convicted, each company faces forfei- ture and a maximum fine of more than US$20 billion. Since 2017, US trade policy and restrictions on semi technology exports have been the key pivot point

In January 2017, an advisory report to the US president called for exec- utive policy focus on semiconductors. The report stated that “if the leading edge is advancing slowly, that makes it easier for China to use industrial policies to get technologically close enough to supplant

Exhibit 36: Proposed overseas PE investments that are pending or have been rejected Date Target Chinese Buyer Technology Deal Size (US$mn) Outcome Dec-16 Aixtron Grand Chip Investment Fund Equipment 7,554 Blocked by US Hua Capital Management, CITIC Jan-16 Omnivision Capital, Golden Stone CMOS image sensor 1,900 Closed Investment Jun-16 NXP (Standard product business) JAC Capital, Wise Road Capital Discrete, Power IC, Logic 2,750 Announced Dec-15 NXP (RF Power business) JAC Capital RF IC 1,800 Closed Linkong Development Jul-15 Bridgelux LED 130 Closed Investment Dec-15 ISSI Uphill Investment SRAM, DRAM 640 Closed

MORGAN STANLEY RESEARCH 27 M FOUNDATION Our assessment of China's effective self-sufficiency by 2020 and 2025, broken down by subsegments

We gauge what portion of semiconductor products is designed by market size. From this we can see that the overall semiconductor local companies, and expand our analysis to self-sufficiency by end market share China possessed in 2017 was only 3.5%, while its market devices. We start by listing the key semiconductor component ven- share in several crucial components such as x86 CPUs and GPUs was dors by product, and calculate the self-sufficiency ratio by estimating virtually nil. On top of that, China's self-sufficiency ratio by end revenue generated by Chinese companies compared to the whole device is mostly below 10%.

Exhibit 37: China's self-sufficiency ratio by end device Industrial and others Drone Gameconsole Basestation Networking equipment Home Appliances Surveillance Auto TV Server PC Feature phone Smartphone

- 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 (US$ mn) China self produced value Produced by other regions

Source: Gartner, IHS Markit, IDC, Morgan Stanley Research Exhibit 38: China's semiconductor share by product

China China China China semi market market market US$mn Global market size Global leading companies China local companies (fabless or IDM) Taiwan companies value share in share in share in 2017 2020 2025 Logic semis x86 CPU 48,292 Intel, AMD Hygon, Loongson VIA Technologies - 0% 0% 1% ARM-based processors 39,679 Qualcomm, MediaTek HiSilicon, Spreadtrum MediaTek 3,120 8% 12% 15% GPU 6,893 NVIDIA, AMD - 0% 0% 0% FPGA / PLD 4,933 Intel, Xilinx, Microsemi Unigroup Guoxin 7 0% 1% 2% MCU 18,020 Renasas, NXP, Microchip, STM Ingenic, CEC Huadam, Unigroup Guoxin Nuvoton, Holtek, Sunplus 983 5% 8% 11% ASIC 26,627 Apple, Broadcom, AMD Hisilicon, Bitmain, Cambricone MediaTek, GUC 2,666 10% 10% 10% DDIC 5,597 Samsung, Novatek, Himax, Synaptics Chipone NovaTek, Himax, Raydium 56 1% 2% 3% Other logics (DSP, Smartcard IC, etc.) 3,013 Broadcom NationZ Realtek, ASMedia, Aspeed 151 5% 5% 5% Total Logic 153,054 6,983 4.6% 6.0% 7.3%

Memory DRAM 72,096 Samsung, SK Hynix, Micron Jin Hua Nanya Tech, Winbond, ESMT - 0% 0% 1% NAND 53,738 Samsung, SK Hynix, Western Digital YMTC Winbond, Macronix, Phison - 0% 3% 9% Other memory 4,447 Cypress, Macronix GigaDevice Macronix 193 4% 7% 10% Total Memory 130,281 193 0.1% 1.5% 4.6% Specialty semis Analog - power IC 21,766 TI, Analog Devices, MediaTek Silergy, Chipone, Joulwatt MediaTek, GMT, Advanced Analog Tech 657 3% 6% 8% Other analogs 11,333 Analog Devices, Cirrus Logic, TI Hangzhou Silan Micro MediaTek, GMT 224 2% 5% 7% Discretes ( Transistors, Diodes) 19,957 Infineon, On Semi, Vishay, Rohm Hangzhou Silan, Good-Ark, Jilin Sino-MicroelectronicPan Jit, TSC, Lite-on Semi 785 4% 7% 9% Wireless/Wired connectivity and RF 42,901 Broadcom, Qualcomm, Skyworks, Qorvo Hisilicon, Vanchip, RDA, ZTE MediaTek, Realtek, Parade 2,327 5% 8% 12% Image sensors 11,496 Sony, Samsung, On Semi GalaxyCore, Omnivison PixArt, Silicon Optronics 1,734 15% 17% 18% Other sensors (Fingerprint, Magnetic, etc.) 10,482 Bosch, Infineon, STM, NXP Goodix Egis, Focaltech, Elan 567 5% 5% 5% Total Specialty 117,935 6,293 5.3% 7.8% 10.2% Others (LED, Laser, etc.) 19,123 Osram, Toshiba, Nichia, Seoul Semi Foshan Nationstar, Guangzhou Hongli, Shenzhen JufEpistar, Everlight, Lextar 1,358 7.1% 10% 13% End products market size 420,393 14,827 3.5% 5.3% 7.6%

Source: Gartner, company data, Morgan Stanley Research

28 M FOUNDATION As some Chinese companies, such as HiSilicon, have shown strong Our conclusion is that full semiconductor self-sufficiency is a long design capabilities, we believe the adoption rate of local semiconduc- way off, considering that China's ultimate goal probably isn't just to tors will rise in future, mainly driven by domestic system brands and satisfy its own needs in entry-level devices. While meaningful consumption. We forecast that China's overall global market share in progress has been made in some products with lower entry barriers, semiconductors will rise to 5.3% in 2020 and 7.6% in 2025, with some more must be done to achieve self-sufficiency. key components like ARM-based processors reaching as much as 15%. In Exhibit 35 , we forecast what Chinese semiconductor companies' market share could be like in a hypothetical world with no manufac- Using smartphones as an example, in 2017 Chinese system brands turing limitations or policy interference. Under this assumption, we accounted for 42% of unit shipments and 30% by value of total believe China's local design abilities could be fully released, and it smartphone consumption globally. However, Chinese-designed and could take as much as 100% market share globally in some semicon- -produced semiconductors accounted for only 11% of the global ductor products such as image sensors and NAND Flash. We provide market by value, with the contribution of several key components this exhibit as we believe it provides a clearer picture of where China's such as DRAM at zero. The self-sufficiency rate for devices that design abilities stand globally and how much self-sufficiency could require less sophisticated semiconductor content, such as feature China achieve in the long run. phones and home appliances, is obviously higher, but that only high- lights the fact that Chinese semiconductor companies are struggling to make gains in more advanced products.

Exhibit 39: China's semiconductor demand breakdown Global production Semi conductor China consumption China brand China production Semi content per box (US$) units (in mn) demand (US$mn) Smartphone 1465 30% 42% 70% 69 101,735 Feature phone 512 20% 30% 80% 16 8,408 PC 255 27% 24% 80% 311 79,454 Server 10 17% 5% 50% 3000 30,569 TV 233 25% 20% 70% 85 19,872 Auto 82 30% 10% 20% 474 39,112 Surveillance 150 40% 30% 70% 20 3,000 Home Appliances 1000 20% 15% 50% 22 21,537 Networking equipment 100 20% 20% 70% 100 10,000 Basestation 1 25% 30% 40% 1500 1,500 Gameconsole 47 15% 5% 70% 157 7,380 Drone 3 20% 49% 80% 101 302 Industrial and others 20% 10% 40% 97,523 Total demand (US$mn) 24% 22% 61% 420,393

Source: Gartner, IHS Markit, IDC, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 29 M FOUNDATION

Exhibit 40: Exhibit 41: Chinese system brands' semiconductor consumption by value China's local production by value

Smartphone Smartphone Feature phone Feature phone PC PC Server Server TV TV Auto Auto Survilliance Survilliance Home Applianes Home Applianes Networking equipments Networking equipments Basestation Basestation

Source: Gartner, IHS Markit, IDC, Morgan Stanley Research estimates Source: Gartner, IHS Markit, IDC, Morgan Stanley Research estimates

10 signposts for China's semi industry 4. HiSilicon turns into a merchant vendor from a captive design house for Huawei evolution over the next 12 months 5. China successfully develops its own CPU cores and commer- cializes the RISC V ecosystem Aside from our top-down framework for China's semiconductor 6. Raw wafer companies achieve a breakthrough in 12-inch epi- development, here are key events that could help us verify whether substrate technology and how soon it is happening: 7. China consolidates its semiconductor industry to few clus- ters 1. More ASIC chip designs driven by China’s internet/software 8. US removes restrictions on semiconductor industry exports companies 9. Progress on China’s overseas M&A 2. Whether China YMTC can successfully claim IPR in 3D NAND 10. More IPOs 3. JCET develops competitive SiP and fan-out technology for heterogeneous chip integration

30 M FOUNDATION Defining effective self-sufficiency

There is a fair bit of confusion around the way 'imports' are defined. We wrote this section to discuss how to count semiconductor pro- For example, the Semiconductor Industry Association (SIA) esti- duction and consumption, and assess what that means for China's mates that US$157bn worth of semiconductors, of 33.6% of global self-sufficiency. Our conclusion is that self-sufficiency can be defined consumption, were sold to China in 2018. However, China customs in different ways. The market and the press mainly measure China's data suggests total imports were US$312bn, which could mean some semiconductor self-sufficiency in terms of the trade deficit – i.e., that some semiconductors were sold to customers outside China, imports minus exports. However, we think a more meaningful but the ultimate shipment destination was factories within China. measure is whether a semiconductor component is locally designed The definition of 'US semiconductor products' is also somewhat or not. blurry. Does that mean semis produced by fabs located in the US, or products designed by US semi companies regardless of where they We estimate that local IC design houses can currently meet 16% of are produced? Some of these issues are discussed by our US col- demand from local system brands. Comparing end-user consump- league Joe Moore in his report Semiconductors: Context on potential tion against local production, self-sufficiency is around 12%, accord- China semiconductor purchase commitments (19 Feb 2019). Our ingly to our analysis. Below, we go into more detail with a graphical assessment is that a significant majority of the US$157bn worth of demonstration of the pros and cons of each self-sufficiency defini- semiconductors shipped into China is going into end products that tion. are ultimately exported from China, and a large portion goes to multi- nationals.

