The Internet of Things – Opportunities and Challenges for Semiconductor Companies May 2015

The Internet of Things – opportunities and challenges for semiconductor companies May 2015

January 2015 This final report is the result of a collaboration between McKinsey and the Global Semiconductor Alliance (GSA)

For semiconductors, the IoT is GSA/McKinsey collaboration

A key growth opportunity Unpaid collaboration between GSA and ▪ The number of connected IoT devices is McKinsey & Company to develop a expected to reach 20 – 30 bn by 2020 perspective on the implications of IoT for the semiconductor industry ▪ A semiconductor growth opportunity exists for servers/network equipment (“Internet”) and 11 GSA member executives overseeing the components for deployed “things” effort as the Steering Committee A new strategic challenge Interviews with 30 C-level executives from ▪ The highly vertical character of the IoT (many semiconductor companies and the broader IoT small niches) requires a new approach on how ecosystem (including semiconductor to address the market customers) ▪ The IoT is starting to happen but is still early Survey of 229 semiconductor executives in its development (e.g., unclear standards, no from GSA member companies “killer application” yet) ▪ IoT devices often have specific technical Supporting rigorous (quantitative) analyses requirements regarding low power consumption, integration, cost points, Final report summarizing findings (ex- connectivity, and sensors clusively available for GSA members)

SOURCE: Gartner; IDC; ABI Research; GSA and McKinsey & Company “IoT collaboration” 1 The joint GSA/McKinsey report on IoT is available in 3 different formats This report

Presentation – executive summary Presentation – fact pack Written report

20-page presentation 70-page presentation 20-page written report ▪ Short executive summary for ▪ Detailed presentation and ▪ Complementary perspective an overview on the report’s documentation of findings with special focus on key findings interview findings

Reports exclusively available for GSA members at http://www.gsaglobal.org/gsa-resources/publications

SOURCE: GSA and McKinsey & Company “IoT collaboration” 2 Editorial

This report was developed as a part of an unpaid collaboration between the Global Semiconductor Alliance and McKinsey & Company between Aug. 2014 and Apr. 2015

Churchill Tower Sophienstraße 26 12400 Coit Road, Suite 650 80333 Munich Dallas, Texas 75251 Germany United States Dr. Harald Bauer, Director, Frankfurt http://www.gsaglobal.org [email protected] Sandro Grigolli Mark Patel, Principal, San Francisco EMEA Executive Director [email protected] [email protected] Jan Veira, Associate Principal, Munich [email protected]

GSA Steering Committee for this report

David Baillie, Fogale Sensation, CEO Dr. Jalal Bagherli, Dialog Semiconductor Plc, CEO Stan Boland, Neul Ltd., CEO Svein-Egil Nielsen, Nordic Semiconductor ASA, CTO Dr. Steven Gray, CSR Plc, CTO Dr. Harald Hamster, Infineon Technologies AG, Head of Strategy Dr. Yannick Levy, Parrot SA, VP Corporate Business Development Dr. Maria Marced, TSMC Ltd., President TSMC Europe (Chair of Steering Committee) Thomas Riener, ams AG, EVP and Head of Marketing & Strategy Dr. Hans Rijns, NXP Semiconductors NV, CTO Remy de Tonnac, INSIDE Secure SA, CEO

SOURCE: GSA and McKinsey & Company “IoT collaboration” 3 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Good progress towards the ▪ Progress for IoT has been made by increasing supplier push and customer realization of IoT demand, technological and cost advancement, and improved infrastructure ▪ Challenges need to be overcome to unleash the full potential of IoT

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Customer ▪ Which opportunities exist for stimulating market demand? 2 demand Key ▪ What is the best way to navigate an environment strategic 3 Standards of immature standards? challenges Platform ▪ How can customers in fragmented markets be reached successfully? 4 strategy

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 4 Internet of Things can be defined in a wider or more focused way ILLUSTRATIVE Focus of this report1

Classic Internet devices Wide definition The IoT is the network of all PCs Tablets physical objects accessed through the Internet Smart- phones

Big data Wearables Focused definition Medical devices The IoT is the network of all smart things communicating Enterprise intranet Autonomous systems over the Internet with Connected cars Data centers something else without Industrial automation human interaction and that have some sort of sensing Cloud “Smart things” mechanism

1 Key findings in this report are; however; also valid with the wider definition

SOURCE: Expert interviews; McKinsey and GSA IoT survey (n=229; VP-level+ executives from semiconductor companies); GSA and McKinsey 5 & Company “IoT collaboration” IoT is an ecosystem, and “things” are only a small part of this IoT technology stack – example energy/smart grid “Things” SELECTED ELEMENTS OF STACK

Exemplary players Customer Integration Smart meter applications Grid apps (e.g., supply/ Use (e.g., advanced (e.g., demand- (e.g., substation demand cases meter side automation) balancing, load infrastructure) management) forecasting)

System Power grid management systems and system integration integration

Meter data Distribution Outage Asset Customer management Software management information manage- information system system (DMS) system (OIS) ment (AM) system (CIS) (MDMS)

Communica- Home-area network WAM, WiMAX, cellular LAN, RF mesh, PLC tion network (HAN), ZigBee

Semi- Processor DSP Memory Analog MEMS conductor

“Things” Switches Transformers Meters Storage

Energy T&D Substation Wires Customers infrastructure

SOURCE: Press clippings; expert interviews; GSA and McKinsey & Company “IoT collaboration” 6 Many new IoT applications have already been established – even more are gaining traction and are on the horizon Proliferation status of IoT applications Medical Industrial Wearables Smart home electronics automation Connected cars Smart cities

Productivity Smart watch Connected lighting Digital patient record In-vehicle infotainment Public surveillance improvement

Fitness accessories Home automation Logistics tracking Telematics Traffic monitoring Estab- lished Smart meter Vehicle tracking

Existing Surveillance today Smart door lock Predictive process Automatic system Smart glasses Hospital management Traffic control Smart thermostat monitoring upgrade

Gaining Connected appliances Vital function monitoring Remote servicing traction

Intelligent lighting Smart pill Predictive maintenance

Vehicle to vehicle / Intelligent Smart clothes and shoes Gardening Patient localization vehicle to Internet Smart grids production lots Emerging over the communication next 3-5 years Location-based Smart implants Predictive maintenance information

Embedded wearables Assisted living Sensor swarms Autonomous driving Predictive maintenance

Autonomous On the horizon Distributed maintenance environmental Agile/individual monitoring manufacturing

SOURCE: IHS; expert interviews; press clippings; GSA and McKinsey & Company “IoT collaboration” 7 Initial applications show that the IoT is already happening

Wearables – Smart watches Connected car – Insurance Smart metering

▪ In 2014 > 22 mn wearable devices ▪ An American car insurance ▪ > 50 mn smart meters (AMI, sold with a volume of ~ USD 12bn company monitors customer advanced meter infrastructure) are ▪ Launch of Apple watch expected driving habits via the mobile installed in the USA which cover in Q2 2015 with volume up to 30 network with a customized device 43% of all US homes mn units in first 12 months1 installed in the car

Road pricing system IoT technology in automotive Smart service in industrial machine

▪ In Singapore, > 93 gantries have ▪ An electrical vehicle manufacturer ▪ Smart system diagnostics have Electronic Road Pricing System to fixed battery charging on ~30,000 improved the first-time fix rate by manage road tolls which are cars without a “physical recall” 5% for an exemplary equipment charged based on location, time with a remote system update via vendor in the flexible materials slots, and day (e.g., weekday vs. the mobile network industries weekend)

1 Prediction by Morgan Stanley

SOURCE: IHS Technology 2013; press search; Navigant Research; SBD; CCS Insight; University of Michigan; The Edison Foundation Institute 8 for Electric Innovation (IEI); GSA and McKinsey & Company “IoT collaboration” Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however semiconductor industry some ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 9 IoT is considered to be a key growth driver for semiconductors ILLUSTRATIVE in the coming years

