TW Semiconductor Report en

Semiconductors – the Next Wave
Opportunities and winning strategies for

semiconductor companies
April 2019
Foreword
The ever-evolving semiconductor scene 5
Future of automotive semiconductors 12
Race is on for AI chips 30
Muted tone for M&A 38
Accessing the Chinese market 46
Digitalization is key to competitiveness 52
2
Foreword
Semiconductors are essential technology enablers that power
many of the cutting-edge digital devices we use today. The global
semiconductor industry is set to continue its robust growth
well into the next decade due to emerging technologies such as
autonomous driving, artificial intelligence (AI), 5G and Internet
of Things, coupled with consistent spending on R&D and
competition among key players.
East Asia (Mainland China, Japan, South Korea and Taiwan) is

where some of the world's most important semiconductor
players are located. The region has become a hotspot for the
semiconductor industry due to its burgeoning economy, the
rise of mobile communications and growth in cloud computing.
China, in particular, commands almost half of overall market
value, split roughly 50-50 between domestic demand and
Taiwan-based, world-leading ODMs (e.g. Foxconn and Quanta) or
foundries (e.g. Taiwan Semiconductor Manufacturing Company
[TSMC]) serving global clients. China also aspires to a self-
sufficient semiconductor industry and ascension to become a
global powerhouse at the same time. Japan, on the other hand,
is an important supplier of semiconductor materials, high-end
equipment and special semiconductors. Meanwhile, South Korea
has a commanding lead in the global HBM (high bandwidth
memory) DRAM (dynamic random-access memory) market.
M&A activity in the semiconductor industry has already peaked

and specialized verticals are becoming the primary focus.
Japan and Korea are seeking to revive their industries through
acquisitions, while the continuing trade war and disputes over
intellectual property will hamper China's ambition to go on a
global spending spree.
3
The semiconductor sector's growth trajectory will flatten

somewhat as demand for consumer electronics saturates.
However, many emerging segments will provide semiconductor
companies with abundant opportunities, particularly
semiconductor use in the automotive sector and AI.
In the automotive sector, the adoption of safety-related

electronics systems has grown explosively. Semiconductor
components that make up these electronic systems will cost
USD600 per car by 2022. Automotive semiconductor vendors
will benefit from a surge in demand for various semiconductor
devices in cars, including microcontrollers (MCUs), sensors
and memory. Automation, electrification, digital connectivity
and security will result in the addition of more semiconductor
content to automotive electronics and subsystems in the next
decade.
The AI semiconductor scene has seen a race not just at the

application level, but also at the semiconductor chip level, where
different architectures are vying for a piece of the pie. The cloud
is the biggest market for AI chips, as their adoption in data
centers continues to increase as a means of enhancing efficiency
and reducing operational cost.
Finally, China has become a source of income for top global

semiconductor companies, many of which generate over half
of their revenue from China. Multinationals trying to access the
Chinese market should consider a multitude of factors such as
policies, technologies, marketing, logistics and global strategies.
It is imperative for a multinational to realize the position it is in
before entering China to come up with the best entry strategy.
4
Semiconductors – the Next Wave | The ever-evolving semiconductor scene
The ever-evolving
semiconductor scene
Robust growth expected and the inclusion of emerging
In the past few years, growth of the technology such as AI in products
global semiconductor industry has and 5G networks, as well as rapid
been driven largely by demand from growth in automotive and industrial
electronics such as smartphones electronics, will be some of the
and the proliferation of applications market's key driving forces. The
including the Internet of Things bulk of semiconductor revenue will
and cloud computing. The global come from processing electronics
semiconductor sector's total revenue (e.g. storage and cloud computing)
is set to increase from USD481 billion and communication electronics (e.g.
in 2018 to USD515 billion in 2019 wireless).
and continue its robust growth well
into the coming decade. Continued
enhancements of existing products
5
Figure: Global semiconductor sales revenue (2016-2022, billion USD)
542.64 billion (2022)

600 0.25
542.64
515.51 509.07 514.19
500 481.09 0.2
420.39 Communication
Electronics: 30.2%
400 0.15
345.85 Data
Processing
Electronics:
300 0.1 34.2%
Consumer Electronics:
200 0.05 9.3%
Automotive Electronics:
100 0 12%
Industrial Electronics:
0 -0.05 Military/Civil Aerospace 13.2%
Electronics: 1.1%
2016 2017 2018 2019 2020 2021 2022
Source: Gartner, Deloitte analysis
Automotive and industrial the semiconductor industry, with revenue

run-away segments from consumer electronics, data
Automotive electronics and industrial processing and communication
electronics are expected to be electronics set to grow steadily.
the fastest growing markets in the
6
Figure: Semiconductor revenue growth by electronic equipment type (2017- 2022)
Head Mounted Display：34.9%
Smart Watches：23.9%
ADAS：23.6% Automotive Electronics