Exhibit 42: SIA's estimates of semiconductor product shipments to China

Product Category Global Shipments to China % of Global Shipments that go to China Discretes $8.9 37% Optoelectronics $13.0 34% Sensors $4.6 35% Analog $21.6 37% Microprocessors $8.8 19% Microcontrollers $4.9 29% Other Lofic $35.3 32% DRAM $41.8 42% NAND $17.3 32%

Source: Morgan Stanley Research, Semiconductor Industry Association

Exhibit 43: China's self-sufficiency ratio: Alternative perspectives

Source: Gartner, IHS Markit, IDC, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 31 M FOUNDATION 1. Self-sufficiency based on the Trade from overseas semi fabs to China OSATs to be packaged or to China OEMs to be assembled (import). Balance method With this method, the self-sufficiency ratio is around 27% – the This is by far the most straightforward method, with the easiest highest among the three. access to data points. If we look at Exhibit 45 , we start with overseas or local IC design houses having design-in with overseas or local The advantage of this method is that it is easy to track and under- system brands like Apple or Huawei. After that, they will then out- stand. However, many imports are destined for assembly businesses source their manufacturing to semi fabs including foundries and rather than local consumption, making China's reliance on semi OSATs. Once the component is made into wafers, this will be the key imports seem more pronounced than it actually is. It also distorts export area within the method. The final wafers could be shipped actual exports from China by including the production of foreign from China semi fabs to overseas OSATs to be packaged (export) or companies.

Exhibit 44: China's imports and exports of semiconductors 350,000 40%

300,000 35%

30% 250,000 25% 200,000 20%

(US$ (US$ mn) 150,000 15% 100,000 10%

50,000 5%

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

China semi export (A; US$mn) China semi import (B; US$mn) Sufficiency Ratio (%) Source: China Customs

Exhibit 45: Self-sufficiency based on Trade Balance

Source: Morgan Stanley Research Note: China semi exports = China semi fabs' exports to overseas OEMs + China semi fabs' exports to overseas semi fabs; China semi imports = China OEMs' imports from overseas semi fabs 32 M FOUNDATION 2. Self-sufficiency based on the Tech the country's total need for semiconductors. Therefore, total semi- conductor demand in China will be all the semiconductor content the Hardware Ecosystem method Chinese system brands require (components from both domestic and overseas design houses). And the total amount of semiconduc- In this method we focus on the design capabilities of China's semicon- tors China is able to produce will be the combined value of output ductor companies to get closer to real demand and supply. We cate- from China's design houses, no matter whether it is used by domestic gorize semiconductor demand based on the components needed for or overseas system brands. Chinese and overseas system brands to build their products. The most important step is how these system brands source semicon- With this method, we estimate the self-sufficiency ratio at 16%. ductor components and how they complete the final product. It mat- ters less where these system brands end up selling their products. The main advantage of this method is that we believe it captures real But the total number of products sold matters, as this can indicate economy activity with less double-counting through exports and imports.

Exhibit 46: Self-sufficiency based on the Tech Hardware Ecosystem

Source: Morgan Stanley Research. Note: China system brand semi demand = China semi designs for China system brands + overseas semi designs for China system brands; China system brand semi product = overseas semi designs for China system brands + China semi designs for China system brands

MORGAN STANLEY RESEARCH 33 M FOUNDATION 3. Self-sufficiency based on the GDP The advantage of this method is that it is very intuitive. However, the disadvantage is that it could overstate China's ability to be totally Creation method self-sufficient if China has many commoditized component technolo- gies but not critical technologies like CPUs and memory. That could The last method is based on GDP creation, where everything is cate- inflate production value, which cannot necessarily translate into the gorized from a manufacturing angle. In this case, total China market percentage of semiconductor content in a box produced in China. consumption is the value of all products that domestic consumers Another problem with this is that the demand and local supply met- buy from Chinese and overseas system brands, while total semicon- rics are not a 100% apples-to-apples comparison, making the final ductor production is the value produced by Chinese foundries for number susceptible to big fluctuations. domestic and overseas system brands.

With this method, the self-sufficiency ratio is around 12%, the lowest among the three.

Exhibit 47: Self-sufficiency based on GDP creation

Source: Morgan Stanley Research Note: China semi consumption = China consumption from China system brands + China consumption from overseas system brands; China semi production = China semi fabs' production for China system brands + China semi fabs' production for overseas system brands

Exhibit 48: Exhibit 49: Our estimates of China's semiconductor consumption at different Our estimates of China's self-sufficiency in semi components stages of the tech supply chain 70% 300 60% 243 250 50%

200 40% 30% 27% 150

(US$bn) (US$bn) 20% 16% 100 91 100 12% 10% 50 0% China factory self- China system brand self- China end-user self- - sufficiency % sufficiency % sufficiency % China factory China system brand China end-user consumption consumption consumption Source: Gartner, company data, Morgan Stanley Research estimates

Source: Gartner, IHS Markit, IDC, company data, Morgan Stanley Research estimates

34 M FOUNDATION China semi investment perspective

Exhibit 50: SWOT analysis for China's semiconductor industry

Source: Morgan Stanley Research

SWOT analysis c. Strong government support

Semiconductors have been an important focus of the Chinese gov- In this section we review the position of China's semiconductor ernment in recent years, with favorable local government policies as industry to present a clear picture of the potential risks and rewards well as the initiation of many SOE-funded investment portfolios such facing the industry. as the Big IC Fund. Another source of support is R&D subsidies, which have benefited companies like JCET and SMIC. It is not clear precisely 1. Strengths where China's focus will be in the next few years, but we believe strong government support will continue. a. Good infrastructure for tech development d. Large end market

With China’s leading research on 5G and existing infrastructure for With its 1.3 billion population, China is one of the most important new technology dissemination (high internet usage, high smart- regions for semiconductor content consumption, from smartphones phone penetration, acceptance of new technologies such as fintech), to PCs to servers. Its market size means China will be able to develop we believe that the country is a healthy environment for the semicon- many applications and semiconductor products for domestic and ductor industry. China's infrastructure development will create overseas use. The country's scale also means semiconductor compa- demand for more local semiconductor components and nurture nies can secure funding and support without leaving the country. technology advances to meet this need. e. Abundant domestic technology applications b. Low operating costs China is generally open to new technologies, which bodes well for China has been the factory of the world for several decades. semiconductor demand. Domestic applications have been growing Although labor costs have been rising in many prosperous cities, quickly, creating opportunities for local semiconductor companies semiconductor companies can still find affordable labor in more rural that are able to produce competitive products. areas. In addition, as the Bitcoin frenzy of 2017 and 2018 showed, electricity costs are low in many Chinese provinces.

MORGAN STANLEY RESEARCH 35 M FOUNDATION 2. Weaknesses Exhibit 51: Global semi companies' R&D ratio (%) vs. R&D expense (US$mn) 14,000 Intel (20.8%, a. Dependency on government support 12,000 13,035)

10,000

Chinese semiconductor companies, even some of the locally leading 8,000 ones, rely heavily on government support in the form of capital injec- 6,000 Qualcomm tions, favorable policies, and expense subsidies such as tax breaks. TSMC (8.3%, (24.6%, 5,504) 4,000 2,722) Applied (12.1%, Without these forms of support, Chinese companies would be less 1,814)

2017 2017 R&D expense(US$mn) 2,000 Nvidia (18.7%, Infineon (7.6%, competitive. TI (10.1%, 1,508) 1,769) - 646) 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 2017 R&D % b. Shorter history of semiconductor research Source: Thomson Reuters, Morgan Stanley Research

A critical factor for any semiconductor industry is developing talent – a process in which there are no shortcuts. It took Taiwan 30-40 Exhibit 52: years to build enough R&D resources and a sufficient talent pool to Chinese semi companies' R&D ratio (%) vs. R&D expense (US$mn) become a key player in the global supply chain. We believe China 450 SMIC (13.8%, 400 427) needs more time to accumulate industry know-how and build up its 350 industry more comprehensively. 300 250 200 c. Lack of a single industry cluster; resources overly Goodix (16.2%, 150 JCET (3.3%, 121) spread to provinces 92) 100 GigaDevice

2017 2017 R&D expense(US$mn) 50 (7.1%, 22) The resources being put into semiconductors are largely spread - across the whole industry, which we think dilutes the focus and slows 0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% 14.0% 16.0% 18.0% progress given the lack of critical IP and talent. In addition, several 2017 R&D % local governments are building their own semi funds, which again Source: Thomson Reuters, Morgan Stanley Research drags down the overall process since it is less productive to set up two SOEs that have the same ultimate goal. rent advantages for leading overseas foundries could narrow since leading-edge development has become more and more challenging, 3. Opportunities and China could potentially leverage this opportunity and catch up. If the gap does narrow, China would benefit by virtue of being one of the largest consumers of semiconductors. a. Disruptive semiconductor technology such as AI semi c. Collaboration with the global semiconductor supply chain Despite lagging overseas companies overall, China can still be a fast mover when any new field emerges. When cryptocurrency mining If globalization and the outsourcing of the semiconductor supply created new semiconductor demand in 2017, Chinese companies chain continue, we believe China's semi industry would gain from quickly developed dedicated ASIC chips to capture this demand, clearly identifying the correct niches and moving into them. China riding a wave of success until the Bitcoin price plunged in late 2018. would then be able to leverage its strengths to capture more market We believe other emerging technologies could create further oppor- share globally. tunities for China's semiconductor industry to catch up. d. Greater capital inflows into China b. Deceleration of Moore's Law Since China is one of the world's largest economies, we believe more Progress in traditional semiconductor manufacturing has slowed, as capital inflows into the industry from external partners are possible. evidenced by the deceleration of Moore's Law. That is to say, the cur- With more capital resources available, we think China would then be

36 M FOUNDATION able to purchase critical technologies from external partners, and foreign technology through acquisitions, it will slow the country's more quickly narrow the gap with global leaders. ability to catch up rapidly and may force semiconductor companies to rely on in-house R&D, which could take years to accumulate. 4. Threats c. Competition for talent from other industries, such as internet a. Critical IP and equipment are controlled by foreign companies Manufacturing was historically a popular choice of profession in Asia as it offered stable work hours and decent pay. However, as the shift Although the industry has been under development for more than to a world of services and the internet continues, the outflow of two decades, with fab capacity increasing, China lacks its own intel- talent to other industries will be a threat to the development of lectual property and manufacturing equipment. This puts the China's semiconductor industry. country in a difficult situation as it either has to seek licensing from overseas companies, gain IP and equipment via acquisitions or joint d. China's system brands prioritize foreign semi sup- ventures, or simply develop on its own from scratch and slow the pliers for better performance development of the semiconductor industry. The leaders in most subsegments of the semiconductor industry are b. Overseas M&A restrictions overseas companies with better-performing products. Therefore, although China has a lot of system brands and manufacturing M&A activity in the industry is becoming more complicated, with demand, local semi suppliers still face intense competition given that many aspects to consider. Although this issue is not specific to China, system brands usually prioritize foreign semi suppliers. it is a threat the country will have to face. If China is unable to acquire