Survey of semiconductor 2014 - 2020 industry experts shows IoT 2007 - 2014 as key growth driver 2000 - 2007 How do you rate IoT as a growth driver? Personal com- Wireless com- % puting/internet munications Internet of Things ▪ PCs ▪ Smartphones ▪ Smart home The number 1 Key ▪ Laptops ▪ Network ▪ Wearables incremental growth 17 ▪ Servers infrastructure ▪ Healthcare growth driver for drivers the industry ▪ Industrial ▪ Connected car ▪ Cloud/big data One of the top 3 incremental 48 Average growth drivers industry 3% p.a. 4% p.a. 3 - 5% p.a. of the industry revenue growth

Just 1 of many I Wireless as key growth II … and IoT is likely to growth factors 33 driver will slow down… pick up for the industry ▪ 16% growth p.a. 2009 - 13 ▪ Considered most important ▪ Market maturity expected growth driver by many executives Current to slow growth down to 3% ▪ IoT installed base to grow by 15- I expect limited trends incremental p.a. 2014 - 18 20% p.a. to 26-30bn devices in growth from the 2 2020 IoT ▪ Economic impact > USD 2 tr in 2025

SOURCE: McKinsey and GSA IoT survey (n=229; VP-level+ executives from semiconductor companies); iSuppli; Gartner; IDC; expert interviews; 10 McKinsey Global Institute; GSA and McKinsey & Company “IoT collaboration” I Semiconductor industry growth has primarily been driven by growth in the wireless market over the last 5 years Semiconductor industry revenue by application, 2000 - 20E, USD bn

CAGR CAGR CAGR 2000-07 2007-14 2014-20

+4% p.a. 439 422 409 30 Wired -6% 4% 6% 396 385 374 41 Automotive 8% 5% 6% +4% p.a. 352 Consumer 47 8% -2% -2% 323 21 electronic 313 309 28 +3% p.a. 304 274 57 Industrial 1% 7% 7% 264 52 16 260 239 229 20 231 221 38 Data 25 58 109 1% 0% 1% 186 processing 12 157 161 35 24 100 22

102 95 154 Wireless 8% 11% 6% 112 53 31

2000 01 02 03 04 05 06 07 08 09 10 11 12 13 14E 15E 16E 17E 18E 19E2020E

Note: Figures may not sum due to rounding

SOURCE: iSuppli (Q3 2014); McKinsey; GSA and McKinsey & Company “IoT collaboration” 11 II The IoT installed base is expected to grow rapidly in the long-term, but executives have different expectations on timing and growth segments

Analysts are optimistic about the rapid rise While executive expectations are of the IoT market positive but more tempered IoT installed base1 “I think IoT is the next natural evolution of Connected devices, billions the semiconductor market, but I don’t see Gartner IDC ABI it providing any revolutionary growth” – Foundry executive

“I believe the IoT will happen and reach +15-20% p.a. significant scale but it’s not yet fully clear 30 28 at what point the critical scale for market 26 inflection will be” – Fabless player executive

9 10 7 “If you aggregate all IoT verticals, the opportunity is huge. Obviously, it will grow, but verticals will behave differently and take-off at different rates” 2013 2020 – IDM VP

1 “Wide” IoT definition used by reports (i.e. including "classic" internet devices; such as PCs; servers).

SOURCE: Gartner; IDC; ABI Research; expert interviews; McKinsey; GSA and McKinsey & Company “IoT collaboration” 12 II IoT is also a significant growth opportunity for semiconductors due to the increasing demand for network and cloud infrastructure

IoT device growth Network and cloud infrastructure growth

▪ Number of connected IoT devices Server market – slow but steady growth expected to grow to 26 - 30 bn by Server shipments, millions 2020 +4% p.a. ▪ Trend exists to shift as much +3% p.a. computing power from the “thing” to 9.7 10.1 10.5 10.9 hubs or into the cloud in order to 8.3 8.6 8.7 9.0 9.3 minimize power consumption and form factor of “things”

2010 11 12 13 14E 15E 16E 17E 2018E Network infrastructure – IoT drives demand Increased demand for network infrastructure and hubs expected due to ▪ Increased number of hubs required to deliver connectivity coverage ▪ Lower computational efficiency of small hubs vs. servers, (i.e., higher demand for silicon for same computation needs) ▪ No virtualization of hubs

SOURCE: IDC; press clippings; analyst reports; expert interviews; GSA and McKinsey & Company “IoT collaboration” 13 II The readiness of IoT for reaching the growth ILLUSTRATIVE inflection point varies by vertical Widely available Partially available

Most surveyed semi- Medical Industry Connected conductor industry experts Wearables Smart home electronics Automation cars Smart cities expect growth inflection point in 1-5 years

When do you expect that the IoT inflection point will Inflection be reached? enablers Percent Technology Current price Further devel- Affordability for (e.g., ad- level only opment of che- complex IoT vancement, feasible for mical sensors, (e.g., autono- Now 1 affordability) premium/enthus miniaturized mous driving) to iast customers implants be created ongoing

1 - 2 years 21 Market Profitable Overarching High Interoperability Highest safety Strong business from now (e.g., frag- use cases integration of certification and security and reliability models needed mentation, searched applications requirements currently only need to be that do not standards, needed; in medical within (small) guaranteed; require big security) missing applications; proprietary legal aspects to investments; 3 - 5 years 53 from now standards initial growth ecosystems, be clarified higher integra- impeding expected from industry wide (potential tion of different growth lifestyle/fitness/ standards regulation) applications assisted living missing needed > 5 years from 10 now Ecoystem/ Capabilities for infrastruc- IoT missing in ture (e.g., hospitals I don't expect products a disruption, 14 available, but steady growth network coverage)

Note: Figures may not sum due to rounding

SOURCE: McKinsey and GSA IoT survey (n=229; VP-level+ executives from semiconductor companies); IHS; expert interviews; 14 GSA and McKinsey & Company “IoT collaboration” II Consumer electronics case studies reveal that inflection of growth pockets occurred when the relevant growth enablers was in place

Consumer electronics technology case studies

Mobile Flat- Social network Inflection phone Smart- Net- MP3 screen ▪ 5 growth enablers (India) phones books player TV Blu-ray Tablet PC Mobile enablers Technology identified driven (2 technology driven, 3 Affordability 2005 2009 2007 2002 2003 2006 2010 <2000 2009 ecosystem driven) Usability <2000 2009 <2000 2001 2003 2006 2010 <2000 2009 ▪ Inflection of growth hap- Ecosystem pened when driven the last relevant Infrastructure 2005 2005 2005 n/a <2000 n/a 2005 2004 2004 enabler of availability growth was in place Content n/a <2000 <2000 2003 <2000 2007 <2000 2006 2006 availability

Standards <2000 <2000 <2000 <2000 <2000 2008 <2000 <2000 <2000

Year of 2005 2009 2007 2003 2003 2008 2010 2006 2009 inflection

SOURCE: McKinsey; GSA and McKinsey & Company “IoT collaboration” 15 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 16 Key enablers from semiconductors and beyond for IoT growth are already in place today

A Key players have positioned themselves along the value Supplier chain and have launched flagship products push B Alliance building and M&A activity are extensively used to strengthen companies’ positions

C Consumer pull is gaining momentum in certain segments, Customer (e.g., for wearables) demand D There is increasing interest in IoT solutions (e.g., Industry 4.0)

E Increased device lifetime has been enabled by reduced Technological power consumption due to semiconductor technology progress enhancement F Cost position of critical components (e.g., sensors) is reaching levels that allow for higher market penetration

G Connectivity infrastructure is widely available Required (e.g., LTE, WiFi) infrastructure H Cloud computing is readily available at rapidly declining prices

SOURCE: GSA and McKinsey & Company “IoT collaboration” 17 A IoT players across the value chain are preparing for EXEMPLARY PLAYERS broad IoT adaption Key movements of selected IoT players

Leading IoT trend based on deep understanding of customer needs and innovative ideas

System integrator IP provider Chip manufacturer Solution provider End retailers (EMS)