Industrial Electronics
Other Wearables：22.9%
Consumer Electronics
EV/HEV：21% Data Processing
Communication Electronics
Security：18.5 % Military/Civil Aerospace Electronics
2017-2022 Other Industrial: 17.5%
GAGR Body：13.3%
Other Consumer：13.0%
Solid-State Lighting: 12.4%
Storage：11.8%
Instrument Cluster：11.2%
Automation：11.2%
Energy Management：8.5%
Infotainment：8.2%
Transportation：7.7%
Fitness wearables：7.7%
5%
0% 10% 15% 20% Mean：7.2%
Other Audio：6.1%
Chassis: 6.1%
Aftermarket：6.1%
Wired Communication：5.1%
Commercial Amusement：4.3%
Medical/healthcare：3.8%
Test/measurement：3.6%
Powertrain：3.4%
Appliances：3.0%
Military/Civil Aerospace Electronics：3.0%
Other Video：2.9%
Wireless Communication：2.6%
Safety：2.6%
Compute：1.9%
TV：0%
Digital Set-Top Box：-2.9%
video game consoles/handhelds：-4.0%
Digital Camcorder：-5.5%
Digital Still Camera：-7.1%
(Blue Laser)DVD Player/Recorder：-12.0%
Portable Media Players：-15.9%
7
Consumption of automotive electronic Head mounted displays will be the Communication electronics include
components for safety, infotainment, main driver of semiconductor growth wired and wireless electronics. In
navigation and fuel efficiency will in consumer electronics. In addition, wireless electronics, traditional phone
increase for years to come due to wearables and smart watches will be and cellular modems will experience
ever-more electronic components new point of growth. However, other large declines, while smart phone
being applied in advanced safety consumer electronics markets, such demand will increase only slightly.
features added to vehicles. Among as DVD and portable media players, Revenue growth from the wireless
applications driving semiconductor will see sharp declines. Hence, the market could be quite slow. In wired
growth, advanced driver-assistance overall revenue increase of consumer communication electronics, enterprise
systems (ADAS) will have the largest electronics will be somewhat limited. WANs (wide area networks) deployed
increase. They will drive demand for as appliances should be the fastest-
ICs (Integrated Circuits), MCUs and Data processing electronics include growing segment.
sensors. computing and storage. Storage,
especially SSDs (solid-state drives), APAC appetite remains strong
Industrial electronics encompass will account for the largest increase. The Asia Pacific region will remain
security, automation, solid-state As the price declines seen in 2018 the world's biggest market for
lighting, transportation and continue, the trend towards greater semiconductor consumption. An
energy management. Security SSD adoption and increased average increasing proportion of Chinese
is the most substantial driver of capacity will retain strong momentum; products, which is stimulating the
industrial electronics. New memory particularly in enterprise SSDs as growth of the whole Asia Pacific
technologies enhance the energy data center demand will remain a key market, will be the major contributing
savings, security and functionality of driver. factor. In addition, more M&A will
IoT devices. benefit semiconductor sector growth
going forward.
Figure: Semiconductor sales by region (2018) Figure: Semiconductor sales growth (2018)
North America 19.60%

14%
10% Americas
Asia Pacific (exclude Japan) 14.50%
EMEA
Japan
6%
Asia Pacific
(exclude Japan) EMEA 13.90%
70%
Japan 9.30%
Source: Gartner
8
In terms of growth, the US had the Figure: Semiconductor sales in East Asia
fastest growth rate in 2018 due to
the rise of DRAM and high demand
for MCUs, especially in the storage
market. The Asia Pacific region
China Mainland Korea Taiwan Japan
has benefited from a boom in the
memory market as the rising price of
Source: Gartner
memory generated great revenue.
The integrated circuit (IC) industry
in Mainland China grew 24.8%,
making a major contribution to China playing catchup of the IT industry. By increasing
Asia-Pacific. Korean semiconductor In East Asia, Japan has held a leading added value, Taiwan should solve the
industry growth depends mainly position in the semiconductor downside in the IC design industry
on IC suppliers, particularly in the R&D and materials industry, with caused by a lack of capital and
memory chip market. On the other semiconductor giants including talent investment. China is taking up
hand, the semiconductor industry Toshiba, Sony and Renesas. South Taiwan's semiconductor market share.
in Taiwan is based on the foundry Korea and Taiwan are strong in Moreover, the expanding Chinese
model. However, price fluctuations memory and foundry, respectively. Mainland market will be a business
have harmed many foundries. This South Korea leads in DRAM and NAND channel for the IC design industry and
has prompted vendors in Taiwan and has many top semiconductor its companies will continue to invest in
to transfer some foundries into companies such as Samsung and SK the semiconductor industry in Taiwan.
Mainland China and re-prioritize to Hynix, mainly thanks to government Firstly, Mainland China can provide
focus on IC design in order to buck support. In terms of NAND, it is market support. The semiconductor
the price decline. As for Japan, as increasingly difficult for new players industry in Taiwan needs to be close
semiconductor companies have split to participate in the competition to the consumer market to support
and reorganized, they have exited due to the required accumulation of product innovation and economic
the DRAM business, which has low the technical know-how. However, of scale. Secondly, talent can be
technical value, and shifted focus South Korea also has challenges. provided to allow Taiwan to focus on
to exploit high value-added system The price of DRAM is falling and higher value-added product R&D.
chips. exports are declining. Hence, South
Korean semiconductor vendors are The Chinese semiconductor industry
devoting themselves to equipment is on an upward trajectory, with
and materials research as they seek to double-digit growth. Even though
expand into other sectors to tilt away the competitiveness of Chinese
from a heavy emphasis on memory. semiconductor vendors has improved
greatly in recent years, the industry
Taiwan has become the world's still relies heavily on key components
leading location for semiconductor from the West, resulting in a self-
foundry manufacturing. Taiwan's sufficiency rate of less than 20%. The
semiconductor foundry industry Chinese government has paid close
is dominated by two contract attention to this issue, promulgating
manufacturers, TSMC and United numerous favorable policies to
Microelectronics Company (UMC). strengthen the semiconductor
Semiconductor foundry is a mainstay industry.
9
Figure: Major players in the Chinese semiconductor industry
National Team Local Team

Headed by the likes of Big Fund and Under the guidance of Large Fund,
Tsinghua Unigroup, has invested many local governments have also
hundreds of billions of dollars into the set up local versions of IC investment
semiconductor ecosystem. fund.
China
Semiconductor
Players
Returnees Team MNC
Many returnees have participated in Intel, TSMC and other MNCs have
the development of semiconductor been in China for a long time. Many
industry in China, often backed by PE/ joint venture has been setup across
VC. China.
Broadly speaking, there are four participated in the development of Looming uncertainties from the
categories of players in the Chinese the semiconductor industry in China, Sino-US trade war
semiconductor industry—the including UNISOC and Veri Silicon, 2019 is destined to be a bumpy one
"National Team", "Local Team", PEVC Giga Device and Montage etc. These for the tech sectors of China and
funds and MNCs—each of them enterprises are normally founded the US. Tit-for-tat tariffs will likely
vying to make China a world-leading by returnees, focus on IC design and continue and could even escalate,
semiconductor powerhouse. The in most cases are backed by PEVC. if no agreement from both side
National Team is spearheaded by Among MNCs, Intel, TSMC and many can be reached in areas such as
Large Fund and Tsinghua Unigroup, other foreign-invested companies intellectual property, technology
which have invested hundreds of have been in China for a long time. transfer and cyber-attacks. The
billions of dollars in the value chain. In recent years, a growing amount of hardest hit sector in the trade war
The "Local team", on the other hand, foreign capital has paid attention to will be semiconductors, where the US
acts under the guidance of Large China opportunities. Global Foundries imports USD2.5 billion worth of goods
Fund, with many local governments has a factory in Chengdu. ARM and a year.
having set up investment funds such Qualcomm have established joint
as the Beijing IC Industry Fund and ventures in China. In particular, after
Shanghai SummitView Capital Fund. Taiwan loosened restrictions on
These local semiconductor funds high-tech investment in the mainland,
are estimated to have exceeded TSMC set up factories in Nanjing and
RMB200 billion. In the PEVC system, Unicom entered Fujian via Jinhua.
many overseas returnees have
10
Figure: China imports: IC chips vs Crude Oil Figure: US deficits with China
Computers & electronics $167.3