Exhibit 53: Volatile memory patent distribution

Source: iRunway, Morgan Stanley Research

Exhibit 54: NAND Flash patent distribution

Source: iRunway, Morgan Stanley Research

MORGAN STANLEY RESEARCH 37 M FOUNDATION

Exhibit 55: China wafer fabs planned or under construction

Source: Gartner, company data, Morgan Stanley Research

38 M FOUNDATION Implications for global semis

Implications for memory and Asian tech TSMC, which are global top-tier companies, but rather second-tier players in the region such as UMC and Nanya Tech. In particular, (Shawn Kim) Taiwan’s semiconductor companies are relatively more focused on consumer electronics products, with sales heavily reliant on Chinese In a report last July, we indicated that China's share of manufacturing OEMs, making them more vulnerable to China's homegrown IC com- in nominal GDP (and in real terms) has decreased since the 2007-08 panies taking share. With the exception of TSMC and Samsung, many financial crisis, from 32% to a 2016 floor of 29%. To improve its pro- companies have not invested heavily in the software or IP side of the ductivity, China has a target that by 2020 manufacturers must make industry and in terms of R&D spend critical for innovation and long- notable progress to a higher level of digitalization. The central gov- term viability as the IC sector matures. ernment lists 'new generation information technology' as one of 10 strategic focus industries, with a particular focus on the internet, AI, …with a more level playing field in the Data Era. China appears to and semiconductors. China is therefore likely to continue to support be better positioned to have a larger impact on the next generation local semiconductor producers. of cognitive technologies. While the development of the internet of things, AI and autonomous vehicles will increase demand for semi- China challenging dominance by Korea and Taiwan. China is the conductors, competition from China is already here and the BATs are world’s biggest consumer of semiconductors but imports make up moving quickly with a greater presence in global markets. For new nearly 90%. It envisions upping its semiconductor localization level chips such as AI, China will not be playing catch-up, as it has done for to 70% from its current 10% by 2025. In chips alone, 40% of Korean decades with more conventional chips. Instead, its existing strength exports and over 50% of Taiwan's head to China. As such, a successful in AI and its unparalleled access to the quantities of data required to chip industry would make China more economically competitive and train AI algorithms could give it an edge in designing chips optimized independent, at the expense of existing chip suppliers in Korea and to run them. Baidu’s ‘Kunlun’ deep-learning chips and Huawei’s Kirin Taiwan. China is the world’s largest and fastest-growing market for 980 or in-house designed ‘Ascend’ AI chip are examples. In the field semiconductors, but given that no Chinese chipmaker has broken of communications, China is taking a lead role in setting the 5G stan- into the top 15 globally in terms of sales, the implication for incum- dard and combining communications with IoT to be a big market. In bent players is likely to be meaningful over time. China poses stra- China the demand for semiconductors is huge; coupled with security tegic risks to Korea's memory and Taiwan’s logic and back-end requirements, the market has become larger. We think China will be industries. This can already be seen in the ramp-up of Chinese pur- in the leading position in IoT, not just general IoT but also in the car chases of precision machinery for its factories. We believe the first networking industry. China is no longer drained of talent and China targets of Chinese semiconductors are not Samsung Electronics or has an edge in Internet-based innovation, which requires a large

Exhibit 56: Exhibit 57: China's semi equipment purchases ranked second in 2018 China's IC production accounts for only 5% of the total 80 (US$ bn) 30 6.0% 5.1% 70 Others Taiwan US Korea Japan Europe China 25 4.7% 5.0% 4.0% 3.8% 60 20 3.4% 3.5% 4.0% 3.0% 50 15 2.6% 3.0% 40 1.8% 2.0% 1.9% 1.9% 10 1.4% 2.0% 30 5 1.0% 20 0 0.0% 10 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

0 China IC Production (US$ bn) China IC Production as % of Total (RHS) 2015 2016 2017 2018e 2019e Source: IC Insight, Morgan Stanley Research Source: SEMI, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 39 M FOUNDATION number of users. This is different from low cost engineering for R&D Implications for Greater China semis and manufacturing. In 2017, China surpassed the US for the first time in global AI funding with 48% share, overtaking the US at 38% and (Charlie Chan, Daniel Yen) 13% for the rest. Recent news reports indicate that China has shifted away from its Korea will face increasing competition over market share. With 'Made In China 2025' policy to give foreign companies greater access the aim to improve self-sufficiency in chips from 35% in 2018 to 80% to its economy. Under the original policy, China aimed to increase its by 2030 or US$305bn, Korea appears most at risk in terms of market self-sufficiency in key components and materials from 40% in 2020 share. Telecommunications firms such as Huawei are moving to to 70% in 2025, reducing its reliance on foreign technology. The develop their own semiconductors to reduce dependence on foreign policy drew heavy criticism from international competitors given suppliers. YMT, along with two other Chinese firms, are planning on that heavy government subsidies would have been used to increase moving toward mass production of NAND this year as well. Both market share. In recent years, China has commoditized solar, LED, firms are investing in R&D to maintain a technological edge over and TFT-LCD products. Success in these areas may encourage China, while also reducing capital expenditures to prevent over- Chinese semiconductor companies to invest in more R&D to build up supply from further depressing prices. Samsung has also begun their own patents. We think US-China trade issues may mean short- moving into new growth markets, such as automotive chips, where term pain for China's local semi companies but long-term gain for its first new chip will power Audi’s infotainment systems from 2021. global semis.

Becoming a semiconductor powerhouse – Korea in 1988 vs. Tariffs are the most obvious feature of US-China trade tensions and China now. It took Samsung and SK Hynix decades to account for have caused some short-term semiconductor demand volatility roughly 75% of the DRAM market, with Samsung alone accounting owing to their effect on consumer confidence, inventory build, and for 44% of the market in 2018. However, to get to dominance took diversification of the production base to prevent further tariff hikes. a period of 10 years between 1994 and 2004. Before that, Japanese The impact of tariffs on the semiconductor supply chain is more on chipmakers (1984-1994) and US chipmakers (before 1984) domi- an operational level, and so far we haven't seen a big impact from nated the memory market. Similar to China today, Korea's dominance tariff issues on semi companies, because most of them already have was made possible by government policy (chaebol support, tax holi- production sites outside of China. days, and access to capital needed for costly fabs and long periods of losses), technical breakthroughs, and economic conditions – all of Over the long term, we are optimistic on China IC product companies which are present in China today. Longer term, the performance gap (such as Goodix) and design services (ASIC outsourcing, IP, EDA is shrinking as Moore’s Law becomes too expensive and of limited (Electronic design automation) tools). On manufacturing, we are benefit to a small set of global semiconductor companies. As a result, more positive on the scalability of backend foundries than wafer low-cost, lagging-edge Chinese technology companies will soon be foundries, whose profitability needs to improve. On memory, we able to address a larger part of the global market. think private company YMTC could become relevant in the NAND Flash industry by 2020.

40 M FOUNDATION Implications for US semis (Joe Moore, Exhibit 58: Global capex Global Capex (mn USD) Craig Hettenbach) 120,000

We expect following impacts to general purpose semis: 100,000

80,000 1. Discretes: Nexperia, acquired from NXP, could become a bigger competitive threat to leading power discrete companies (ON 60,000 Semiconductor) over time. Industry pricing for discretes has been 40,000 benign in the last two years due to a strong upcycle. But a normaliza- tion in the cycle and potential competitive pressure from China could 20,000 change this. 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019E NAND DRAM Other

2. NOR: Cypress Semiconductor has already repositioned its NOR Source: Company data, Morgan Stanley Research portfolio away from low density, focusing mostly on mid to high den- sity auto and industrial applications. If companies like Gigadevice was only one new entrant over that time, Intel, which was about 5% were to catch up at the mid and high end, that could pose a longer- of the total investment; the other 95% was primarily technology term competitive threat. transition. We believe that these businesses only work economically at large scale, given that the large R&D burden is not volume-de- 3. RF power amplifiers for wireless infrastructure: JAC Capital pendent. With multiple entrants in both types of memory, we esti- bought NXP’s RF power business and NXP retained the legacy mate that Chinese memory investment could reach US$15-20bn in Freescale RF power. China could represent a long-term risk to market spending by 2025, even if profits don't follow immediately. share and margins. Our long-term preference is for NAND exposure, for several rea- We are more positive than consensus on the growth outlook of sons: 1) NAND spending has already come down meaningfully for global semi cap names. While many of the fab announcements have 1H19 and investor expectations are even lower; 2) NAND will be the dollar amounts associated with them that appear inconsistent with largest push from China initially; 3) NAND is relatively immune from relatively small cleanroom size, it is clear that to be relevant on a 'EUV deflation', where EUV technology will reduce patterning TAM global scale is an enormous expense, and it is one of the most rele- for all markets but NAND; and 4) it drives a more consistent service vant drivers of semi cap profits. business. All of that leads to our preference for Lam over Applied Materials and Tokyo Electron (covered by Kazuo Yoshikawa) glo- As shown in Exhibit 58 , global memory capital spending in the last bally. Still, we acknowledge that Lam also faces the issues outlined five years has been US$171bn; over that time wafer fab level spending above. comprised 60% of total capital spending, or about US$102bn. There

MORGAN STANLEY RESEARCH 41 M FOUNDATION Implications for Japan semis (Kazuo Exhibit 59: Japan's semi cap equipment makers: Sales breakdown by customer Yoshikawa) device type

100% 90% 17% 21% 16% In the near term, amid ongoing trade tensions between the US 29% 80% 37% DRAM 14% 51% and China, we expect companies in China's semiconductor 70% 27% 14% supply chain to be incentivized to reduce their reliance on sour- 60% 50% 43% 29% cing from US firms, and to increase their procurement from 40% NVM 65% 70% 28% 30% 56% non-US suppliers: Indeed, we have found in talking to Japanese semi 20% 34% cap equipment and components makers that a number of those that 10% 28% 21% 0% Non- compete with US firms have expectations for increased business with Tokyo SCREEN Hitachi Advantest Tokyo Micronics memory Electron Holdings High-Tech Seimitsu Japan Chinese companies. (Needless to say, however, replacing wafer fab WFE Assembly & Test equipment is no easy matter; construction of wafer fab lines without Tokyo Electron, SCREEN and Hitachi High-Tech sales exclude field solution sales. Advantest sales break- the systems of US equipment makers Applied Materials, Lam down is based on test system sales. Micronics Japan sales breakdown is based on probe card sales. Source: Company data, Morgan Stanley Research Research and KLA-Tencor is impossible.) Within our coverage, Tokyo Electron benefits substantially from From this perspective, we believe Advantest is likely to be favorably NVM capex (since roughly 50% of its sales come from etching sys- positioned to win market share in an environment of trade tensions tems and deposition equipment) – although as our US team points between the US and China, since its leading competitor is the US firm out, Lam Research is more preferable than Applied Materials and Teradyne, and the time required for selecting SoC/memory testers Tokyo Electron in terms of China's NVM capex. companies is shorter than for wafer fab equipment companies. Under the New Breed scenario, leading-edge SoC design activity In three of the four scenarios outlined earlier in this report – by China's IC designers would likely burgeon over the longer Industry Upgrade, New Breed, Global Commoditization – we term: Brisker leading-edge SoC design activity would fuel prospec- expect capex by China to increase from 2019 (YMTC 3D-NAND tive demand expansion for photomask-related materials and produc- 64-layer process capacity investment is likely to come first): tion equipment. Almost all of the equipment makers in our coverage already have a track record of supplying equipment for the R&D lines of YMTC in Key Japanese players here include HOYA (semiconductor mask 2018, and should be well positioned to win business in the upcoming blanks), NuFlare Technology (photomask writing systems) and JEOL phase of investment in mass production. (manufactures and sells multi-beam photomask writing systems in partnership with IMS Nanofabrication).