▪ Develop and launch ▪ Develop and launch ▪ EMS develop IoT ▪ Announce OS and ▪ Various start-ups low-power solution IoT-optimized chip products for non- standard platform emerged for IoT for IoT (e.g., ARM by integrating traditional electronics for IoT (e.g., Apple devices Cortex-M, multiple companies (e.g., HomeKit, Google ▪ Diverse non- Bluetooth LE) components into 1 Google Glass by Nest) traditional ▪ Codevelop extreme chip (e.g., Intel Foxconn, Nike electronics low-power core Quark) FuelBand by companies launch architecture with ▪ Partnership among Flextronics) and adopt IoT devices chip manufacturer semiconductor (e.g., connected light, (e.g., MIPS-Ineda) players for IoT chip insurances) (e.g., Atmel-Bosch collaboration)

SOURCE: Press clippings; analyst reports; GSA and McKinsey & Company “IoT collaboration” 18 B Key players extensively use alliances and M&As NOT EXHAUSTIVE Device Chip M&As and to strengthen their IoT market position alliances Recent IoT-related M&As, alliances and commercial activity

Qualcomm Apple announced Samsung launched Gimbal HomeKit and announced Simband beacon HealthKit for iOS 8 open platform

Nest launched the Samsung and LG Intel announced Apple Nest Protect smart launched smart shirts announced smoke detector connected lighting product Apple Watch

Aug Oct Dec Jan Feb Mar Apr May Jun Aug Oct

2013 2014

Intel launched Intel Toshiba Quark SoC launched launched appli- Microchip Qualcomm and IoT Edison IoT cation processor acquired acquired business group platform for wearables ISSC CSR

ARM Google Verizon, GE, acquired acquired and Intel Sensinode Nest announced IoT alliance

SOURCE: Expert interviews; company web pages; press clippings; team analysis; GSA and McKinsey & Company “IoT collaboration” 19 C IoT verticals are expected to grow due to increasing customer pull

IoT verticals growth Number of shipments, mn Customer pull (examples) 135 “In a telephone poll of 1,011 Americans, young consumers aged 18 - 34 are the most excited 100 about wearable technology. 57% said they Wearables 69 would purchase or wear smart glasses, and 42 22 53% said they are interested in 10 a smart watch’ –Opinion Research

124 133 “Consumers are asking for competitive, low- or 107 115 92 99 no-carbon, highly efficient power. We’re finding Smart that clean power generation through smart grid meters is the most competitive” –Galvin Electricity Initiative

41 “Despite concerns, about ¾ of respondents 32 believe that connected vehicles will reduce the Connected 24 number and severity of crashes, improve vehicles 17 emergency response times and result in better 13 11 fuel economy. In addition, more than 60% expect less traffic congestion, shorter travel times, and lower vehicle emissions” 2013 14E 15E 16E 17E 2018E –University of Michigan poll

SOURCE: Navigant Research; SBD; CCS Insights; University of Michigan; GSA and McKinsey & Company “IoT collaboration” 20 D IoT is currently gaining broad interest – Industry 4.0 example

Advantages of Industry 4.0

Based on Enables Digitization Of production Simpler coordination processes of fragmented, ▪ Manufacturing in the DACH worldwide value region (Germany, Austria, and chains Switzerland) has to catch up on the road to Industry 4.0 (IoT in Networks/ In a Flexible, autonomous manufacturing) connectivity cyberphysical transport systems system (CPS) and efficient ▪ Opportunities with analytics are with auto- warehouse logistics far from exhausted nomous Intelligent products exchanges of that manage their information and own production ▪ German SMEs – the "Mittelstand" automated process – could increase their revenue action even more and create 670,000 triggering new jobs if they would make more use of the latest IT Advanced For example, Flexibility of the technologies 3D printing, production lines down advanced to a lot size of 1 ▪ Industry 4.0 is today largely a robotics matter for big companies technology, sensorics

SOURCE: Expert interviews; Press research; GSA and McKinsey & Company “IoT collaboration” 21 D Industry 4.0 example – condition-based maintenance allows increased utilization due to fewer breakdowns

Condition-based maintenance (CBM) Typical impact of CBM1 Detecting early signs of Prioritizing and Increasing unit Machinery -50% problems for timely optimizing mainte- availability due to breakdowns correc-tion at minimal nance resources using shorter costs real-time data downtimes Failure rate Spare part -20% Per 1,000 Remote sensor already inventory signals replacement need at 24 months Total machine -50% Early lifetime Failure downtime standard failure rate rate suggests to repair replacement at Overtime -20% 48 months expenses

Estimated need for Machine life +20 - 40% repair at 48 months

Overall +20 - 30% 12 24 36 48 Time productivity Months

1 Aggregated average figures from expert interviews – not for a specific company

SOURCE: Expert interviews; ARC Advisory Group; VibroSyst; WiHART System; Schaeffler; Department of Defense; GE; GSA and McKinsey & 22 Company “IoT collaboration” E Semiconductor technology has made good progress NOT EXHAUSITIVE on enabling lower power consumption of ICs

Technology1 Description ▪ System on chip (SoC) combines multiple dies in 1 package through interpose to save on 2007 and SoC power, cost, and footprint of the package earlier.. packaging ▪ Compared to nonintegrated design, SoC reduces power consumption by ~ 35%, package size by ~ 40%, and overall cost by 25%

▪ High-K Metal Gate (HKMG) lowers power consumption of ICs compared to SiO2 gate 2007 HKMG oxide

▪ Leakage current of HKMG is ~ 90% lower than SiO2 gate

ARM big- ▪ New heterogeneous multicore architecture couples low-power processor cores with high- 2011 LITTLE performance, high-power cores architecture ▪ Comparing to Cortex-A15, Cortex-A15∙Cortex-A7 saves 75% of the CPU energy

▪ FinFET is a 3D transistor with stronger gate control from a trigate design and depleted channel 2011-2012 FinFET ▪ Compared to planar MOSFET, FinFET is 30 - 40% faster, and the leakage current is reduced up to 90% ▪ Fully depleted silicon on insulator (FD-SOI) with a depleted channel and and additional 2013 FD-SOI2 ultrathin insulator layer (buried oxide) provides strong gate control and low leakage current ▪ FD-SOI is 20 - 30% faster and has 25 - 50% lower power consumption than bulk CMOS ▪ 3D through silicon via (TSV) is a new packaging technology for multiple types of ICs so that they can be stacked horizontally and vertically in a single package 2014 3D TSV3 ▪ For example, 3D TSV for DRAM is expected to match performance plus 20 - 30% lower power consumption compared to standard DRAM

1 Technologies are (partly) not “additive” but complementary 2 Volume production time for 28nm FDSOI 3 For mobile DRAM

SOURCE: Press research; Company websites; GSA and McKinsey & Company “IoT collaboration” 23 E TSV technology is key for 2.5/3.0D IC integration BACKUP

▪ 2.5DIC technology connects the die to ▪ 3.0DIC technology connects multiple dies vertically, the substrate through a TSV interposer directly through TSV within each die – an extra passive layer with no ▪ True 3D technology eliminates extra processing transistor of interposer ▪ Interposer lowers the risk of TSV ▪ Advanced packaging process control required induced degradation but adds 1 to avoid damage during handling extra layer DieMold 1 Die 1 Die 2 Die 2 Interposer Die 3 VS. Substrate Substrate

PCB PCB

2.5DIC 3.0DIC Image 3D stacked Applications CPU FPGA GPU sensors MEMS DRAM 2010 - 2013 - 14 2015 - 16 2012 stacked 2012 - 13 launch 2015 launch Mass pro- remains nice with 28 nm expected BSI CIS of multi-ASIC expected using duction application for process with 16/20 MEMS sensor ICs HBM and HMC start high-end nm process technology servers

IBM/ Xilinx NVIDIA/ Sony Bosch Hynix/Micron Examples Oracle AMD

SOURCE: Bernstein Research; Yole; press clippings; GSA and McKinsey & Company “IoT collaboration” 24 F MEMS are replacing most conventional sensors needed Details on next slide in IoT devices at a lower cost and better performance