300
Electrical equipment $39.9

250
Misc.manufacturing $38.6
200
Apparel $29.3
150 Machinery $25.7
Furniture $23.4
100
Fabricated metal $20.3
50
Leather $19.8
0 Plastics & rubber $15.6

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Textiles $11.6
Crude oil imports (billion dollars)
Chip imports (billion dollars) 05 01 00 1502 00
Source: Deloitte analysis
Currently, China imports over USD200 In the grand scheme of things, however,
billion worth of IC chips from the US, far the short-term impact of the Sino-US
exceeding its imports of crude oil. Any trade war on China's high-tech industry
hiccups in the semiconductor value chain may not be as big as it appears, since
will have a severe impact on other parts of the majority of IC chips made in China
the industry. As the trade war continues are absorbed domestically. The back-
to intensify, a handful of domestic and-forth trade war will in some ways
semiconductor giants and MNCs have force Chinese enterprises to innovate
begun to re-evaluate their positions in independently and accelerate the process
the supply chain. Apple, for example, has of domestic substitutions to ease future
long used China as its production base for risks.
everything from its signature iPhones to
iPads accessories. The company's supply
chain now spans hundreds of companies.
However, these suppliers may consider
shift some iPhone production away
from China should tariffs on US imports
continue to rise.
11
2
The future of
automotive
semiconductors
Surge in automotive electronic regulation and customer demand, the
systems adoption of safety-related electronics
The automotive industry has come a systems has exploded. Most of the
long way in terms of providing factory- automotive innovations taking place
installed electronics meant for safety today arise from electronics rather
and comfort. Back in 2004, only a than mechanics. The cost contribution
quarter of cars had airbags and less of automotive electronics increased
than 50% had factory-installed power from around 20% in 2007 to about
seats. However, due to government 40% in 2017.
12
Semiconductors – the Next Wave | Future of automotive semiconductors
Figure: Factory-installed electronics
Interior Voice/Data
Lightering Communications
Accident Entertainment
Recorder Auto- Active System
Driver Dimming Cabin Noise Cabin DSRC
Battery
Alertness Mirror Suppression Environment Management
Monitoring Controls
Instrument Event Repair
Night Cluster Data
Vision Recorder Lane Correction
Head-Up
Display Electronic Toll
Collection
Windshield Digital Turn

Parental Wiper Control Signals
Controls
Navigation
Engine System
Airbag
Control
Deployment Differential
Adaptive Front EV/HEV

Lighting Security System
Adaptive Cruise Active Exhaust
Control Noise Suppression
Automatic Active Suspension
Braking
Hill-Hold
Electric Power Regenerative Control
Steering Braking
Electronic Throttle
Control Antilock
The Pressure Braking
Electronic Valve Monitoring
Timing Idle Stop/
Start OBDII Electronic
Drive Blindspot stability
Shaft Parking
Direction Seat control system
Cylinder Position
De-activation Remote Lane
Active control
Keyless Departure Active Yaw
Vibration
Entry Warning Control
Control
Transmission
Control
ADAS: uses sensors installed on the EV/HEV: hybrid vehicle that combines a Body: represents for vehicle components
vehicle to sense the surrounding conventional internal combustion engine itself, including power door, power
environment at any time of driving (ICE) system with an electric propulsion window, climate control and mirror wiper
process, which can collect data, detect system. control.
and track potential dangers.
Instrument Cluster: a control panel Chassis : the framework of an artificial
Powertrain: refers to series of parts and usually located directly ahead of a object, which supports the object in its
components that generate power on the vehicle's driver, displaying instrumentation construction and use.
vehicle and transmit it to the road surface. and controls for the vehicle's operation.
Infotainment: provides a combination of
Safety: includes active and passive safety Aftermarket: concerned with the information and entertainment based on
systems that reduce the risk of accidents, manufacturing, remanufacturing, vehicle integrated information processing
as well as their consequences. distribution, retailing, and installation system.
of all vehicle parts after the sale of the
automobile by the original equipment
manufacturer (OEM) to the consumer.
13
The automotive industry has come a long way in terms of providing factory-installed electronics meant for
safety and comforts.
100% 100% 100%

80% 80% 80%
60% 60% 60%
40% 40% 40%
20% 20% 20%
0% 0% 0%
ABC Stability Controal Side Airbags
2004 2017 2004 2017 2004 2017
100% 100% 100%

80% 80% 80%
60% 60% 60%
40% 40% 40%
20% 20% 20%
0% 0% 0%
Rear Object Sensors Rear Object Sensors Blind Spot Sensors
2004 2017 2004 2017 2004 2017
Electronic systems as % of total car cost
2030F 50%
Advanced driver assist Active-passive safety

2010 35%
Powertrain Radar/ vision infotainment
2000 22% Airbag ABS/ESP
1990 15%
1980 10% Electronic fuel injection
1970 5%
0% 10% 20% 30% 40% 50% 60%
14
The cost of semiconductor content, from a surge in demand for various

i.e. the components that make semiconductor devices in cars, such
up electronic systems, has grown as MCUs, sensors, memory and more.
from USD312 per car in 2013 to By 2022, the figure is expected to
around USD400 today. Automotive reach close to USD600 per car.
semiconductor vendors are benefiting
Figure: Cost contribution of automotive electronics and semiconductor

content per car
700 Semiconductor price 50%

(per car) 45%
45%
600 40%
Automotive electronics cost 600 40%
(% of total car cost)
500 35%
475
27% 30%
400
25%
300
18% 300 20%
200 15%
150 10%
100
5%
0 0%
2000 2010 2020 2030
Source: IHS, Deloitte Analysis
Semiconductor vendors play a entire ecosystem for years to come.