42 M FOUNDATION Appendix 1: Potential breakthrough points for China semis Algorithm-driven chip design (software + Heterogeneous chip integration hardware integration) What comes after Moore’s Law? The Bitcoin mining chip is an example of software + hardware integra- tion. The circuits are designed and optimized to run the SHA-256 Heterogeneous integration is the combination of separately manu- algorithm efficiently. The Bitcoin mining chip vendor, Bitmain, gener- factured semiconductor components into one package using ated US$600mn of revenue in 2017 and was one of TSMC’s top three bumping, wirebonding, 3D stacking, and other assembly techniques customers at the peak of Bitcoin mining in 1H18. AI semiconductors to achieve optimal performance, power consumption, latency, and have many similarities with cryptomining chip designs. They are both cost reduction. Comparing to the race for the smallest transistor size algorithm-driven, while the circuit design (or hardware) itself isn't as in the Moore’s Law era, heterogeneous integration is more about difficult as a smartphone SoC. Therefore much of the development system design capabilities and packaging. and investment is at the software level, and requires integration between software and hardware. In the past few years since ITRS (International Technology Roadmap for Semiconductor) set the vision for future semiconductor improve- Exhibit 60: ment progress, leading semiconductor companies have frequently Horizon Robitics BPU design mentioned their dedication to heterogeneous integration. For example, TSMC, the leading foundry, has been aggressive in devel- oping 2.5D CoWoS, fan-out InFO, and 3D WoW packaging solutions while ASE, the leading OSAT, has been proving its capabilities to do complex SiP (system in package) projects.

By far, the most important milestone for heterogeneous integration is the introduction of the Apple Watch, assembled and tested by a coalition of ASE and USI (an EMS subsidiary of ASE). For the latest Apple Watch Series 4, the companies have to pack more than 700 different components into one single semiconductor package, including the main CPU, memory, passive components, RF items, and others. It has sold around 20mn units per year since introduction and remains the most important SiP or heterogeneous integration Source: Bitmain market. Exhibit 61: Horizon Robotics' chip for edge AI Other than creating small computer systems like Apple Watch and accommodating non-uniform shaped items like the various IoT, wear- able devices, another important implication for SiP technology is its time-to-market advantage. If we take another look at USI and its col- laboration with Qualcomm, we can find important advantages to using SiP. Looking at the breakdown of an application processor for smartphones (SoC; system-on-chip), we can see that most of the functions are done with one go at the foundry side with the most advanced node at the time. However, is it really necessary for every component to be at the most expensive node? On top of that, many carriers might need further qualifications for the same RF compo-

Source: Horizon Robotics nents because the application processor model is different. Because

MORGAN STANLEY RESEARCH 43 M FOUNDATION

Exhibit 62: Exhibit 63: ASE on SiP (System-in-Package) YMTC's visualized X-stacking technique

Source: YMTC

Source: ASE Commune IDM (CIDM) of this, going SiP will isolate components that need performance and power consumption improvements by going leading edge while Commune IDM is a business model promoted by Dr. Richard Chang keeping legacy ones at a more mature node to ensure fast turn- for China's semiconductor production industry to compete with around with repetitive qualification from carriers no longer required. global leaders. According to SiEn (Qingdao) Integrated Circuits Co., this sharing model allows IC design companies, end-product compa- YMTC X-stacking 3D NAND nies, and semiconductor manufacturers to integrate multiple parties through the establishment of a joint IDM company and enjoy shared benefits. Yangtze Memory Technologies (YMTC) has been developing for years its own advanced memory chips aiming to become a game- The CIDM model enables IC design companies to have exclusive pro- changer in the NAND industry. The X-stacking 3D NAND technique duction capacity and technical support for chip manufacturing. At the that YMTC announced in August 2018 is to put NAND and I/O arrays same time, the IC manufacturers can enjoy and share the market on two separate dies, which are then packaged together to become potential, resources, and risk. On the flipside, this CIDM model cre- a single chip. YMTC claims this approach would increase NAND bit ates more difficulties and challenges on partners' integration in the density which could actually achieve better cost per bit than compet- joint IDM company. For example, the IC manufacturer would have to itors, and would shorten the product development and manufac- allocate capacity resources appropriately to ensure all business part- turing cycle times by 20%. ners' demand is fulfilled, while the IC design participants could have conflicts of interest if competing in the same sub-sectors. However, some critics have suggested that the cost will be signifi- cantly higher given the double amount of wafers and additional The first Chinese CIDM company – SiEn Integrated Circuits – is cur- backend costs, so the actual adoption rate and success of the tech- rently targeted to enter mass production by the end of 2019. If this nology is yet to be proved. However, we still think this is another project turns out to be successful, we believe it could be a break- demonstration of new semiconductor techniques invented locally, through for China's semi industry given this could concentrate R&D which could potentially disrupt or become a breakthrough point for resources within the CIDM company and solve its weakness in terms China's semiconductor industry. of diversified resources.

44 M FOUNDATION Appendix 2: US/China/Taiwan semiconductor policy

Over the last 30 years the semiconductor industry has The PCAST thus believes that in order to maintain a strong and com- been driven by heavy sovereign investment that exac- petitive semiconductor industry, the US not only needs an economic erbated boom/bust cycles, and the next 5-10 years are and policy environment that ensures US's leadership in semiconduc- likely to repeat that history. Sovereign investment tors, but also needs to take protective action to respond directly to driven by national priorities in Japan, then Korea, then Chinese policies that violate trade rules and distort the global Taiwan characterized many of the boom/bust cycles within market. semis, each time driving a major investment cycle followed by multiple years of substantial oversupply. China's actions On November 19, 2018, the US Department of Commerce's Bureau to date look similar, though there are some differences – of Industry and Security published a document addressing its review the barriers to entry in memory were high during Taiwan's of controls for certain emerging technologies, including AI and push and seem even higher now after industry consolida- machine learning, microprocessor technologies, and cloud technolo- tion. Recent trade tension may serve as a double-edged gies. The document shows the government's intention to control the sword – it could either give China the motivation to export of advanced semiconductor components that could result in develop its semiconductor industry or it could cause pre- national security threats. vious efforts to stall. China semi policy trends US semi policy trends

China included semiconductor policy in its 12th and 13th Five-Year US caution has been rising since China increased its efforts to acquire Plans. Local governments also provide different levels of tax and manufacturing technologies abroad and narrow the gap between land benefits. itself and the world. Ensuring U.S. Leadership and Innovation in Semiconductors was published on January 6, 2017 by the President’s Several Chinese memory players have emerged, with overlapping Council of Advisors on Science and Technology (PCAST). The report ownership and incentive structures. Investment thus far has been assesses the challenges facing the semiconductor industry and their small from the memory makers, as generally the companies are consequences for the US economy and national security. The PCAST trying to establish viable pilot production before ramping up. believes innovation in semiconductors has slowed considering fun- damental technological limits, while a concerted push by China to reshape the market has increased the risks to maintaining its global leadership.

MORGAN STANLEY RESEARCH 45 M FOUNDATION

Exhibit 64: Local government semiconductor policies Overseas Government Fund size Favorable Investment Land & Credit line R&D subsidy M&A Tape-out IPR fund (Rmb$bn) rate incentive building support Beijing v v v v Shanghai v v v v v Shenzhen v3v vvvvvv Tianjing v 0.2 v v Hefei vv vv Nanjing v6 v v Wuxi 5 vv v v Wuhu v v vv Chonqing v 20 v vv Chengdu v v v Wuhan v 10 v v Changsha vv vv Xiamen v 50 v vv vv Jinjiang v 50 vvvvvvv Hanzhou v v v v

Source: Morgan Stanley Research

Exhibit 65: Important semiconductor policies in China since 2011

Source: TrendForce, Morgan Stanley Research

46 M FOUNDATION Taiwan semi policy trends Per the latest regulations, Taiwan requires prospective mainland Chinese investment in the semiconductor industry to be approved by the Investment Commission of the Ministry of Economic Affairs. Taiwan started the development of its semiconductor industry four Semi equipment, foundries, and OSAT are currently open to main- decades ago. Back then, the US welcomed both Taiwan and Japan to land Chinese investment, with the proviso that the investor submits develop semiconductor technology, for example through the RCA a strategic collaboration proposal and cannot have majority control program. Taiwan successfully established semiconductor foundries over the investment. Investments are subject to approval by the in the early 2000s, and this was followed by a boom in its IC design Investment Commission. IC design, however, is not open to mainland industry (Sunplus, Realtek, Novatek, MediaTek). Taiwan used to Chinese investment. Any exceptions to this must be approved the invest in the DRAM sector by licensing technology from Infineon, Executive Yuan. Elpida, and Micron. And although there were some lawsuits related to IPR, in general the US was open to Taiwan’s semi industry develop- ment because it made the US tech supply chain more comprehensive and efficient. Taiwan therefore accumulated talent in semiconduc- tors and companies like MediaTek built up IP.

Exhibit 66: Taiwan's current investment policy on mainland Chinese investment in the semiconductor industry Rules for mainland Chinese Subsegment Proviso investment

1. The investor must submit a proposal on strategic collaboration for the Investment Commission's approval Semi equipment Open with provisos 2. The investor cannot have controlling power* over the target

1. The investor must submit a proposal on strategic collaboration for the Investment Commission's approval IC Manufacturing Open with provisos 2. The investor cannot have controlling power over the target

1. The investor must submit a proposal on strategic collaboration for the Investment Commission's approval OSAT Open with provisos 2. The investor cannot have controlling power over the target

IC Design Not allowed

Discrete components Open

*Controlling power is defined as: 1) >30% holding; 2) >50% of board seats; 3) >50% of voting rights; or 4) Overarching control of strategy/direction in terms of finance, operations, and human resources

Source: Investment Commission, Ministry of Economic Affairs, Morgan Stanley Research

MORGAN STANLEY RESEARCH 47 M FOUNDATION Appendix 3: Exploring China's semi companies - our tour guide

How to use this guidebook

We provide profiles for a number of semiconductor subsegments, showing: l the competitive landscape and aggregate market share by country l profiles of Chinese and overseas groups in terms of growth, profitability, R&D spending, and end-market exposure l market share ranking of global players and their revenue size l headquarters of Chinese companies l other supplemental data.