Key trends in MEMS can be used across all IoT verticals MEMS suitable MEMS

▪ Cost and size of Wear- Smart Medical Indus- Connec- Smart ables home electro- trial auto- ted cars cities MEMS are nics mation decreasing while performance is increasing Microphone Integration of ▪ BAW filter2 MEMS into 1 building block is Pressure sensors on-going (e.g., IMU1 combos) Accelerometer ▪ Integration of Magnetometer MEMS with Gyroscope logic expected in next 5 years Lab-on-chip Flow sensor Temperature

1 Inertial measurement unit; consisting of up to 10 different accelerometers; gyroscopes; magnetometers; and pressure sensors 2 Bulk acoustic wave filters

SOURCE: Yole; iSuppli; IHS; expert interviews; GSA and McKinsey & Company “IoT collaboration” 25 F The recent price and volume developments of key MEMS components can be seen as indicators that IoT growth is picking up

Price and volume development of key MEMS devices (normalized to 2013)

Microphones Inertial combos Oscillators Past examples demand/cost correlation Gyroscopes RF MEMS Smartphones Average selling price (ASP) ASP1 touch controller ASP1 WLAN IC CAGR 2.00 2009 - 18, % Annual number of smartphones sold 1.58 USD Units, mn -5 1.25 4 1,000 -8 3 1.00 2 500 -9 0.79 1 0 0 0.63 -14 2003 04 05 06 07 08 09 10 11 12 2013 0.50 -13 2009 10 11 12 13 14 15 16 17 2018 Led bulbs LED price/klm Volume Market size (LED 10.00 CAGR lamp, LED luminaire) 2009 - 18, % USD USD bn 57 40 80 1.00 80 30 60

17 20 40 0.10 35 10 20

6 0 0 0.01 2007 08 09 10 11 12 13 14 15 16 17 18 2020 2009 10 11 12 13 14 15 16 17 2018

SOURCE: Analyst reports; expert interviews; GSA and McKinsey & Company “IoT collaboration” 26 G Wireless infrastructure is prepared for the rising demand in wireless communications

Example 1: 4G LTE coverage in the US Example 2: global wireless coverage Total penetration1 Region Sep 2014

US/Canada 111%

Mexico / South America 115%

Western Europe 122%

Eastern Europe 146%

Middle East 113%

Africa 76%

Asia-Pacific 90%

World average 96%

Currently: 98% of the US The world is nearing 100% population is covered by LTE wireless connectivity

1 Wireless penetration is measured as subscribers/population

SOURCE: Ovum, Verizon; GSA and McKinsey & Company “IoT collaboration” 27 H Case example: Google is passing on economies of scale to ILLUSTRATIVE end customers by dropping prices for cloud storage and big data

Cloud -68% p.a. storage 0.8 USD/ month/ GByte 0.3 Google announces massive price drops for its cloud computing services and storage, introduces sustained-use discounts Big data -86% p.a. analysis (Big- 35.0 Query) USD/ TByte 5.0

Before After April 1, 2014

SOURCE: “The Cloud Service Provider Market”; TechCrunch; press clippings; GSA and McKinsey & Company “IoT collaboration” 28 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 29 Challenges remain for unlocking full potential of the IoT for semiconductor industry – survey results

Key challenges for success in IoT1 Average score from all participants (out of a total of 100 points)1 Security and 1 Security/privacy issues 19.3 privacy are seen as key 2 Low customer demand challenges to 16.4 (no killer apps) IoT growth

3 Lack of common standards 16.3

4 Addressable market challenges (too 15.5 fragmented, unknown/new customers, ...) 5 Value extraction challenges (margins too low, costs too high, 14.8 The industry is value captured by others, ...) optimistic about semiconductor 6 Semiconductor technology challenges 12.1 technology (low power, integration,...) challenges

1 Participants had a total of 100 points to distribute to challenges based on severity/importance of challenge (more points = more important)

SOURCE: McKinsey and GSA IoT survey (n=229; VP-level+ executives from semiconductor companies); GSA and McKinsey & Company “IoT 30 collaboration” The ease to resolve critical gaps varies by vertical INDICATIVE Ease to resolve critical gaps Easy to solve/no challenge Medium effort required Difficult to solve/major challenge

Wearables Smart home Medical Industrial Connected Smart cities a Higher inter- electronics automation cars operability required to increase benefit from home Critical gaps automation b High certification Security 1 c requirements and and privacy varying regional regulations slow Customer down market 2 demand development; scattered market with many small 3 Standards a players c Highest reliability and safety required Fragmented for application in 4 b markets cars (especially autonomous driving) Value d Devices still require 5 extraction significant technological advancement (e.g., power Technological consumption and 6 d advancement form factor)

SOURCE: Expert interviews; GSA and McKinsey & Company “IoT collaboration” 31 Ease of resolution of critical gaps strongly Ease to resolve critical gaps INDICATIVE Easy to solve/no challenge Medium effort required differ by vertical (1/2) Difficult to solve/major challenge BACKUP

Selection of verticals Wearables Smart home Medical electronics Industrial automation Connected cars Smart cities

▪ Data is often not critical ▪ Security is important to ▪ Privacy, especially for ▪ Security is crucial ▪ Security is essential to ▪ Security is critical for prevent unauthorized private medical data, for mission-critical prohibit accidents smart grid and traffic ▪ Security and privacy entry into home and 2nd needs to be guaranteed operations (e.g., control Security have comparably low ▪ Regulatory level attacks on home automated 1 and relevance ▪ Life supporting requirements raise the ▪ Security is less critical automation manufacturing) privacy functions have highest security requirements for functions like energy security requirements ▪ For functions like data efficiencies or lighting tracking or optimization security not as critical

▪ Customer demand ▪ Industry players have ▪ IoT demand for specific ▪ Adoption will be driven ▪ Reliability and ▪ Smart meters and grids is increasing with key been focusing on applications already by economic benefit technology capability are building up in players pushing (e.g., specific products, like exists are critical gaps for America and Europe ▪ Customer demand is Apple, Samsung, Intel) thermostats and fire broader demand ▪ Adoption will be driven currently very ▪ However, public beacons ▪ Expect to take off in by efficiency application specific, ▪ Early prototypes institutions are risk Customer 2 2015/16 ▪ All-in-one products improvements, resulting in a scattered available but no large- averse, budget demand required to further economic and market scale rollout yet restricted, and not increase customer convenience benefits economically ▪ Some customer excitement for customers incentivized to adopt segments are new IoT applications traditionally conservative/ slow with innovations

▪ Wearables can use ▪ Very scattered ▪ Government regulations ▪ Standards are ▪ System and behavioral ▪ Standards are required existing connectivity landscape of competing and system standards advantageous for ease standards still required to ensure standards from standards needed for seamless of integration for autonomous driving communication consumer electronics device integration between different ▪ Large pool of ▪ Proprietary systems are ▪ Limited number of like mobile phones devices but that is only 3 Standards players from different ▪ Medical devices can feasible in industrial automotive OEMs make critical within same ▪ Large consumer industries leverage standards environment but limit standard alignment city/region (e.g., traffic electronics players can from established speed of adoption easier ▪ Interoperability of monitoring, traffic light leverage existing consumer ecosystems devices would be very controls and weather ecosystems and established IT important stations) infrastructure

SOURCE: Expert interviews; GSA and McKinsey & Company “IoT collaboration” 32 Ease of resolution of critical gaps strongly Ease to resolve critical gaps INDICATIVE Easy to solve/no challenge Medium effort required differ by vertical (2/2) Difficult to solve/major challenge BACKUP

Selection of verticals Wearables Smart home Medical electronics Industrial automation Connected cars Smart cities

▪ Market has already ▪ Today, market is ▪ Regulatory ▪ Market of industrial ▪ Only limited amount of ▪ Specific sub-segments been created by mobile developing, although requirements can automation is players for connected are developed (smart device players fragmented with many potentially slow down fragmented and slow cars, and most have meters) and relatively Frag- players for different market development moving, which takes a clear market direction standardized 4 ▪ Potentially dominated mented smart home products long time to create (big OEMs, few new by a few large ▪ Scattered market with ▪ Many end markets markets (long tail) entrants) consumer electronics ▪ In the future potentially many players and highly fragmented players few consumer different types of electronics players products dominating