critical role in the automotive Advances in technology such as AI,
industry supply chain. In the electric vehicles (EVs), autonomous
classic automotive ecosystem, driving, energy storage, and cyber
semiconductor vendors sell to security; social awareness of topics
Tier 1 electronic systems vendors, such as safety and ride-sharing;
which then integrate technology environmental concerns like pollution;
into modules and send these to the and economic considerations
automaker (OEM) for assembly. In including infrastructure spending and
recent years, the automotive industry growth in Asian markets are all set to
has been undergoing massive reshape the automotive industry.
transformations that will re-invent the
15
Figure: Role of semiconductors in the automotive ecosystem
~10% <10% ~40%-70% <10% ~20% Cost for

automakers
R&D & Material Parts Parts Marketing

Assembly
Design Supply Sourcing Integration & Services
Material Component Distributor &

Standardizers Integrators OEM
Suppliers Specialists Exporter
Semiconductor role
Tier 2 Tier 1
EDA Semiconductor Electronic OEM
Vendors Systems
IP Vendors Bosch, Delphi,

Nvidia, NXP, Kostal,
GM, Ford, Tesla,
Renesas, Harman,
Audi, BMW,
Panasonic, Denso,
Honda, Toyota,
Foundry Toshiba, Siemens,
Nissan, Volvo,
Infineon, Continental,
Hyundai,
STMicro, TI, Valeo, Gentex,
Daimler
Samsung Johnson
Control, Magna
OSAT
$400M $30B $120B $2T

CAGR: CAGR: CAGR: CAGR:
4% 6% 5.5% 5.5%
Source: Cadence, IHS, Deloitte analysis
16
Key words: Automation, electrification, connectivity

and security
Four mega trends will result in more semiconductor content
being added to automotive electronics and subsystems in the
next decade.
Figure: Mega automotive semiconductor trends
s
do r
en Au
o rV to
m
u ct Au t
o ak
ic D r i v no m er
nd l e c t r le s ing o
E c h ic ( A us
ico
s
Ve DA
m
S)
Se
Au
n
to
t io
m
ic a
at
I n te
Elec anced
ic s
io Ar tif ence
rif
n
t ro n
ct
llig
e
i c ia l
El
Ad v
Telco
or s
Co
Di e c t i
d
end
ce y
nn
gi t v i
Pl a
n
va r i t
In f t fo
al t y
hV
da t y
yer
Ad e cu
Pla
s
ot r m
an r i
rd
S
St ecu
ai
s
m
Te
n
s e
S
L ig h t w e ig h t
Technolog y
L ig h t y
weight Technolog
Trends Players Technology Transfomation
17
1. Automation Safety is a key selling point for will play a critical role in the shift to
Widely considered as the ultimate autonomous vehicles. However, autonomous driving. More sensors
goal for the future of mobility. getting to full autonomy (Level are required to enable autonomous
Automakers and Tier-1 suppliers, as 5), requires advancements in capabilities. The number of sensors
well as technology providers (e.g. technologies such as ADAS safety used in vehicles will increase as
semiconductor vendors) and smart systems that can reduce the number autonomy levels rise. At Level 4, the
mobility companies (e.g. ride-sharing of traffic accidents, including number of sensors could reach 29.
companies) not traditionally involved electronic stability control, lane These features will not be limited to
in the automotive industry, are all departure warning, anti-lock brakes, high-end cars, but will move into mid-
racing to develop and invest in related adaptive cruise control and traction range and economy models where
technologies. Semiconductor vendors control. All of these require complex volume movement is going to be
in particular are actively developing electronic components that include much higher in the next few years.
a wide-range of microchips, fusion high-speed processors, memory,
and system-on-a-chip devices controllers, sensors and data links
incorporating AI and machine learning to ensure the reliability and safety
technologies. of a vehicle. For example, sensors
Figure: Number of sensors for different levels of autonomous driving
AC C: Active Cruise Control Level 5 (2040)

LDWS: Lane Departure Warning System
AP Anywhere
LKA: Lane Keep Assist
PA: Parking Assist
AEB: Automatic Emergency Braking
Level 4 (2028)
DM: Driver Monitoring
AP: Auto Pilot Sensor Fusion
TJA: Traffic Jam Assist
~32
AP Highway
Ultrasonic 10
Radar LRR 2
Level 3 (2022)
~29 Radar SRR 6
AEB DM Camera for surround 5
Ultrasonic 10 Long distance cam 4
TJA Radar LRR 2 Stereo camera 2
Level 2 (2015) Radar SRR 6 Ubolo 1
Level 1 (<2012)
PA LKA ~13 Camera for surround 5 LIDAR 1
Long distance cam 2 Dead reckoning 1
ACC LDWAS
Ultrasonic 4 Stereo camera 1
~6 Radar LRR 1 Ubolo 1
Ultrasonic 4 Radar SRR 4 LIDAR 1
Radar LRR 1 Camera for surround 4 Dead reckoning 1
Camera for surround 1
18
2. Electrification 3. Digital Connectivity especially have an advantage in this