As a guide, we summarize the most similar global semi companies and their industry peers in China.

Exhibit 67: Semi-similars vs. 'the new breed'

Company Ticker Product Global counterpart Taiwan counterpart Type

A-share Goodix 603160.SS IC design Synaptics Egis New Breed USI 601231.SS Module Murata New Breed JCET 600584.SS Backend foundry Amkor ASE GigaDevice 603986.SS NOR/MCU Cypress Macronix Sinyang 300236.SZ Chemical Sumco GWC Silan 600460.SS Power IC STMicro GMT NationZ 300077.SS Smartcard IC NXP H-share SMIC 0981.HK Foundry Globalfoundries UMC HHGrace 1347.HK Foundry Globalfoundries Vanguard Huada Semi 0085.HK Smartcard IC NXP US listed Alpha & Omega AOSL.O Power Discrete ON Semi ACM ACMR.O Equipment Applied Materials Taiwan listed Silergy 6415.TW IC design Texas Instrument MediaTek New Breed Focal Tech 3545.TW IC design Synaptics Novatek Private Bitmain IC design Bitfury New Breed Cambricon AI Semi Wave Computing New Breed Robotic Horizon AI Semi Wave Computing New Breed YMTC Memory IDM Micron Nanya Tech New Breed Zhaoxin x86 CPU Intel VIA New Breed HiSilicon IC design Qualcomm MediaTek Spreadtrum IC design Qualcomm MediaTek Amlogic IC design Actions Semi Rockchip IC design MediaTek Vimicro IC design Qualcomm MediaTek Actions Semi IC design Qualcomm MediaTek BCD Power IC Maxim Richtek San-An IC GaAs foundry Skyworks WIN Semi AMEC Equipment Applied Materials ChipONE IC design Skyworks Novatek Omnivision CMOS ON Semi Himax GalaxyCore CMOS ON Semi Himax SGIT Smart meter, charger Maxim Source: Morgan Stanley Research 48 M FOUNDATION #1 China's wafer foundries

Exhibit 68: Foundries: Competitive landscape Region China Taiwan Singapore Korea Japan USA SMIC TSMC* Samsung GlobalFoundries HuaHong UMC Dongbu XMC Vanguard MagnaChip CanSemi Powerchip Navtech WIN Semi AWSC Global market share 7% 70% 0% 10% 1% 10% *Global leader Source: Morgan Stanley Research estimates

Exhibit 69: Exhibit 70: Foundries: 2018 market share ranking Foundries: Industry profile Mainland Company Country Revenue (US$mn) Market Share Overseas 1 TSMC Taiwan 35,840 56% China 2 GlobalFoundries US 6,080 10% Revenue CAGR (2015-2019) 10% 7% 3 UMC Taiwan 5,440 9% 4 Samsung South Korea 4,928 8% Gross margin % 20% 40% 5 SMIC* China 3,456 5% R&D % of revenue 18% 10% 6TowerJazz Israel 1,536 2% 7 Powerchip Taiwan 1,152 2% Net margin % 3% 30% 8 Vanguard Taiwan 896 1% Domestic exposure 60% 10% 9 HuaHong* China 896 1% 10 Dongbu HiTek South Korea 768 1% Others 3,008 5% Total industry value 4,290 59,710

Source: IC Insights, Morgan Stanley Research estimates Source: Company data, Morgan Stanley Research estimates

Exhibit 71: Foundries: Locations of headquarters

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 49 M FOUNDATION

Exhibit 72: Exhibit 73: Foundries: China players' gross margin trend Foundries: 2018 market share 40 Others 5% HuaHong 1% Dongbu 1% 30 Vanguard 1% Powerchip 2% 20 TowerJazz 2%

10 SMIC 5%

-

(10) Samsung 8%

(20)

(30) Mar-05 Apr-06 May-07 Jun-08 Jul-09 Aug-10 Sep-11 Oct-12 Nov-13 Dec-14 Jan-16 Feb-17 Mar-18 TSMC 56% Foundry UMC 9% Source: Bloomberg, Morgan Stanley Research

GlobalFoundries 10%

Source: IC Insights, Morgan Stanley Research estimates

Exhibit 74: Summary of 12" fabs in China Year of initial Planned Capacity Location Company Codes Status Business production (kwpm) Memory Fabs JinJiang Fujian Jinhua Integrated Circuit - 2018 60 Planned Memory C2F 2018 80 Planned Memory Wuxi SK Hynix C2 2006 125 Completed Memory Hefei Hefei Changxin - 2018 125 Planned Memory Wuhan - 2019 300 Planned Memory YMTC Nanjing - 2019 100 Planned Memory Fab 68 2010 200 Completed Memory Dalian Intel - 2019 100 Planned Memory Xi'an Samsung China Fab 2014 120 Completed Memory Logic Fabs Hefei Nexchip Semiconductor Jinghe Fab 2017 40 Planned IDM/Foundry Wuxi Huahong Grace Semi - 2020 30 Planned Foundry Fab 2 2020 40 Planned Foundry Shanghai Shanhhai Huali Fab 2 2011 35 Completed Foundry Chengdu Globalfoundries Fab 11 2018 65 Planned Foundry SH 300mm phase 2/3 2018 70 Completed Foundry Shanghai SH 300mm Fab 1 2007 25 Completed Foundry BJ 300mm Fab 1 2004 50 Completed Foundry SMIC Beijing BJ 300mm Fab 2 p1 2015 35 Completed Foundry BJ 300mm Fab 2 p2 2018 35 Planned Foundry Shenzhen SZ 300mn Fab 2017 40 Planned Foundry Qingdao SiEN Fab 2 Phase 1 2019 40 Planned CIDM Nanjing TSMC TSMC (Nanjing) 2018 20 Planned Foundry Xiamen UMC Fab 12X 2016 50 Completed Foundry Wuhan XMC Fab A/Fab B 2008 60 Completed Foundry

Source: Gartner, Morgan Stanley Research

50 M FOUNDATION #2 China's backend foundries

Exhibit 75: OSATs: Competitive landscape

Region China Taiwan Singapore Korea Japan USA JCET ASE* UTAC Amkor TFME Powertech Tianshui Chipbond KYEC ChipMOS

Global market share 24% 50% 5% <5% <5% 15% *Global leader Source: Morgan Stanley Research estimates

Exhibit 76: Exhibit 77: OSATs: 2018 market share ranking OSATs: Industry profile Mainland Company Country Revenue (US$mn) Market Share Overseas China 1 ASE Taiwan 7,890 28% 2 Amkor US 4,060 15% Revenue CAGR (2015-2019) 20% 5% 3 JCET* China 3,200 12% Gross margin % 12% 20% 4 Powertech Taiwan 1,890 7% R&D % of revenue 4% 4% 5 Tianshui* China 1,060 4% 6 TFME* China 910 4% Net margin % 1% 8% 7 KYEC Taiwan 675 3% Domestic exposure 80% <50% 8 UTAC Singapore 673 3% 9 ChipMOS Taiwan 600 2% 10 Chipbond Taiwan 580 2% Total industry value 7,018 21,982 Others 7,462 26% Source: Company data, Morgan Stanley Research estimates *Chinese player

Source: Trendforce, Morgan Stanley Research estimates

Exhibit 78: OSATs: Locations of headquarters

Nantong Fujitsu (Nantong)

JCET (Jiangsu) Tianshui Huatien (Tianshui)

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 51 M FOUNDATION

Exhibit 79: Exhibit 80: OSATs: China players' gross margin trend OSATs: 2018 market share 30

Others 26% 25 ASE 28%

20

15

10

5 Chipbond2% ChipMOS 2% - Mar-05 Apr-06 May-07 Jun-08 Jul-09 Aug-10 Sep-11 Oct-12 Nov-13 Dec-14 Jan-16 Feb-17 Mar-18 UTAC 3% OSAT

Source: Bloomberg, Morgan Stanley Research KYEC 3% Amkor 15% TFME 4%

Tianshui 4%

Powertech 7% JCET 12% Source: Trendforce, Morgan Stanley Research estimates

52 M FOUNDATION #3 China memory

YMTC is the Chinese memory company that owns the IP for 3D Alternatively, China can run a foundry business model for memory, NAND, licensed from Cypress Semiconductor. The company is con- just like the Inotera JV of Formosa Group and Micron. Outside of structing a new fab to produce 3D NAND, but we think more effective DRAM, we see an opportunity in NAND, because China acquired an IP capacity will come in 2020 when the company mass produces its 96L license from Cypress four years ago, before trade tensions started. If 3D NAND. In 2018, we think the total capacity for 3D NAND was only China were to build up a competitive NAND fab, it could potentially 5k wafers per month for 64L 3D NAND, and this could expand to 20k disrupt the current NAND supply chain. in 2019. China has built up local OSAT supply, led by JCET. China has also On DRAM, Innotron (which is funded by Gigadevice) is developing developed several strong IC design companies, such as HiSilicon for 19nm DRAM technology. But without a big R&D team for support, we smartphone processors and SoC. Bitmain is the industry leader for think the capacity could be only 1-2 k wafers per month. We also think mining ASICs, while Goodix is monetizing fingerprint sensors. UMC/Jinghua's 25nm DRAM technology development is quite behind. Overall, we don't think China memory fabs will move the needle on memory supply in 2019.