▪ Business model in ▪ Large markup on ▪ Suppliers can possibly ▪ Systems and solutions ▪ OEMs still have to ▪ Suppliers possibly can wearables is well- hardware; system and extract fair share of cover a large part of the define business model extract fair share of defined from supplier to solutions cover a large value with medical- added value how to extract value value for highly Value OEM part of the added value grade products (healthy from connectivity/data specialized products 5 ▪ Possibly, new business extraction margins) (with right business ▪ Suppliers extract fair ▪ Possibly, new business models necessary for ▪ Suppliers extract fair model) share of value with models necessary component suppliers share of value with components automotive grade products

▪ Single devices already ▪ Even though many ▪ Potentially, very ▪ Cost is less critical as ▪ Technology changes ▪ Technology available have high level of devices are connected specific technical IoT is only a small for low-power designs today maturity; technology to the power network, requirements portion of total are not needed as cars ▪ Low-power designs for advancement will technological changes equipment and have their own power Techno- ▪ Chemical sensors, many use cases not further improve usability for low-power design machinery cost generation logical miniaturization, and crucial, as devices are 6 can extend use cases advance- ▪ Embedded systems still lower power connected to the power and lower cost ment need technological consumption still network advancement in terms ▪ Lower cost is an required to enable ▪ Long-term autonomous of power consumption enabler to reach large some use cases applications exist that and size for certain volumes faster need better energy use-cases efficiency

SOURCE: Expert interviews; GSA and McKinsey & Company “IoT collaboration” 33 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 34 1 The Internet of Things faces new security and NOT EXHAUSTIVE privacy challenges

New security concepts and Security is critical for all elements of the IoT stack solutions needed

Security on device level is ▪ End-to-end security challenging due to low across the entire stack performance/memory ▪ Safe encryption requiring minimal computing LAN resources IoT partly covering mission- ▪ Agile adaption to new critical functionality (e.g., con- emerging threats without Cloud nected car, electricity grid) user interaction ▪ Tampering/hacking Hub Sensitive private or detection commercial data (e.g., health ▪ Compliance with data) regulatory requirements (e.g., on privacy)

Exemplary security risks ▪ Snooping/blocking of commands/data transmission ▪ Injection/alteration of data or commands ▪ Delay of legitimate commands ▪ Use of hacked devices for other attacks

SOURCE: Expert interviews; GSA and McKinsey & Company “IoT collaboration” 35 1 Case example – spam botnets are starting to include smart IoT devices, such as TVs and refrigerators

Cyber attack using IoT 25% of attached devices were Few e-mails per 3 waves of spam devices misused during the attack device per day ▪ Large attack between Dec 23, 2013 Home- and Jan 6, 2014 networking routers ▪ Besides “things,” also conventional devices, Connected such as desktop multi-media computers, laptops, centers and mobile device TV sets used ▪ Attack using simple exploits, such as default passwords and At least 1 misconfigurations refrigerator

Total of > 100,000 devices < 10 emails > 750,000 spam in botnet per device e-mails

Increased out-of-the-box security needed for IoT devices

SOURCE: Proofpoint; GSA and McKinsey & Company “IoT collaboration” 36 1 Ensuring privacy – IoT-driven increase of data collection requires end-to-end security and potentially regulation IoT as a “threat” for privacy

Description Example Illegal exploitation through Espionage for private information Illegal data A Multiple scenarios backdoors or vulnerabilities (e.g., by criminals, hostile exploitation organizations)

Data given to corporations cannot Data is resold or is inherently Loss of control be easily deleted connected to operational data without consumer knowledge

Companies have detailed Detection and use of very private Usage of private Imbalance customer information without information (e.g., pregnancy by information by consent retailers) companies in a B legal way but to Analyses have inherent faults Refusal of credit due to duplicate consumer’s Wrong data possibly causing discrimination names disadvantage forecasts against individual users

Companies have the know-ledge Loss of health insurance (e.g., due to Discrimination/ to discriminate against individual detection of high cancer risk) exclusion customer groups

Each cause needs specific solutions ▪A Strong end-to-end security is a must for all IoT devices ▪B Transparent regulations needed to ensure appropriate data usage and to provide legal clarity for companies

SOURCE: Österreichische Bundesarbeitskammer; GSA and McKinsey & Company “IoT collaboration” 37 1 Security requirements for IoT vary by segment NOT EXHAUSTIVE

Level of security Security concepts/examples Typical requirements by segment

Con- Light Wear- Smart Indus- nected ables home trial car ▪ Public key cryptography, e.g., SSL Trans- VPN Software mission ▪ IPsec encryption ▪

▪ Secure memory Data storage ▪ Certificate management encryption

▪ Software obfuscation Device operating ▪ Secure bootstrap system encryption Hardware ▪ Hardware-implemented de/encoding algorithms Hardware encryption ▪ Hardware random number generators

▪ Physically unclonable Additional regulatory Physical Non digital, functions requirements exist in physical security elements ▪ Quantum transmissions some regions1 Strong

1 For example; Germany requires smart meter data to be hardware encrypted

SOURCE: Expert interviews; GSA and McKinsey & Company “IoT collaboration” 38 1 loT security is an opportunity for a vertically integrated play from semiconductor companies through growth or collaboration “Organic" opportunity Partnership opportunity

End-to-end security needed for IoT stack A) Move upwards B) Expand from the middle C) Move downwards

Many key competencies in Applica- Knowledge transport layer Application tion/ of low-level security design designer key cloud (hardware) are applicable competency security to the application provides layer advantage in Control of designing application higher-level interfaces Trans- (software) Network equipment and/or port/ End-to-end security manufacturer core security offer customer hub competency access can Hardware design provide an capabilities are advantage in needed for defining low- offering an level “Thing” Semiconductor integrated designs/arch- manufacturer solution itectures core competency

Organic development of capabilities may not be in a short time ▪ Collaborations, partnerships, or M&A are viable strategic options

SOURCE: GSA and McKinsey & Company “IoT collaboration” 39 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 40 2 A lack of market demand is a growth inhibitor for IoT – semiconductor players can help customers to create demand

Status inflection enablers today by IoT vertical

Wear- Smart Medical Industry Con- Smart ables home electro- auto- nected cities Necessary approach to nics mation cars help stimulate demand Inflection enablers ▪ Provide open platforms and an ecosystem to Technology facilitate use case development ▪ Educate customers and Market build customer capabilities (e.g., with dedicated sales force, FAEs1) Ecosystem ▪ Complement semiconductors with software to enable Lack of profitable use cases customers to use products ▪ Verticals have IoT products, but lack a profitable use cases more easily Missing customer capabilities ▪ Develop different levels ▪ Use cases with economic added value exist, but customers have yet to buy in for of development support IoT’s added value depth depending on ▪ Many B2B customers do not know how to leverage the connectivity of “things” customer needs and how to create economic value

1 FAE:Field application engineer

SOURCE: Navigant; IHS; SBD; GSA and McKinsey & Company “IoT collaboration” 41 2 Major semiconductor players are building software eco-systems around their products

IoT strategy (outside-in perspective) Released iNEMO Partnered with Joined ARM mbed Announced new filtering and predictive Thingsquare wireless project in open development Partnering strategy software tool in Mar connectivity platform Dec 2013 platform in Nov with players across 2011 in Mar 2013 2014 IoT value chain

M&A strategy targe- Acquired Mashery’s Acquired Aepona’s API Acquired Wind Acquired McAfee Relaunched Intel ting verticalization of API management exposure and River embedded security software Services Division in IoT by building software in May monetization software software in in Feb 2011 Jul 2014 software/service Jul 2009 2013 platforms in May 2013 capabilities in house

Updated BeeKit Released Onebox IoT Updated Code- Development of wireless gateway Warrior software in connectivity soft- connectivity toolkit Sep 2013 Apr 2014 ware to complement in Jan 2013 processor/MCUs