The need for cars to be more Then there is digital connectivity— area, due to their core capabilities
fuel-efficient, and the impetus advanced vehicle connectivity (V2I, and aggressive capital investment.
of government regulation to V2V, in-vehicle)—that connects both From a consumer's perspective, in-
lower emissions, are driving up the inside and outside of a car and car digital connectivity and content
semiconductor demand for both makes it a part of the Internet of will become the norm, particularly
traditional vehicles and EVs/hybrid Things. Automakers have begun to with Asian customers who consider
electric vehicles (HEVs). In today's provide operating systems (OS) to act entertainment a basic automotive
traditional internal combustion (ICE) as platforms for potential app stores, function with flawless integration
engines, there is still huge potential to as well as developed tailored apps, of their personal mobile devices.
lower CO2 emissions. Many sensors services and media content. Digital Connectivity is not just entertainment;
and controls are required to run an technology players are adapting vehicle-to-vehicle communication is
engine efficiently and significant their mobile platforms to cars and a critical technology for autonomous
improvements can still be achieved. developing in-car entertainment cars to realize driverless technology
The Chinese government, for example, platforms. Some media-streaming and avoid accidents. It is expected
will introduce the new "6A" emission services and device manufacturers that by 2023 over 90% of vehicles
standard by 2020 to further lower have formed partnerships with produced will be connected.
vehicle emissions. automotive OEMs. Digital tech players
At the same time, EV/HEV advances

require progress on the electrification Figure: Components of the connected car
of power trains. Governments in
many countries are beginning to
ban, or in the process of banning,
ICE-based cars completely. China is
Vehicle
targeting increased EV production by communication
automakers (10% of total production Firmware busses
from 2019). Many global automakers

have also set goals to have EVs
account for 15%-25% of their sales
Advanced/
within 10 years, thereby bringing autonomous Mobile
vehicle applications
EVs to the masses. As major OEMs systems
pursue their EV goals, growth of the
semiconductor industry will increase
Wireless
proportionally. Thus, conditions communications
are ripe for innovation in electronic
Connected
powertrain technologies that will vehicle
be used to reduce total emissions, services
thereby accelerating more demand for

semiconductors.
Infotainment Integrated
systems vehicle security
19
4. Security vendors today are more focused than secured with no introduced back
As vehicles become more connected, ever on the safety and reliability of doors or Trojans, assess vulnerabilities
hardware and software platforms vehicles. in software and firmware, provide
will be increasingly exposed to over-the-air (OTA) updates and have
hacking risks. A failure in one vehicle Two levels of protection are being communications links.
component could have an avalanche considered to deter any potential
effect. For example, if vehicle threats. The first is developing policy Future states of mobility will
communications were attacked and establishing cybersecurity coexist
maliciously, advanced vehicle systems standards, so that manufacturers We envision that these
would not be able to receive vital can follow a rigorous process for transformations will create four
situational awareness information and establishing the safety of connected co-existing states in future mobility,
integrated vehicle security systems vehicles. Standards alone may not be where vehicle ownership diverges
(control braking, acceleration and sufficient, however, as automakers between the shared and the personal,
crash avoidance, etc.) would fail to and technology players will need to while vehicle control continuously
react. Thus, automotive electronics build components that are highly moves towards complete automation.
Figure: Future states of mobility
Vehicle
Control
Technology Landscape
Level 3 & Level 4 of Tech players and startups play
1 autonomy. Advanced a critical role with automakers.
Autonomous
driver assistance system Autonomous vehicle become

(ADAS). Cheaper economical.
1. Private 3. Shared technology.
Autonomous Autonomous
Connected vehicle. Level Little change to the existing
2 0 & Level 1 of ecosystem. Automakers
autonomy, adaptive continue to push unit sales.
speed steering, ABS, Tech players emerges.
speed control.
On-demand mobility High asset utilization, high

3
Driver-assist
services, entertainment, efficiency autonomous ride-

navigation, apps. sharing network. Greater
collaboration between players
2. Private 4. Shared in disruptive environment.
Automation Automation
Mobility and data-driven Ride-sharing proliferates.
4 services, mobile devices Multi-vehicle households
such as smartphone on reduces. Tech players to offer
Vehicle the rise. more personalized services.
Ownership
Private Shared
20
In Private-Autonomous (state 1), automakers, technology players, fleet vehicles, such as ADAS and in-
vehicle ownership remains private, owners and regulatory bodies to vehicle infotainment (IVI), becoming
but the automation level has achieved create a complex urban ecosystem. an important growth market for
an all-time high due to decreases semiconductor players.
in the cost of technology, thereby In Shared-Automation (state 4),
boosting the number of autonomous where vehicle ownership is shared, a Asia Pacific the most attractive
vehicles. Collaboration between OEMs rapid increase in adoption of shared region
and technology players is the norm in mobility will emerge. Point-to-point Overall automotive semiconductor
this state. transportation will be created via revenue is projected to reach USD40
ride sharing and cost per-mile will billion (an all-time-high) in 2018,
In Private-Automation (state 2), where decrease. The landscape in this state before hitting USD60 billion in 2022.
vehicles are personally owned with will allow technology players to offer The Asia Pacific region is expected
limited ADAS. Auto OEMs will continue more enhanced and personalized to lead the world with 41% growth,
to focus on unit sales and bring in services. driven by supportive government
incremental technological advances. policy and consumer demand for
There will be little change in the Opportunities in automotive safety, particularly in China. China
ecosystem. semiconductors are emerging is now the largest market for the
Although mobile is, and will remain, automotive industry, with almost 29
In Shared-Autonomous (state 3), the largest market for semiconductor million cars sold in 2017. China is also
the majority of cars are shared with companies, overall growth has becoming a major auto-manufacturing
autonomous capabilities. In this been saturated for many years. The center, attracting global automakers,
landscape, there will be a rise in only exception is the automotive accounting for close to 29% of
on-demand mobility services such semiconductor segment, where worldwide light vehicle production. All
as entertainment, and we will see demand is strong as more electronic this makes the region highly attractive
deepened collaboration between components are being added to to semiconductor vendors.
Figure: Automotive semiconductor revenue and unit production by region
65
Automotive semiconductor revenue(billion of US$)
55
45
35
25
15
-5 2016 2017 2018 2019 2020 2021 2022
Powertrain Infotainment Body EV/HEV Chassis Safety Aftermarket ADAS Instrument Cluster
21
ADAS to lead growth sensors and power semiconductors,

The growth of the automotive because in hybridization and electrical
semiconductor market depends vehicles, semiconductors are the
on electronic equipment and primary driver for the efficiency of the
semiconductor content growth electric drive train.
in vehicles. The ADAS application
segment is expected to have
the highest growth rate. ADAS
Asia/Pacific
semiconductor content will rise
along with the level of automation.
In fact, partially automated cars will EMEA
have about USD100 in added-in
semiconductor content, with a highly Americas
automated car potentially having
USD400 added. A fully automated
Japan
car would have about USD550 of
semiconductor content. In other
0.00% 10.00% 20.00% 30.00% 40.00%
segments such as powertrains, there
is huge demand for microcontrollers, Source: Gartner, Deloitte analysis
Figure: Average ADAS semiconductor content at different levels of automaton
Partial automation High automation Full automation
Level 2 Level 3 Level 4
$400 10% $550