Exhibit 81: Memory projects in China

Invested Company Products Capital Capacity progress and plans Key investors (US$bn)

China Big IC Fund, TsingHua Unigroup, Hubei IC Fund, Hubei YMTC NAND, DRAM >24bn - Mass production in 2020 Science Technology Investment Group - Risk production in late 2018; monthly capacity 60k Fujian Electronics and Information Group, Jinjiang Energy Fujian Jinhua Specialty DRAM 5.7bn - Monthly capacity to reach 240k in 2025 Investment Co. - Mass production by end of 2018 Innotron Memory 8GB DDR4/LPDDR4 7.2bn Hefei Industry Investment Group - Monthly capacity to reach 20k in 2019

Source: Digitimes, Morgan Stanley Research

Exhibit 82: Memory IDM: Competitive landscape

Region China Taiwan Singapore Korea Japan USA YMTC (NAND) Nanya Tech (DRAM) Samsung (DRAM, NAND)* Toshiba (NAND) Micron (DRAM, NAND) GigaDevice (NOR) Macronix (NOR) SK Hynix (DRAM, NAND) Cypress (NOR) Winbond (DRAM, NOR) Powerchip

Global market share 0% 2% 0% 80% 3% 15% *Global leader Source: Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 53 M FOUNDATION

Exhibit 83: Exhibit 86: Memory IDM: 2018 market share ranking Memory IDM: 2018 market share Powerchip 1% Company Country Revenue (US$mn) Market Share Winbond 1% Macronix 1% 1 Samsung South Korea 64,418 43% GigaDevice 0% Cypress 2% 2 SK Hynix South Korea 35,995 24% 3 Micron US 30,383 20% Nanya Tech 2% Toshiba 7% 4 Toshiba Japan 10,041 7% 5 Nanya Tech Taiwan 2,778 2% 6 Cypress US 2,471 2% 7 Winbond Taiwan 1,700 1% Samsung 43% Micron 20% 8 Macronix Taiwan 1,226 1% 9 Powerchip Taiwan 876 1% 10 GigaDevice* China 355 0% *Chinese player

Source: Gartner, Morgan Stanley Research estimates

Exhibit 84:

Memory IDM: Industry profile SK Hynix 24%

Mainland Source: Gartner, Morgan Stanley Research estimates Overseas China Revenue CAGR (2015-2019) 25% 7% Gross margin % 38% 60% R&D % of revenue 9% 7% Net margin % 20% 40% Domestic exposure 95% 10%

Total industry value 400 130,000

Source: Company data, Morgan Stanley Research estimates

Exhibit 85: Memory IDM: Location of headquarters

Source: Morgan Stanley Research

54 M FOUNDATION #4 China IC design

Exhibit 87: IC design: Competitive landscape

Region China Taiwan Singapore Korea Japan USA Hisilicon MediaTek Broadcom Samsung Socionext Qualcomm Spreadtrum Novatek MegaChips Nvidia Bitmain Realtek Apple Goodix Himax AMD Unigroup Xilinx NationZ Marvell Silergy Global market share 5% 14% 6% 3% 4% 55%

Source: Morgan Stanley Research estimates

Exhibit 88: Exhibit 89: IC design: 2018 Market share ranking IC design: Industry profile Mainland Company Country Revenue (US$mn) Market Share Overseas 1 Qualcomm US 17,082 6% China 2 Broadcom US 16,176 6% Revenue CAGR (2015-2019) 20% 5% 3 Nvidia US 9,402 3% 4 MediaTek Taiwan 7,838 3% Gross margin % 30% 30% 5 Apple US 6,960 3% R&D % of revenue 15% 15% 6 AMD US 5,329 2% 7 Hisilicon* China 4,715 2% Net margin % 20% 20% 8 Xilinx US 2,475 1% Domestic exposure 70% 10% 9 Marvell US 2,394 1% 10 Unigroup* China 2,050 1% Others 195,579 72% Total industry value 13,860 256,140 *Chinese player Source: Company data, Morgan Stanley Research estimates Source: IC Insights, Gartner, Morgan Stanley Research estimates

Exhibit 90: IC design: Locations of headquarters

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 55 M FOUNDATION

Exhibit 91: Exhibit 92: IC design: China players' gross margin trend IC design: 2018 Market share 50

45

40

35

30

25

20

15

10

5

- Mar-05 Apr-06 May-07 Jun-08 Jul-09 Aug-10 Sep-11 Oct-12 Nov-13 Dec-14 Jan-16 Feb-17 Mar-18

IC Design

Source: Bloomberg, Morgan Stanley Research

Source: IC Insights, Gartner, Morgan Stanley Research estimates

56 M FOUNDATION Share Prices

SMIC (0981.HK): HK$8.50 Tokyo Electron (8035.T): JPY$14,960

JCET (600584.SS): Rmb$15.61 HOYA (7741.T): JPY$7,054

USI (601231.SS): Rmb$13.05 NuFlare (6256.T) JPY$5,930

Goodix (603160.SS): Rmb$101.31 Applied Materials (AMAT.O): US$38.51

GigaDevice(603986.SS): Rmb$108.11 Lam Research (LRCX.O): US$172.01

Silergy (6415.TW): NT$517.00 Intel (INTC.O): US$53.35

TSMC (2330.TW): NT$ 230.50 Xilinx (XLNX.O): US$121.94

UMC (2303.TW): NT$11.45 Texas Instruments (TXN.O): US$106.45

Vanguard (5347.TWO): NT$65.50 NVIDIA (NVDA.O): US$161.14

ASE Holding (3711.TW): NT$62.20 Micron (MU.O): US$39.03

Nanya Tech (2408.TW): NT$60.40 AMD (AMD.O): US$22.96

GUC (3443.TW): NT$201.00 Analog Devices (ADI.O): US$107.68

GlobalWafers (6488.TWO): NT$303.50 Cypress (CY.O): US$15.13

Samsung Electronics (005930.KS): KW$43,650 ON Semi (ON.O): US$22.61

SK Hynix (000660.KS): KW$66,600 MediaTek (2454.TW): NT$270

Share prices as of March 11, 2019

MORGAN STANLEY RESEARCH 57 M FOUNDATION Disclosure Section

The information and opinions in Morgan Stanley Research were prepared or are disseminated by Morgan Stanley Asia Limited (which accepts the responsibility for its contents) and/or Morgan Stanley Asia (Singapore) Pte. (Registration number 199206298Z) and/or Morgan Stanley Asia (Singapore) Securities Pte Ltd (Registration number 200008434H), regulated by the Monetary Authority of Singapore (which accepts legal responsibility for its contents and should be contacted with respect to any matters arising from, or in connection with, Morgan Stanley Research), and/or Morgan Stanley Taiwan Limited and/or Morgan Stanley & Co International plc, Seoul Branch, and/or Morgan Stanley Australia Limited (A.B.N. 67 003 734 576, holder of Australian financial services license No. 233742, which accepts responsibility for its contents), and/or Morgan Stanley Wealth Management Australia Pty Ltd (A.B.N. 19 009 145 555, holder of Australian financial services license No. 240813, which accepts responsibility for its contents), and/or Morgan Stanley India Company Private Limited, regulated by the Securities and Exchange Board of India (“SEBI”) and holder of licenses as a Research Analyst (SEBI Registration No. INH000001105); Stock Broker (BSE Registration No. INB011054237 and NSE Registration No. INB/INF231054231), Merchant Banker (SEBI Registration No. INM000011203), and depository participant with National Securities Depository Limited (SEBI Registration No. IN-DP-NSDL-372-2014) which accepts the responsi- bility for its contents and should be contacted with respect to any matters arising from, or in connection with, Morgan Stanley Research, and/or PT. Morgan Stanley Sekuritas Indonesia and their affiliates (collectively, "Morgan Stanley").

For important disclosures, stock price charts and equity rating histories regarding companies that are the subject of this report, please see the Morgan Stanley Research Disclosure Website at www.morganstanley.com/researchdisclosures, or contact your investment representative or Morgan Stanley Research at 1585 Broadway, (Attention: Research Management), New York, NY, 10036 USA.

For valuation methodology and risks associated with any recommendation, rating or price target referenced in this research report, please contact the Client Support Team as follows: US/Canada +1 800 303-2495; Hong Kong +852 2848-5999; Latin America +1 718 754-5444 (U.S.); London +44 (0)20-7425-8169; Singapore +65 6834-6860; Sydney +61 (0)2-9770-1505; Tokyo +81 (0)3-6836-9000. Alternatively you may contact your investment representative or Morgan Stanley Research at 1585 Broadway, (Attention: Research Management), New York, NY 10036 USA. Analyst Certification

The following analysts hereby certify that their views about the companies and their securities discussed in this report are accurately expressed and that they have not received and will not receive direct or indirect compensation in exchange for expressing specific recommendations or views in this report: Charlie Chan; Craig Hettenbach; Shawn Kim; Joseph Moore; Daniel Yen, CFA; Kazuo Yoshikawa, CFA.

Unless otherwise stated, the individuals listed on the cover page of this report are research analysts. Global Research Conflict Management Policy

Morgan Stanley Research has been published in accordance with our conflict management policy, which is available at www.morganstanley.com/institutional/research/conflictpolicies. A Portuguese version of the policy can be found at www.morganstanley.com.br Important US Regulatory Disclosures on Subject Companies

As of February 28, 2019, Morgan Stanley beneficially owned 1% or more of a class of common equity securities of the following companies covered in Morgan Stanley Research: ASE Technology Holding Co. Ltd., ASMedia Technology Inc, Chipbond Technology Corp, Egis Technology Inc, Lam Research Corp, MediaTek, Novatek, ON Semiconductor Corp., Parade Technologies Ltd, Phison Electronics Corp, Realtek Semiconductor, Silergy Corp., UMC, Vanguard International Semiconductor, Xilinx.

Within the last 12 months, Morgan Stanley managed or co-managed a public offering (or 144A offering) of securities of Micron Technology Inc., Texas Instruments.

Within the last 12 months, Morgan Stanley has received compensation for investment banking services from Cypress Semiconductor Corp, Micron Technology Inc., NVIDIA Corp., SK Hynix, Texas Instruments.

In the next 3 months, Morgan Stanley expects to receive or intends to seek compensation for investment banking services from Advanced Micro Devices, Analog Devices Inc., Applied Materials Inc., ASE Technology Holding Co. Ltd., Cypress Semiconductor Corp, HOYA, Intel Corporation, Lam Research Corp, MediaTek, Micron Technology Inc., Nanya Technology Corp., NuFlare Technology, NVIDIA Corp., ON Semiconductor Corp., Samsung Electronics, Silicon Motion, SK Hynix, SMIC, Te Connectivity Ltd, Texas Instruments, Tokyo Electron, UMC, Xilinx.

Within the last 12 months, Morgan Stanley has received compensation for products and services other than investment banking services from Advanced Micro Devices, Analog Devices Inc., Applied Materials Inc., ASE Technology Holding Co. Ltd., Cypress Semiconductor Corp, Intel Corporation, MediaTek, Micron Technology Inc., Nanya Technology Corp., Novatek, NVIDIA Corp., ON Semiconductor Corp., Realtek Semiconductor, Silicon Motion, SMIC, Te Connectivity Ltd, Texas Instruments, Tokyo Electron, UMC, Winbond Electronics Corp.

Within the last 12 months, Morgan Stanley has provided or is providing investment banking services to, or has an investment banking client relationship with, the following company: Advanced Micro Devices, Analog Devices Inc., Applied Materials Inc., ASE Technology Holding Co. Ltd., Cypress Semiconductor Corp, HOYA, Intel Corporation, Lam Research Corp, MediaTek, Micron Technology Inc., Nanya Technology Corp., NVIDIA Corp., ON Semiconductor Corp., Samsung Electronics, Silicon Motion, SK Hynix, SMIC, Te Connectivity Ltd, Texas Instruments, Tokyo Electron, UMC, Xilinx.

Within the last 12 months, Morgan Stanley has either provided or is providing non-investment banking, securities-related services to and/or in the past has entered into an agreement to provide services or has a client relationship with the following company: Advanced Micro Devices, Analog Devices Inc., Applied Materials Inc., ASE Technology Holding Co. Ltd., Cypress Semiconductor Corp, Intel Corporation, Macronix International Co Ltd, MediaTek, Micron Technology Inc., Nanya Technology Corp., Novatek, NVIDIA Corp., ON Semiconductor Corp., Realtek Semiconductor, Silicon Motion, SMIC, Te Connectivity Ltd, Texas Instruments, Tokyo Electron, UMC, Winbond Electronics Corp, Xilinx.