2009 2010 2011 2012 2013 2014

Launched AllJoyn Released Internet of Released 2net Acquired 2,400 IoT software Everything Devel- Mobile software software and mobile Development of framework in Feb opment Platform in platform in OS patents from HP in connectivity soft- 2011 Jan 2013 Sep 2013 Jan 2014 ware to complement processor/MCU products, selected acquisitions to Merged with SiRF in Released Harmony Updated µEnergy Released VibeHub Released CSRmesh strengthen Jun 2009, inte- software framework BT platform in networked audio BT protocol in Feb technology portfolio grating its GPS in Sep 2010 Jun 2013 platform in Jan 2014 2014 software tools

SOURCE: Company websites; press search; GSA and McKinsey & Company “IoT collaboration” 42 2 Open-source hardware can be an enabler for start-ups and hobbyists to enter into IoT device development

Rele- Open-source Openness levels Implications for Business vance hardware Applications Mech. Firm- Circuit Elect. IC semiconductor model/concept for IoT players for product design ware layout circuit design players Basic enabling of openPicus, IoT devices Can stimulate IoT devices (e.g., Electric Imp, built by start- demand from providing connec- Spark Labs ups, larger smaller cus- tivity backbone for device tomers such as IoT devices made manufacturers start-ups due to of hardware, or individuals reduced devel- software, and/or opment costs and cloud service) generally reduced time to market

Supply for maker Arduino, DIY/Maker/ho Low unit counts community (i.e., Adafruit, memade not worth their own components, frame- SparkFun, IoT devices development, can works, and instruct- OLIMEX be a (small) tions for home- market for made electronics, existing ICs many of which are IoT “things”)

SOURCE: Press clippings; GSA and McKinsey & Company “IoT collaboration” 43 2 IoT customers have a broad range of demands with a need PRELIMINARY for tailored platform offerings and targeted customer support High customer involvement No customer involvement

Level of customer involvement Types of IoT end Company Product con- Solution Device Key customer Requirements for customer example cept design development manufacturing needs product offering ▪ Low cost Highly flexible ▪ Fast time-to- solution offering market Start-ups August using off-the-shelf ▪ High variability components ▪ Ease of Frequent use of Mostly development standard OSs1, outsourced e.g. iOS, Android Large ▪ Performance Full customer consumer ▪ Reliability and support to enable electronics security maximum ▪ Compatible with customization/ legacy system performance Large ▪ Low total cost industrial of ownership

Traditional electronics Traditional electronics Industrial- specific OS1 (e.g. RTOG) ▪ Help with product design Full turnkey Nontraditional ▪ Integrated solution from companies solution project chip design to Lack of experience in device/ convenience manufacturing solution development (opportunity for partnerships with third parties)

1 Operating systems

SOURCE: GSA and McKinsey & Company “IoT collaboration” 44 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 45 3 IoT standards are not yet mature – Widely adopted NOT EXHAUSTIVE connectivity example New standard Established, adoption ongoing Current wireless connectivity standard landscape Data rate, Log scale Power consumption, indicative 1 Many competing standards 1 Gbps High for low and medium-low data rate hinder growth for many IoT applications 100 Mbps 4G ▪ Interoperability missing ▪ Consortia wars might 10 Mbps be emerging 1 ▪ Additional incompatibilities 2 in higher communication 1 Mbps layers (e.g., 6LoWPAN vs. LTE Cat. 01 Zigbee) 2 Standard whitespace for 100 Kbps 802.11ah low-data rate, low-power, high-range applications such as smart grid 10 Kbps ▪ WiFi/LTE power consumption is too high 100 Bps ▪ Alternatives with low power and a wide range (e.g., LTECat. 0, Low 10 Bps 802.11ah, SIGFOX, On- 10 m 100 m 1 km 10 km 100 km Ramp) are in very early stages Range, Log scale

1 Preliminary specs

SOURCE: Expert interviews; company websites; press research; GSA and McKinsey & Company “IoT collaboration” 46 3 Different approaches exist for developing ILLUSTRATIVE – SELECTED EXAMPLES ONLY standards for the Internet of Things Interest group Open standard initiative Industry player IoT standardization efforts

Wearables Smart Connected Medical Industrial Smart cities home cars electronics automation

Apple1 Google All- Industrial Internet Consortium (IIC) Intel Gateway HomeKit/ /Nest Joyn Solutions Appli- HealthKit cation

Qivi- con

Open Inter- API/ connect Ser- Consortium vices (OIC)

Conne- ctivity Wi-Fi, cellular IoT, LTE-MTC, Bluetooth, ZigBee, Z-Wave, SIGFOX, RFID

1 Apple is using a Bluetooth specification that requires special connectivity chips

SOURCE: Press clippings; company websites; GSA and McKinsey & Company “IoT collaboration” 47 3 Semiconductor players need to pursue well-thought out Deep-dive strategies in IoT landscape of immature and missing standards

Development Good standards of standards is must fulfill key Depending on company type, different still in progress requirements approaches are required ▪ Multiple ▪ Holistic roll-out ▪ All players should competing with many – Pursue a hedging strategy with a standards are participants/ focus on standards that meet the available in some devices criteria for high likelihood of success domains ▪ Strong – Actively support industry ▪ Standardization ecosystem associations in promoting standards is still missing in available Use collaborations with other many areas in – industry players to drive verticals and layers ▪ Clear value standardization of the stack proposition for key stakeholders ▪ Large players can act more independently due to sufficiently large volumes and can potential be driving forces in establishing standards

SOURCE: GSA and McKinsey & Company “IoT collaboration” 48 3 Successful standards exhibit common key characteristics NOT EXHAUSITIVE BACKUP

Analysis of connectivity standards Common factors for successful standards ▪ Success of 5 different connectivity Successful rollout standards in the US was investigated Rapid and scaled deployment LTE, Bluetooth, WiFi, and 3G CDMA ▪ Ease of implementation came out as winners while WiMax did not succeed Push & commitment from lead players ▪ Winners shared key characteristics in Strong partnerships their strategies Strong ecosystem Highly available products and services Large accessible market Open/accessible standard Common value across key stakeholders Clear value to stakeholders Low cost/investment to adopt Strong interest group or association to align players

SOURCE: Expert interviews; GSA and McKinsey & Company “IoT collaboration” 49 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 50 4 IoT device specifications vary significantly within a EXEMPLARY IoT VERTICALS multidimensional solution space, depending on application

Smart Smart Industrial meter watch automation

Every Every Permanent Power consumption year few days Recharge cycle Mbit/s Connectivity Transmission speed Connectivity 1m 100m >1km Transmission distance Data processing Low High Computation speed Data processing < MB MB GB Memory capacity None Light Strong Communication security

Irrelevant cm mm Form factor

USD 0.01 USD 1 > USD 100 Price of IoT chipset

SOURCE: Press research; GSA and McKinsey & Company “IoT collaboration” 51 4 Many small IoT market segments cannot be profitably addressed with a traditional business model for integrated devices

▪ IoT device specifi- A platform strategy Breakeven volume in different market environments1 cations vary along is necessary multiple dimensions Annual shipments, mn ILLUSTRATIVE ▪ Allows to cover many – Power CALCULATION IoT markets with small requirements, annual volume lifetime Low- – Connectivity complexity 20 ▪ Requires design identification of IoT – Form factor device archetypes, – Performance even over different – … High- verticals ▪ Traditional R&D and complexity ▪ Needs definition of a marketing approach design common-ground would include the 74 platform for each of design of an these archetypes application-specific product Most IoT Wearables: Smart grid: verticals: very ~22M units – ~100M units ▪ Minimum volume is small volumes required to breakeven per specific with each product – application with minimum volume a few thousand depends on product to a few million complexity units per year

1 Assumptions: R&D cost USD 12 mn for low-complexity IC design and USD 45 mn for high-complexity design; 5-year product lifetime; market with 10 competitors; ASP USD 3.00 in year 1; thereafter 10% price decline per year; gross margin of 55%