12%
20%
35% 8%
60% $100 5%
25%
45%
40%
40%
a
M
da
da
da
da
da
da
or
or
er
er
er
sio
sio
sio
O
O
at
at
m
m
Ra
Li
Ra
Li
Ra
Li
lB
lB
lB
Fi
Fi
Fi
tu
tu
Ca
Ca
Ca
ta
ta
ta
r
r
Ac
Ac
so
so
so
to
to
to
n
n
Se
Se
Se
Source: Infineon
From a device perspective, different will generate large demand for example, when vehicles reach Level
types of components will be sensors and microcontrollers as 4 or 5, systems need to be able to
required for increasingly complex well as processing for each of these process all sensor data to create a
functionality sets. Some segments sensors. The industry is also in the holistic view that allows the car to
will grow faster than others do. process of developing more powerful make the right judgement.
For example, autonomous driving MCUs/MPUs to handle the data. For
22
Figure: Automotive semiconductor application and device growth (2022)
Segment Growth Size Sub segment

(2017-2022) (2022)
Blind spot detection/Collision warning/Park

ADAS 23.6% 11b
assistance/V2X/Front view camera
Aftermarket 6.1% 3b Vehicle parts/Equipment/Service repair/Collision repair
Power door/Power window/Climate control/Mirror wiper

Body 13.3% 12b
control
USD 600
by Application （Gartner)
semiconductor Chassis 6.1% 7b Suspension/Differential/drive shaft

content per car
by 2022
EV/HEV 21% 6b Hybrid vehicle
Infotainment 8.2% 9b Connectivity/Telematics/Car navigation/Car audio
Instrument Cluster 11.2% 4b Instrument panel, instrument wiring harness
Powertrain 3.4% 6b Engine Control/Transmission
Safety 2.6% 5b EPS/ABS/Airbags/Traction control/Tire pressure monitor
General-Purpose Logic IC Data Converter, Switch, Multiplexer, Voltage Regulator,

12.5% 2b
Reference
Memory IC 10.9% 4b DRAM, Emerging Memory, Flash Memory, NAND
Optoelectronics 18.9% 8b Image Sensor, LED, Photosensor
Discrete 9.9% 9b Power transistor, Diodes

by Device（Gartner)
Nonoptical Sensors Environmental Sensors, Fingerprint Sensors, Inertial

7.2% 6b
Sensors, Magnetic Sensors,
Micro-Component IC 7.7% 11b Digital Signal, Microcontroller, Microprocessor
Data Converter, Switch, Multiplexer, Voltage Regulator,

Analog IC 7.4% 4b
Reference
ASIC 6% 3b ASIC
ASSP 12.3% 19b ASSP
23
Seizing the opportunities these opportunities, leaders in the failure rate might be acceptable in
The automotive segment is not really semiconductor industry should the mobile space for suppliers, but
a new market for semiconductor consider the following actions and automakers will want less than one
vendors. In fact, many of them have approaches to the market: defective part per billion (DPPB) for 15
been participating in this market for to 20 years.
several years. However, it was not 1. Understanding the automotive
considered a big revenue source since sector's unique hurdles Furthermore, in the mobile world,
there were not enough customers, When it comes to requirements, the designs regularly reach 3GHZ,
it took a very long time to qualify automotive electronics market is where frequency and speed are
a process and volume was low vastly different from the consumer top priorities. On the other hand,
compared to consumer electronics. electronics market. For example, automotive frequencies and speeds
However, sentiment has changed consumers seek the latest, leading vary. But, as 5G-connected cars
due to rising demand for automotive edge technologies in cellphones. soon become a reality, the two
electronics such as ADAS, AI, However, in automotive, some domains might eventually merge, with
connectivity and sensors. To seize sensors are still being manufactured connected cars in a way become cell
at 150nm. This is because automotive phone-like. Automotive chips also
design requires substantial need to operate at a wider range of
redundancy and size is less of a temperatures: -40°to 155°centigrade
concern, so there is little incentive versus 0°to 40°centigrade for a
to move to 7nm sensors like those in mobile device. Voltages also vary
cellphones. In addition, the failure rate greatly between automotive and
requirement is much more stringent mobile. In mobile applications,
in the automotive space, because voltages are kept low to conserve
when a cellphone goes down a simple battery life. However, there are
reset can fix the issue, which is not high voltages in cars and many
possible for a car on the road. A 10% semiconductors are still analog, which
requires devices to operate correctly
in a wider range.
24
Figure: Automotive IC requirements versus mobile This is especially important since

future autonomous driving will rely
on components working together
perfectly. How devices will fare over
time in a harsh external environment
28nm  7nm Process >180nm  7nm
is yet to be determined. A typical
100M+ Design Sizes(μP/C) <1M  1000M+ car model can be on the market for
10-15 years, a refresh cycle that is
900MHz  2.7GHz Frequencies 30MHz  5.9GHz much longer than that in consumer
electronics.
0.5V to 1.8V Voltages ~1V  >60V
0  40℃ Temperatures -40  155℃ It will be incumbent on semiconductor