58 M FOUNDATION Morgan Stanley & Co. LLC makes a market in the securities of Advanced Micro Devices, Analog Devices Inc., Applied Materials Inc., ASE Technology Holding Co. Ltd., Cypress Semiconductor Corp, Intel Corporation, Lam Research Corp, Micron Technology Inc., NVIDIA Corp., ON Semiconductor Corp., Silicon Motion, Te Connectivity Ltd, Texas Instruments, TSMC, UMC, Xilinx.

The equity research analysts or strategists principally responsible for the preparation of Morgan Stanley Research have received compensation based upon various factors, including quality of research, investor client feedback, stock picking, competitive factors, firm revenues and overall investment banking revenues. Equity Research analysts' or strategists' compensation is not linked to investment banking or capital markets transactions performed by Morgan Stanley or the profitability or revenues of particular trading desks.

Morgan Stanley and its affiliates do business that relates to companies/instruments covered in Morgan Stanley Research, including market making, providing liquidity, fund management, commercial banking, extension of credit, investment services and investment banking. Morgan Stanley sells to and buys from customers the securities/instruments of companies covered in Morgan Stanley Research on a principal basis. Morgan Stanley may have a position in the debt of the Company or instruments discussed in this report. Morgan Stanley trades or may trade as principal in the debt securities (or in related derivatives) that are the subject of the debt research report.

Certain disclosures listed above are also for compliance with applicable regulations in non-US jurisdictions. STOCK RATINGS

Morgan Stanley uses a relative rating system using terms such as Overweight, Equal-weight, Not-Rated or Underweight (see definitions below). Morgan Stanley does not assign ratings of Buy, Hold or Sell to the stocks we cover. Overweight, Equal-weight, Not-Rated and Underweight are not the equivalent of buy, hold and sell. Investors should carefully read the definitions of all ratings used in Morgan Stanley Research. In addition, since Morgan Stanley Research contains more complete information concerning the analyst's views, investors should carefully read Morgan Stanley Research, in its entirety, and not infer the contents from the rating alone. In any case, ratings (or research) should not be used or relied upon as investment advice. An investor's decision to buy or sell a stock should depend on individual circumstances (such as the investor's existing holdings) and other considerations. Global Stock Ratings Distribution

(as of February 28, 2019)

The Stock Ratings described below apply to Morgan Stanley's Fundamental Equity Research and do not apply to Debt Research produced by the Firm.

For disclosure purposes only (in accordance with NASD and NYSE requirements), we include the category headings of Buy, Hold, and Sell alongside our ratings of Overweight, Equal-weight, Not-Rated and Underweight. Morgan Stanley does not assign ratings of Buy, Hold or Sell to the stocks we cover. Overweight, Equal-weight, Not-Rated and Underweight are not the equivalent of buy, hold, and sell but represent recommended relative weightings (see definitions below). To satisfy regulatory requirements, we correspond Overweight, our most positive stock rating, with a buy recommendation; we correspond Equal-weight and Not-Rated to hold and Underweight to sell recommendations, respectively.

Other Material Investment Services Clients Coverage Universe Investment Banking Clients (IBC) (MISC) Stock Rating Count % of Total Count % of Total IBC % of Rating Category Count % of Total Other MISC Category Overweight/Buy 1115 36% 296 42% 27% 526 38% Equal-weight/Hold 1379 44% 325 46% 24% 639 46% Not-Rated/Hold 46 1% 6 1% 13% 6 0% Underweight/Sell 585 19% 82 12% 14% 225 16% Total 3,125 709 1396

Data include common stock and ADRs currently assigned ratings. Investment Banking Clients are companies from whom Morgan Stanley received investment banking compensation in the last 12 months. Due to rounding off of decimals, the percentages provided in the "% of total" column may not add up to exactly 100 percent. Analyst Stock Ratings

Overweight (O). The stock's total return is expected to exceed the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Equal-weight (E). The stock's total return is expected to be in line with the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Not-Rated (NR). Currently the analyst does not have adequate conviction about the stock's total return relative to the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Underweight (U). The stock's total return is expected to be below the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Unless otherwise specified, the time frame for price targets included in Morgan Stanley Research is 12 to 18 months.

MORGAN STANLEY RESEARCH 59 M FOUNDATION Analyst Industry Views

Attractive (A): The analyst expects the performance of his or her industry coverage universe over the next 12-18 months to be attractive vs. the relevant broad market benchmark, as indicated below.

In-Line (I): The analyst expects the performance of his or her industry coverage universe over the next 12-18 months to be in line with the relevant broad market benchmark, as indicated below.

Cautious (C): The analyst views the performance of his or her industry coverage universe over the next 12-18 months with caution vs. the relevant broad market benchmark, as indicated below.

Benchmarks for each region are as follows: North America - S&P 500; Latin America - relevant MSCI country index or MSCI Latin America Index; Europe - MSCI Europe; Japan - TOPIX; Asia - relevant MSCI country index or MSCI sub-regional index or MSCI AC Asia Pacific ex Japan Index. Stock Price, Price Target and Rating History (See Rating Definitions)

60 M FOUNDATION

MORGAN STANLEY RESEARCH 61 M FOUNDATION Important Disclosures for Morgan Stanley Smith Barney LLC Customers

Important disclosures regarding the relationship between the companies that are the subject of Morgan Stanley Research and Morgan Stanley Smith Barney LLC or Morgan Stanley or any of their affiliates, are available on the Morgan Stanley Wealth Management disclosure website at www.morganstanley.com/online/researchdisclosures. For Morgan Stanley specific disclosures, you may refer to www.morganstanley.com/researchdisclosures.

Each Morgan Stanley Equity Research report is reviewed and approved on behalf of Morgan Stanley Smith Barney LLC. This review and approval is conducted by the same person who reviews the Equity Research report on behalf of Morgan Stanley. This could create a conflict of interest. Other Important Disclosures

Morgan Stanley & Co. International PLC and its affiliates have a significant financial interest in the debt securities of Advanced Micro Devices, Analog Devices Inc., Applied Materials Inc., ASM Pacific, Cypress Semiconductor Corp, Intel Corporation, Lam Research Corp, Micron Technology Inc., NVIDIA Corp., ON Semiconductor Corp., Samsung Electronics, Texas Instruments, Xilinx.

Morgan Stanley Research policy is to update research reports as and when the Research Analyst and Research Management deem appropriate, based on developments with the issuer, the sector, or the market that may have a material impact on the research views or opinions stated therein. In addition, certain Research publications are intended to be updated on a regular periodic basis (weekly/monthly/quarterly/annual) and will ordinarily be updated with that frequency, unless the Research Analyst and Research Management determine that a different publication schedule is appropriate based on current conditions.

Morgan Stanley is not acting as a municipal advisor and the opinions or views contained herein are not intended to be, and do not constitute, advice within the meaning of Section 975 of the Dodd-Frank Wall Street Reform and Consumer Protection Act.

Morgan Stanley produces an equity research product called a "Tactical Idea." Views contained in a "Tactical Idea" on a particular stock may be contrary to the recommendations or views expressed in research on the same stock. This may be the result of differing time horizons, methodologies, market events, or other factors. For all research available on a particular stock, please contact your sales representative or go to Matrix at http://www.morganstanley.com/matrix.

Morgan Stanley Research is provided to our clients through our proprietary research portal on Matrix and also distributed electronically by Morgan Stanley to clients. Certain, but not all, Morgan Stanley Research products are also made available to clients through third-party vendors or redistributed to clients through alternate electronic means as a convenience. For access to all available Morgan Stanley Research, please contact your sales representative or go to Matrix at http://www.morganstanley.com/matrix.

Any access and/or use of Morgan Stanley Research is subject to Morgan Stanley's Terms of Use (http://www.morganstanley.com/terms.html). By accessing and/or using Morgan Stanley Research, you are indicating that you have read and agree to be bound by our Terms of Use (http://www.morganstanley.com/terms.html). In addition you consent to Morgan Stanley processing your personal data and using cookies in accordance with our Privacy Policy and our Global Cookies Policy (http://www.morganstanley.com/privacy_pledge.html), including for the purposes of setting your preferences and to collect readership data so that we can deliver better and more personalized service and products to you. To find out more information about how Morgan Stanley processes personal data, how we use cookies and how to reject cookies see our Privacy Policy and our Global Cookies Policy (http://www.morganstanley.com/privacy_pledge.html).

If you do not agree to our Terms of Use and/or if you do not wish to provide your consent to Morgan Stanley processing your personal data or using cookies please do not access our research.

Morgan Stanley Research does not provide individually tailored investment advice. Morgan Stanley Research has been prepared without regard to the circumstances and objectives of those who receive it. Morgan Stanley recommends that investors independently evaluate particular investments and strategies, and encourages investors to seek the advice of a financial adviser. The appropriateness of an investment or strategy will depend on an investor's circumstances and objectives. The securities, instruments, or strategies discussed in Morgan Stanley Research may not be suitable for all investors, and certain investors may not be eligible to purchase or participate in some or all of them. Morgan Stanley Research is not an offer to buy or sell or the solicitation of an offer to buy or sell any security/instrument or to participate in any particular trading strategy. The value of and income from your investments may vary because of changes in interest rates, foreign exchange rates, default rates, prepayment rates, securities/instruments prices, market indexes, operational or financial conditions of companies or other factors. There may be time limitations on the exercise of options or other rights in securities/instruments transactions. Past performance is not necessarily a guide to future performance. Estimates of future performance are based on assumptions that may not be realized. If provided, and unless otherwise stated, the closing price on the cover page is that of the primary exchange for the subject company's securities/instruments.

The fixed income research analysts, strategists or economists principally responsible for the preparation of Morgan Stanley Research have received compensation based upon various factors, including quality, accuracy and value of research, firm profitability or revenues (which include fixed income trading and capital markets profitability or revenues), client feedback and competitive factors. Fixed Income Research analysts', strategists' or economists' compensation is not linked to investment banking or capital markets transactions performed by Morgan Stanley or the profitability or revenues of particular trading desks.

The "Important US Regulatory Disclosures on Subject Companies" section in Morgan Stanley Research lists all companies mentioned where Morgan Stanley owns 1% or more of a class of common equity securities of the companies. For all other companies mentioned in Morgan Stanley Research, Morgan Stanley may have an investment of less than 1% in securities/instruments or derivatives of securities/instruments of companies and may trade them in ways different from those discussed in Morgan Stanley Research. Employees of Morgan Stanley not involved in the preparation of Morgan Stanley Research may have investments in securities/instruments or derivatives of securities/instruments of companies mentioned and may trade them in ways different from those discussed in Morgan Stanley Research. Derivatives may be issued by Morgan Stanley or associated persons.