SOURCE: McKinsey Numetrics; GSA and McKinsey & Company “IoT collaboration” 52 4 Platform strategy if it is possible to identify device archetypes EXEMPLARY that cover several different IoT verticals

IoT requirements by vertical/application

Processing power Connectivity range eMemory Security Low High Med. Low Long Short > 1Mb < 1Mb Enhanced Basic Power > 700 < 100 Mobile BLE, Zighee, MHz Mhz broadband WiFi High-level IoT device Accessories with screen archetypes Wear- Accessories w/o screen ables High performing Clothing/ shoes ▪ Very high (AP- like) processing Lighting system ▪ Long/short- range RF Home automation Smart ▪ > 1Mb memory Enhanced home Smart metering ▪ security Home surveillance Medium-end Hospital patient/asset ▪ Med processing Medical management ▪ Short/long- elec- range RF tronics Personal health ▪ < 1Mb memory monitoring ▪ Enhanced Con- Entertainment system security nected cars Field monitoring Low-end Industrial Preventive maintenance commodity auto- Low processing mation Supply chain monitoring ▪ ▪ Short-range RF < 1Mb memory Smart Public surveillance ▪ cities Traffic control

Simplified and illustrative view – semiconductor players will need to analyze which applications can be served by 1 archetype platform on a more granular level

SOURCE: Interviews; press clippings; company websites; GSA and McKinsey & Company “IoT collaboration” 53 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 54 5 Semiconductor companies need to identify new ways how to extract “fair share” of overall value generated by IoT for themselves

▪ IoT is expected to create an overall Opportunities for value extraction value of > USD 5 trillion by 2025 ▪ Complement components with software ▪ End-users and consumers will to provide a more comprehensive solution benefit from this value creation ▪ Use security to span all elements of the through IoT applications technology stack and extract value from ▪ Companies within the IoT offering end-to-end solution technology stack will also be able to ▪ Offer system integration services for IoT capture a share of the created value devices ▪ How to extract a fair share of the ▪ Investigate opportunities for new business overall created value for models in IoT that go beyond selling semiconductor companies is an hardware unresolved question ▪ High value extraction by IoT service providers and cloud players might diminish available value for semiconductor companies

SOURCE: McKinsey Global Institute; Cisco; expert interviews; GSA and McKinsey & Company “IoT collaboration” 55 5 The semiconductor industry has started to integrate M&A Inhouse along value chain and is building up software capabilitites Collaboration

Software/API functionality – selected players1

▪ High share of overall Applica. Connec- Interop- created value by IoT is specific Sensor Security tivity erability OS created in Internet/ cloud part of IoT stack (expert estimates: up to 80 - 90%) ▪ Primary focus ▪ Small share of value is on created in things is a connectivity risk for semiconductor and inter- players failing to operability capture fair share of ▪ Most overall created value companies complement ▪ Adding in-house complementary software software to sold development semiconductor devices with M&A is an opportunity to and/or increase captured collaborations value for semiconductor players via a more holistic IoT solution

1 Outside-in analysis

SOURCE: Company websites; press search; GSA and McKinsey & Company “IoT collaboration” 56 NOT EXHAUSTIVE 5 With proliferation of IoT, there are opportunities for completely new business models – potentially also for semiconductor players

Description Example of business model Generate revenue from the sale of ▪ Google/Nest offers a home automation hardware Hardware 1 devices that are sometimes bundled with product (smart thermostat) that can be controlled centered a free service remotely from any device Generate revenue from licensing out IP ▪ ARM offers licenses for computer processor 2 IP centered building blocks; additional revenue by architecture built around its own proprietary RISC supporting development of IP technology Leverage customer/system data to offer ▪ GE uses operational data from sensors on its 3 Data driven better services/pricing, improve product industrial machinery and aircraft engines to provide design, and optimze performance services such as efficiency optimization Provide a software-based infrastructure ▪ PTC’s Axeda machine cloud offers its enterprise 4 Platform for enterprises to deploy IoT solutions customers a complete M2M and IoT data intergration and application development platform Provide data analytics, IT consulting, ▪ Microsoft’s cloud offering Azure, has specialized IoT Professional 5 business intelligence, and cyber security service offerings that enable enterprises to perform services to B2B customers real-time analytics, machine learning, etc. Tightly integrate hardware, software, ▪ Apple controls the entire user experience by selling Closed and services to incentivize customers a set of hardware products with a layer of free 6 ecosystem to stay in a company’s closed ecosystem software and paid services on top of products Provide a free basic service and attempt ▪ Fitbit provides free software with the sale of its to upgrade customers to a paid, premium hardware 7 Freemium service ▪ Additionally, users can upgrade to the premium app that offers additional features Utilize information gained to open, ▪ Google’s wearable product Glass will enable Adjacent rev- 8 adjacent revenue streams (e.g., location- companies to target customers based on location enue streams based advertising, data selling)

SOURCE: Press research; Business Model Generation by Osterwalder/Pigneur; GSA and McKinsey & Company “IoT collaboration” 57 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards 6 advancement low-power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 58 5 Technological advancement in Iow power, device performance and device size is necessary to unleash full potential of IoT c Details on next page

Cost advancement of semiconductors has gone a long way

10,000 10,000  0.03 100 Storage But still more development USD per GB 1 needed 0.01 ▪ Today, many IoT applications 1992 2000 10 2015 technically solvable but have long development times, 10,000 222  0.01 high product cost, or still Computation 100 immature features (e.g., USD per 1 form factor, run time) 1 mn 0.01 ▪ Technological transistors 0.0001 advancement is required to 1992 2000 10 2015 unleash te full potential of IoT –a Lower power 10,000 1,200  0.63 consumption, increased run time Connection 100 –b Integration, smaller USD per 1 Mbps form factor 0.01 2000 10 2015

SOURCE: Deloitte; drpeering.net; cmu.edu; postscapes.com; Texas Instruments; GSA and McKinsey & Company “IoT collaboration” 59 6a Emerging battery technologies may help enable low-power IoT applications, but they are not sufficient alone to enable the IoT

Evolution of battery technologies Emerging technology Pioneers (examples)

Energy density, Wh/kg Thin-film batteries are printable batteries for applications that 2,000 Li-air require a thin form factor and ~2,000 flexibility

Lithium ion with Wirelessly chargeable Si nanowire 400 batteries that can be powered 400 over the air (e.g. using a RF signal) Lithium ion 110 - 140 Nickel-metal Fuel cells convert chemical 100 hydride 50 - 75 energy from fuels (e.g., Nickel cadmium hydrogen) into electricity 35 - 60

Nickel iron 50 30 - 40 Energy harvesting is the Lead acid process of capturing minute 25 - 45 amounts of energy from natural sources (e.g., solar, thermal, 0 wind) and could enable battery- 1860 1910 1960 2010 2020 2030 free devices

SOURCE: International Energy Agency; expert interviews; company websites; whitepapers GSA and McKinsey & Company “IoT collaboration” 60 6a 2 technology options are available for ultralow power in leading edge chips: FinFET and FDSOI

16/14 nm FinFET 28 nm FD-SOI1 Thin silicon film

Source Gate Drain

Insulator Description 3D transistor design that is more Conventional planar transistor design space- and power-efficient than but with an additional ultra-thin conventional planar designs insulator layer (buried oxide) New processes (e.g. double patter- Current planar production lines can be Manufacturing ning) and tools required to create 3D used with small modifications shapes at 20nm node size Intel, TSMC, Samsung, and STMicroelectronics, Samsung, and Vendors GlobalFoundries GlobalFoundries plan to start production

New technologies have an up to 70% reduced power consumption and at the same time up to 60% improved performance compared to 28nm bulk

1 Fully depleted silicon on insulator – smaller FD-SOI nodes are currently in development

SOURCE: Expert interviews; manufacturer websites; press clippings; whitepapers; GSA and McKinsey & Company “IoT collaboration” 61 6b SoC and SiP solutions have significant advantages but are often only viable for large-scale solutions

Technology comparison SiP (system in package) SoC (system on chip) Better ▪ Fewer and shorter interconnects ▪ Highest integration – lowest Power – lower power consumption power consumption consumption