vendors to conduct more simulation,
1-3 years Operation Lifetime ~10-15 years inspection and testing of larger
sample sizes to ensure reliability
<10% Target Field Failure Rates Target Zero Failure
while controlling time and added cost
throughout the process. Suppliers
Source: Synopsys, Deloitte analysis
have to be ready and plan for
long product life cycles to support
manufacturing and maintenance.
2. Expect stringent and These requirements create an
challenging qualification assortment of challenges for 3. Leverage M&A to jumpstart
requirements semiconductor vendors. For example, market entry
In the automotive industry, semiconductor foundries normally For many semiconductor vendors,
chips, components, modules and develop a process technology and gaining access to the automotive
subsystems must meet stringent qualify it using a relatively normal semiconductor market is not a
specifications on quality, reliability, sample size. In automotive however, trivial task. They must weigh the
cost, power and safety, since foundries need to conduct more pros and cons of building in-house
operating conditions in cars are much inspection, testing and screening to capabilities versus acquisition. With
more demanding than in typical achieve higher levels of quality and semiconductor vendors seeking
consumer electronics. For instance, yield. Latent reliability is another new areas of growth and expansion,
the Automotive Electronics Council focus, in which failure may not M&A deal activity has been quite
(AEC) has developed a failure stress occur until components have been active over the last few years. For
test qualification (AEC-Q100) for operating out on the road during example, NXP and Freescale's
integrated circuits (IC). Qualified ICs a vehicle's useful lifetime, due to merger enabled the resulting entity
should met a series of reliability stress age, latent manufacturing defects, to become a leader in automotive
tests (electrical, lifetime, etc.), as well thermal stress or electromagnetic semiconductor solutions, while Intel's
as be subjected to testing under interference. acquisition of Mobileye allowed it to
various intense temperature ranges. enter the market and filled gaps in its
Electronics and semiconductor automotive product offering.
suppliers are expected to support
qualification testing pre-shipment as
well as failure analysis post-shipment.
Ultimately, product development and
manufacturing are all linked back to
safety regulations and everything
should be traceable.
25
Table: Automotive semiconductor related M&A
M&A Amount (USD) Date
NXP/Freescale 118 million March 2015
Avago/Broadcom 370 million May 2015
Intel/Altera 167 million June 2015
Western Digital/Sandisk 178 million October 2015
Analog Devices/Linear 130 million July 2016
Softbank/ARM 302 million July 2016
Qualcomm/NXP 459 million October 2016
Intel/Mobileye 153million March 2017
Source: Mergermarket, Deloitte analysis
Thus, M&A should be considered an the best chance of producing high is also seen by Samsung as a new
integral part of an overall strategy quality components. M&A allows and profitable revenue stream.
to maintain competitive edge. The quicker access to this expertise to Samsung will expand its reach across
potential benefits M&A brings include strengthen competitiveness and the connected car market, ADAS,
filling gaps in product line, accessing increase market share. cybersecurity and over the air (OTA).
state-of-the-art technology and The acquisition also aligns with
gaining access to a wider customer Oversaturation of the smartphone Samsung's IoT universe strategy to
base. These are especially important market has pushed many big achieve volume.
in the automotive semiconductor semiconductor players to explore
industry, because it takes great effort possibilities in the automotive Meanwhile, Panasonic is refocusing
to secure long-term partnership semiconductor market to broaden its work from home electronics to
agreements with automotive their portfolios and diversify revenue high-tech auto parts, leveraging
companies. Added to this is the fact streams. The connected car provides its expertise in electronics to
that most automotive semiconductor an entry point for the expanded role build advanced capabilities in the
projects are lengthy, often taking semiconductor vendors can play. automotive electronics market.
years from concept initiation, product Panasonic has made several
development and qualification For example, Samsung acquired acquisitions in the last few years,
process all the way to production. Harman to build its infotainment coupling these with building
In addition, customers are seeking capability in connected cars. The autonomous driving capability
integrated, comprehensive solutions move allowed Samsung to gain a in-house. Panasonic has begun
from one company instead of many. foothold in the automotive market, testing autonomous cars, teamed
as Harman is a global leader in up with Google and Qualcomm on
From a technology perspective, connected autonomous vehicles. infotainment, and formed a joint
automotive semiconductor The acquisition not only allowed venture ( JV) in China to provide key EV
components have to adhere to Samsung to position itself as a key components.
stringent quality standards, and supplier of infotainment systems
semiconductor vendors with cutting- using an established brand name,
edge manufacturing capabilities have but the autonomous vehicle market
26
4. Rethink collaboration models Players' roles are shifting and new or JVs, the roles in the supply chain
and roles players are rising. Customers are are blurring. As vehicles become more
The ecosystem and collaboration becoming partners or competitors autonomous, systems within them are
models of the traditional automotive and vice versa. From acting as a ever more connected and cannot run
semiconductor are no longer static supplier and forming strategic in isolation.
but more intertwined than ever. partnerships to entering M&A deals
Figure: The evolving automotive semiconductor supply chain
Classic automotive semiconductor supply chain New automotive semiconductor supply chain
Tier 1
Electronic
Systems
Tier 2 Tier 1
Semiconductor Electronic OEM
Vendors Systems
Tier 2
Tech & Digital
Semiconductor OEM
Players
Vendors
27
This market landscape adds a layer applications. New tech entrants