With the exception of information regarding Morgan Stanley, Morgan Stanley Research is based on public information. Morgan Stanley makes every effort to use reliable, comprehensive information, but we make no representation that it is accurate or complete. We have no obligation to tell you when opinions or information in Morgan Stanley Research change apart from when we intend to discontinue equity research coverage of a subject company. Facts and views presented in Morgan Stanley Research have not been reviewed by, and may not reflect information known to, professionals in other Morgan Stanley business areas, including investment banking personnel.

Morgan Stanley Research personnel may participate in company events such as site visits and are generally prohibited from accepting payment by the company of associated expenses unless pre-approved by authorized members of Research management.

62 M FOUNDATION Morgan Stanley may make investment decisions that are inconsistent with the recommendations or views in this report.

To our readers based in Taiwan or trading in Taiwan securities/instruments: Information on securities/instruments that trade in Taiwan is distributed by Morgan Stanley Taiwan Limited ("MSTL"). Such information is for your reference only. The reader should independently evaluate the investment risks and is solely responsible for their investment decisions. Morgan Stanley Research may not be distributed to the public media or quoted or used by the public media without the express written consent of Morgan Stanley. Any non-customer reader within the scope of Article 7-1 of the Taiwan Stock Exchange Recommendation Regulations accessing and/or receiving Morgan Stanley Research is not permitted to provide Morgan Stanley Research to any third party (including but not limited to related parties, affiliated companies and any other third parties) or engage in any activities regarding Morgan Stanley Research which may create or give the appearance of creating a conflict of interest. Information on securities/instruments that do not trade in Taiwan is for informational purposes only and is not to be construed as a recommendation or a solicitation to trade in such securities/instruments. MSTL may not execute transactions for clients in these securities/instruments.

Certain information in Morgan Stanley Research was sourced by employees of the Shanghai Representative Office of Morgan Stanley Asia Limited for the use of Morgan Stanley Asia Limited.

Morgan Stanley is not incorporated under PRC law and the research in relation to this report is conducted outside the PRC. Morgan Stanley Research does not constitute an offer to sell or the solicitation of an offer to buy any securities in the PRC. PRC investors shall have the relevant qualifications to invest in such securities and shall be responsible for obtaining all relevant approvals, licenses, verifications and/or registrations from the relevant governmental authorities themselves. Neither this report nor any part of it is intended as, or shall constitute, provision of any consultancy or advisory service of securities investment as defined under PRC law. Such information is provided for your reference only.

Morgan Stanley Research is disseminated in Brazil by Morgan Stanley C.T.V.M. S.A. located at Av. Brigadeiro Faria Lima, 3600, 6th floor, São Paulo - SP, Brazil; and is regulated by the Comissão de Valores Mobiliários; in Mexico by Morgan Stanley México, Casa de Bolsa, S.A. de C.V which is regulated by Comision Nacional Bancaria y de Valores. Paseo de los Tamarindos 90, Torre 1, Col. Bosques de las Lomas Floor 29, 05120 Mexico City; in Japan by Morgan Stanley MUFG Securities Co., Ltd. and, for Commodities related research reports only, Morgan Stanley Capital Group Japan Co., Ltd; in Hong Kong by Morgan Stanley Asia Limited (which accepts responsibility for its contents) and by Morgan Stanley Asia International Limited, Hong Kong Branch; in Singapore by Morgan Stanley Asia (Singapore) Pte. (Registration number 199206298Z) and/or Morgan Stanley Asia (Singapore) Securities Pte Ltd (Registration number 200008434H), regulated by the Monetary Authority of Singapore (which accepts legal responsibility for its contents and should be contacted with respect to any matters arising from, or in connection with, Morgan Stanley Research) and by Morgan Stanley Asia International Limited, Singapore Branch (Registration number T11FC0207F); in Australia to "wholesale clients" within the meaning of the Australian Corporations Act by Morgan Stanley Australia Limited A.B.N. 67 003 734 576, holder of Australian financial services license No. 233742, which accepts responsibility for its contents; in Australia to "wholesale clients" and "retail clients" within the meaning of the Australian Corporations Act by Morgan Stanley Wealth Management Australia Pty Ltd (A.B.N. 19 009 145 555, holder of Australian financial services license No. 240813, which accepts responsibility for its contents; in Korea by Morgan Stanley & Co International plc, Seoul Branch; in India by Morgan Stanley India Company Private Limited; in Indonesia by PT. Morgan Stanley Sekuritas Indonesia; in Canada by Morgan Stanley Canada Limited, which has approved of and takes responsibility for its contents in Canada; in Germany by Morgan Stanley Europe S.E., regulated by Bundesanstalt fuer Finanzdienstleistungsaufsicht (BaFin); in Spain by Morgan Stanley, S.V., S.A., a Morgan Stanley group company, which is supervised by the Spanish Securities Markets Commission (CNMV) and states that Morgan Stanley Research has been written and distributed in accordance with the rules of conduct applicable to financial research as established under Spanish regulations; in the US by Morgan Stanley & Co. LLC, which accepts responsibility for its contents. Morgan Stanley & Co. International plc, authorized by the Prudential Regulatory Authority and regulated by the Financial Conduct Authority and the Prudential Regulatory Authority, disseminates in the UK research that it has prepared, and approves solely for the purposes of section 21 of the Financial Services and Markets Act 2000, research which has been prepared by any of its affiliates. RMB Morgan Stanley Proprietary Limited is a member of the JSE Limited and regulated by the Financial Services Board in South Africa. RMB Morgan Stanley Proprietary Limited is a joint venture owned equally by Morgan Stanley International Holdings Inc. and RMB Investment Advisory (Proprietary) Limited, which is wholly owned by FirstRand Limited. The information in Morgan Stanley Research is being disseminated by Morgan Stanley Saudi Arabia, regulated by the Capital Market Authority in the Kingdom of Saudi Arabia , and is directed at Sophisticated investors only.

The information in Morgan Stanley Research is being communicated by Morgan Stanley & Co. International plc (DIFC Branch), regulated by the Dubai Financial Services Authority (the DFSA), and is directed at Professional Clients only, as defined by the DFSA. The financial products or financial services to which this research relates will only be made available to a customer who we are satisfied meets the regulatory criteria to be a Professional Client.

The information in Morgan Stanley Research is being communicated by Morgan Stanley & Co. International plc (QFC Branch), regulated by the Qatar Financial Centre Regulatory Authority (the QFCRA), and is directed at business customers and market counterparties only and is not intended for Retail Customers as defined by the QFCRA.

As required by the Capital Markets Board of Turkey, investment information, comments and recommendations stated here, are not within the scope of investment advisory activity. Investment advisory service is provided exclusively to persons based on their risk and income preferences by the authorized firms. Comments and recommendations stated here are general in nature. These opinions may not fit to your financial status, risk and return preferences. For this reason, to make an investment decision by relying solely to this information stated here may not bring about outcomes that fit your expectations.

The following companies do business in countries which are generally subject to comprehensive sanctions programs administered or enforced by the U.S. Department of the Treasury's Office of Foreign Assets Control ("OFAC") and by other countries and multi-national bodies: Samsung Electronics.

The trademarks and service marks contained in Morgan Stanley Research are the property of their respective owners. Third-party data providers make no warranties or representations relating to the accuracy, completeness, or timeliness of the data they provide and shall not have liability for any damages relating to such data. The Global Industry Classification Standard (GICS) was developed by and is the exclusive property of MSCI and S&P.

Morgan Stanley Research, or any portion thereof may not be reprinted, sold or redistributed without the written consent of Morgan Stanley.

Indicators and trackers referenced in Morgan Stanley Research may not be used as, or treated as, a benchmark under Regulation EU 2016/1011, or any other similar framework.

MORGAN STANLEY RESEARCH 63 M FOUNDATION INDUSTRY COVERAGE: Greater China Technology Semiconductors

COMPANY (TICKER) RATING (AS OF) PRICE* (03/12/2019)

Charlie Chan

ASE Technology Holding Co. Ltd. (3711.TW) O (09/21/2018) NT$62.90 ASM Pacific (0522.HK) E (09/21/2018) HK$87.05 Chipbond Technology Corp (6147.TWO) O (07/20/2017) NT$68.20 Chunghwa Precision Test Tech (6510.TWO) E (08/07/2018) NT$413.00 Global Unichip Corp (3443.TW) U (11/05/2018) NT$202.00 GlobalWafers Co Ltd (6488.TWO) U (01/14/2019) NT$303.50 Jiangsu Changjiang Electronics Tech (600584.SS) U (09/22/2015) Rmb16.12 MediaTek (2454.TW) E (01/31/2019) NT$270.50 Nanya Technology Corp. (2408.TW) U (03/03/2019) NT$60.80 Phison Electronics Corp (8299.TWO) E (12/14/2018) NT$266.00 Silergy Corp. (6415.TW) U (08/16/2018) NT$512.00 Silicon Motion (SIMO.O) E (12/14/2018) US$42.46 SMIC (0981.HK) U (09/21/2018) HK$8.45 TSMC (2330.TW) E (11/16/2018) NT$235.50 UMC (2303.TW) U (11/16/2018) NT$11.65 Universal Scientific Ind. (Shanghai) (601231.SS) O (08/04/2015) Rmb13.95 Vanguard International Semiconductor (5347.TWO) O (02/08/2018) NT$66.30 WIN Semiconductors Corp (3105.TWO) U (07/24/2018) NT$162.00

Daniel Yen, CFA

ASMedia Technology Inc (5269.TW) U (02/25/2019) NT$560.00 Aspeed Technology (5274.TWO) O (03/03/2019) NT$731.00 Egis Technology Inc (6462.TWO) O (12/19/2018) NT$212.50 GigaDevice Semiconductor Beijing Inc (603986.SS) U (04/30/2018) Rmb114.00 Macronix International Co Ltd (2337.TW) U (08/13/2018) NT$21.80 Novatek (3034.TW) O (01/21/2019) NT$174.00 Parade Technologies Ltd (4966.TWO) O (03/03/2019) NT$531.00 Realtek Semiconductor (2379.TW) O (08/03/2018) NT$167.50 Shenzhen Goodix Technology Co Ltd (603160.SS) O (07/24/2018) Rmb99.76 Winbond Electronics Corp (2344.TW) U (08/05/2018) NT$14.20 WPG Holdings (3702.TW) O (07/31/2017) NT$39.10 Stock Ratings are subject to change. Please see latest research for each company. * Historical prices are not split adjusted.

64 © Morgan Stanley 2019

The Americas Europe Japan Asia/Pacific 1585 Broadway 20 Bank Street, Canary Wharf 1-9-7 Otemachi, Chiyoda-ku 1 Austin Road West New York, NY 10036-8293 London E14 4AD Tokyo 100-8104 Kowloon United States United Kingdom Japan Hong Kong Tel: +1 (1) 212 761 4000 Tel: +44 (0) 20 7 425 8000 Tel: +81 (0) 3 6836 5000 Tel: +852 2848 5200