▪ Vertically stacked chips can ▪ Highest integration – smallest Package reduce footprint footprint size

Worse Non- SiP SoC ▪ Higher economies of scale ▪ Highest economies of scale inte- ▪ Complex packaging – higher ▪ Simple packaging – lower Product grated assembly costs assembly costs cost ▪ Yields are typically lowest – ▪ Yields are typically lower – ~ 80% ~ 90% Better ▪ Up to 1.5x R&D effort ▪ Up to 3x R&D effort Develop- ▪ High design costs ▪ Highest design costs ment cost

Time to ▪ Longer TTM means additional ▪ Longest TTM as silicon respin is market complexity required for each die change (TTM)

Worse ▪ Individual components can be ▪ Change of components requires Non- SiP SoC replaced redesign of whole chip Flexibility integrated

SOURCE: Product data sheets; ITRS; expert interviews; GSA and McKinsey & Company “IoT collaboration” 62 6b Integration of components into SiP or SoC requires access to several technologies – close collaboration of several suppliers is necessary EXEMPLARY\ Traditional electronics value chain IoT chip scenario COMPANIES Chip Devices/ Devices/ IP IP Chip supplier supplier solution solution Semiconductor chip manufacturers

IP providers 1 Key trend for IoT 1 IP provider collaborate with chip manu- AP/MCU facturer to build an extreme low ▪ pushes power core integration of architecture for 2 own MCU IoT solution Connec- technology tivity (e.g. with Bosch 2 Different chip Sensortec) component ▪ Is developing suppliers will SigFox collaborate to Sensors transceiver/ develop single chip MCU SoCs for IoT 3 Turnkey service from chip to device Memory PCB assembly will be needed for nonelectronic IoT Buy 3 customers SoC OEM “Collaboration with leading sensor manufacturers will enable Atmel to provide customers with the most appropriate sensor solutions, hereby reducing overall time to market,” – Senior Product Marketing Manager, Atmel Corporation

SOURCE: Expert interview; GSA and McKinsey & Company “IoT collaboration” 63 Contents

▪ IoT spans a broad field of applications and is starting to happen now Introduction

IoT as a growth driver for the ▪ IoT is expected to be a key growth driver for semiconductors; however some semiconductor industry ambiguity exists about the timing and the magnitude of growth

Security and ▪ How can security be ensured across the IoT stack? 1 privacy ▪ How can privacy requirements be met for IoT?

Value ▪ How can a suitable business model be chosen? 5 extraction ▪ How can a semiconductor company capture value?

Technological ▪ What can be done to continue the technological improvements towards low- 6 advancement power and low-cost IoT devices?

Implications for semiconductor ▪ Semiconductor players have to pick their role in an evolving market players and embrace challenges as opportunities

SOURCE: GSA and McKinsey & Company “IoT collaboration” 64 IoT is an opportunity for the semiconductor sector, but it calls for new business and operating models

Strategic needs for individual semiconductor players

I Identify application- and vertical-specific growth pockets ▪ Perform an in-depth assessment of market opportunity and requirement for each IoT vertical/application ▪ Identify growth pockets that fit well with specific products/ capabilities of each respective semiconductor company ▪ IoT presents an ▪ Make bold moves if required to enter a specific opportunity for new application/vertical growth beyond II Seek value beyond silicon mobile ▪ Understand application- and vertical-specific value drivers for ▪ IoT creates potential the end-user (how does the consumer or enterprise derive opportunities for value) value capture ▪ Identify opportunities beyond silicon (e.g., system integration, beyond silicon software) to generate value for customers ▪ Test alternative business models to capture more of the generated value (e.g., usage based) III Revisit the operating model to cultivate and support IoT innovation across more fragmented products and markets (e.g., setup of organization, R&D investment approach)

SOURCE: GSA and McKinsey & Company “IoT collaboration” 65 I Identify application- and vertical-specific growth pockets – ILLUSTRATIVE diverse opportunities exist for IoT plays

Many opportunities Verticals Medical Industrial Con- for players to find Wear- Smart Smart electron- auto- nected profitable niches, ables home cities Products ics mation cars (depending on a company’s specific End-user capacity profile), for instance: Appli- applications cations ▪ Player A Cloud services Player C Expert in consumer products with full system integration Servers/ capability infrastructure ▪ Player B System Expert in leading inte- Devices edge IC with high gration performance for pro- Software/alg- sumer applications orithms Player B ▪ Player C Player A Expert in industrial Processing high reliability and chips security ICs ▪ Player D Connectivity Player E Silicon Player D Connectivity enabler chips ▪ Player E Processing provider (Embedded) with vertical-specific sensors connectivity solution

SOURCE: GSA and McKinsey & Company “IoT collaboration” 66 II Seek value beyond silicon – Opportunities further ILLUSTRATIVE up the value chain need to be investigated Partnerships/alliances Semiconductor players’ approaches to the IoT business model Integration/M&A

Classic hardware-oriented IoT as an opportunity to IoT as a new business model Value chain semiconductor business sell more silicon across the entire stack End-user Appli- applications cations Cloud services

Servers/ infrastructure System inte- Devices gration Software/ algorithms Processing chips Connectivity Core semiconductor Core semiconductor Core semiconductor Silicon chips business business business (Embedded) sensors

Software today consid- Differentiation by Full integration to capture ered prerequisite to sell applications focused on all value – targeted by silicon – rarely useable end user – currently key some players but market to capture more value strategy of many players still too nascent to judge

SOURCE: GSA and McKinsey & Company “IoT collaboration” 67 III Revisiting the operating model – enablers for growth in IoT

From To ▪ Limited number of large BUs ▪ Multi-market approach to diverse end ▪ Direct sales and FAE markets Organization ▪ Application-specific R&D ▪ Leverage distribution sales channel programs ▪ Platform approach to R&D ▪ Largely HW based pricing ▪ Explore business models based on usage ▪ Differentiation on features and and value (e.g., usage based) Business model functionality in HW ▪ Differentiation in SW and services (e.g., secure access) Approach to ▪ Application-focused roadmaps ▪ Platform approach to address multiple IoT portfolio / (e.g., Automotive) use cases across verticals technology ▪ BU driven R&D investments and ▪ Leverage third-party IP to build platform development IP (e.g., connectivity, image processing) ▪ Core IP development in design ▪ IP integration for IoT platform designs ▪ Application development with ▪ Platform development: tools design kits, Capabilities medium to large customers developer communities ▪ System (board) level design capability ▪ Limited number of large portfolio ▪ Significant platform investments (tools, bets decided by BU lead user community, platform-level ICs) Investment ▪ Typically, biased investment in ▪ Small bets in many application markets approach core products vs. application ▪ Integration (M&A, alliances, partnerships) development distant from core as a key capability1

1 It is still contested in the industry if integration in unfamiliar territories (e.g., applications/product design, cloud services ) is an advantage or loss of focus

SOURCE: GSA and McKinsey & Company “IoT collaboration” 68 IoT will have particular implications on industry sub-segments and players Selected themes for specific segments and players based on value chain role

Players need to choose strategy for IC design IoT requires bold ▪ Platform approach – offer multiple chip platforms to serve across moves from IDM verticals; derive value from scale and fabless players ▪ Specialist approach – go deep into very selected verticals: derive value from specialist ‘high performance’ applications

Processor and MCU players migrate to compete for IoT SoC IoT drives a new ▪ Processor players “move down” and offer more power and model for cost-efficient mid/low-end solutions with existing integrated competition of AP functionality for connectivity/RF and sensor I/O and MCU players ▪ MCU players need to “move up” to add connectivity/RF and increased functionality for sensor integration

Foundry players have potential to both drive IoT innovation IoT recasts a and benefit from increased demand at nodes beyond n-1 foundry opportunity ▪ Foundry innovation is required on lagging nodes: RF integration, beyond leading NVM integration, and new structures, ultra low power edge architectures ▪ New volume for fabs at 40nm and above

SOURCE: Executive interviews; GSA and McKinsey & Company “IoT collaboration” 69