of complexity to the existing supply are positioned to take the lead in
chain. Some automakers are now software-focused elements. Digital
designing their own IC (e.g. Tesla) and players are adapting their smartphone
extending their activities beyond core platforms to car-specific customer
hardware business into the provision needs, integrating infotainment OS
of software OSs designed to serve as and software platforms into car
platforms for potential app stores, as systems and cars' human-machine
well as developing specific apps and interfaces. Media-streaming
other services or media content. services and end user equipment
Other Tier 1 players are also manufacturers have already formed
designing ICs and getting involved in partnerships with some automotive
software. Continental's purchase of OEMs. Semiconductor vendors
Elektrobit is an example of this trend. can expand their roles by not only
Furthermore, semiconductor vendors serving traditional automakers and
are now developing electronic control their suppliers, but also through
units, as are some IC houses. collaborating with technology vendors
to offer more products.
Automotive suppliers are also trying
to become less dependent on OEMs At the same time, semiconductor this involved a few critical automotive
by establishing direct relationships vendors are increasing collaboration products such as infotainment
with end customers. For example, with both automakers and tier-one and display drivers, with critical
Bosch's app allows customers to automotive suppliers. For example, components such as power train
monitor car functions and provide Nvidia and Audi are collaborating or chassis control components still
a direct connection to the nearest to build an AI platform with deep made by IDMs themselves. However,
Bosch repair center. learning technology for autonomous the trend is shifting, as IDMs are now
driving. Utilizing neural networks outsourcing critical applications to
Technology players are applying to understand the surrounding foundries. For example, ADAS requires
their existing capabilities to digital environment and determine safe advanced MCUs but many IDMs do
automotive platforms. Tech players routing to incorporate into Audi's not have the in-house capabilities to
have significant advantages, including line of L3 cars. This type of strategic create them.
capabilities, operating models and partnership brings in skills from both
capital for aggressive investment, side that complement one other and
as they focus on making horizontal create a mutually beneficial situation.
moves to create new revenue
models. Several high-tech players Meanwhile, many IDMs have stopped
are developing autonomous-driving building leading-edge fabs and
systems that are quite likely to merge turned to "fabless/fablite" over the
into operating systems. Leading last 20 years due to leading-edge
online and technology companies fabs' prohibitive cost. But, they have
are focusing on in-car entertainment in general kept their proprietary
platforms, which they hope will processes in-house while outsourcing
become the standard for such some production to foundries. Initially,
28
5. Keep an eye on start-ups and vibration and shock, wear and tear,
their disruptions and cleaning that is normal to the
Start-ups are a hotbed of activity harsh conditions of driving. Finally,
in connected cars, VRV/V2X there are still edge cases to be
communication, mobility services, resolved, such as bright sun against
cybersecurity and the AI/machine a white background, blizzards that
learning space. Investments in causes whiteout conditions and early
these auto tech companies have morning fog.
also seen huge increases in recent
years. There is a wide variety of There are several reasons why
startups in the industry to address semiconductor vendors should
different pain points. For example, pay attention to startups. First, the
new hardware startups are looking emergence of these startups in the
at ways to address the pain points connected vehicle space affords
associated with light detection and the opportunity to cooperate
ranging (LiDAR), which is integral to with them. Second, gaining access
the safety of autonomous vehicle. to technologies such as digital
First, LiDAR is still quite expensive and connectivity and AI will be increasingly
many big automakers do not use it; important for semiconductor vendors
second, (dynamic) range is a big issue and OEMs, as R&D cost and risk are
with the technology, i.e. how near, high. Third, these startups could
far and widely can LiDAR accurately be potential acquisition targets for
build a 3D picture from millions of semiconductor vendors to acquire
pixels. Robustness is another issue, innovative technologies or enter niche
with LiDAR having to withstand the markets.
Figure: Automotive startups' areas of specialization
Driver
Connected Car V2V/V2I/ V2X assistance/ Fleet Telematics
robotics
Safety Mobility LIDAR/ mapping

Cybersecurity
tools/system Services systems
Machine Computer
Autonomous
learning/ AI/ Vision/ UX/ Design
platforms
Deep learning/ Detection
29
3
The race is on for AI

chips
The race for AI chip domination "Depth" refers to the layers and
The artificial intelligence framework number of nodes in a neural network
can be broadly characterized into model. Over the years, the complexity
three layers. The infrastructure layer between layers and the number of
includes the core AI chips and big nodes have grown exponentially.
data that support the sensing and This presents a significant challenge
cognitive computational capabilities of for computation. Traditional central
the technology layer. The application processing units (CPUs) excel at
level sits at the apex, providing general workloads, particularly if they
services such as autonomous driving, are rules-based. However, CPUs can
smart robotics, smart security and no longer keep up with the parallelism
virtual assistance. AI chips form the required of AI algorithms.
heart of the AI technology chain
and are central to the processing of
AI algorithms, particularly for deep
neural networks (DNN).
30
Semiconductors – the Next Wave | Race is on for AI chips
Figure: AI chips' role in the layers of AI

Flexibility
• Robotics
• Autonomous Driving CPU
• Business Intelligence (Intel)
• Smart Factory
Application • Personal Assistance
• Customer Service
……
GPU
• Machine Learning (NVIDA,AMD)
• Deep learning
Technology • AI Platform
• Speech Recognition
• Image Recognition FPGA
• Biometric (Xilinx, Altera,
…… Microsoft…)
• AI chips
• Sensors
Infrastructure • Computing ASICs
• Data (Google, Horizon
…… Robotics, Bitmain,
Cambrian…)
Efficiency
Source: Microsoft, Deloitte analysis
There are two major ways to address GPUs used to process graphic and flexibility, especially when an
the parallelism issue. One is to add intensive tasks such as games are AI algorithm has not been finalized.
a dedicated accelerator based on built with parallelism in mind. GPUs This allows vendors to optimize
existing computational architecture. have very high performance suitable custom chips for their applications
The other is to re-develop completely for deep learning AI algorithms that while avoiding the cost and potential
to create a new architecture that require a lot of parallelism. This new technology obsolescence of the ASIC
simulates the neural networks of role makes GPUs a great choice for approach.
human brains. The latter approach is AI hardware. GPUs are now widely
still in the early phase of development used in cloud and data centers for ASIC AI chips, on the other hand,
and unsuitable for commercial rollout. AI training. They are also used in the have dedicated architecture for AI
The addition of AI accelerators will automotive and security sectors. The applications. ASIC-based AI chips have
therefore be the primary method. GPU is currently the most widely used, many variations, including TPU, NPU,
Numerous types of AI chips are most flexible AI chip available. VPU and BPU, etc. These are all aimed
available for acceleration: mainstream at diverse, computer-intensive, rules-
ones including GPUs, field- FPGAs, meanwhile, are programmable based workloads with high efficiency
programmable gate arrays (FPGA) and arrays suitable for clients that and performance and the flexibility
application specific integrated circuits want to reprogram based on their of a CPU. Typically, ASIC AI chips have
(ASIC), with variations including TPU, own requirements. FPGAs are higher efficiency, a smaller die size, as
NPU, VPU and BPU, etc. Each has its characterized by a faster development well as lower power consumption than
own strengths and weakness. cycle (versus ASIC) and low power GPUs and FPGAs. But, ASIC chips'
requirements (compared to GPUs). development cycle is longer and less
But, their flexibility makes their cost flexible, which has contributed to its
relatively high. FPGAs can be seen as a slow commercial adoption.
good compromise between efficiency
31
There are two distinct AI deployments The training phase requires a phase is typically done via hardware
in deep learning: training and tremendous amount of computational in the cloud. The inference phase, on
inference. AI utilizes big data as a power because it requires the the other hand, can be handled either
foundation to "train" neural network application of a huge data set to a in the cloud or on devices (products)
models, using training datasets to neural network model. This requires at the edge. Compared to training
obtain these newly trained models. high-end servers that have advanced chips, inference chips require more
A newly trained model is then armed paralleled performance to be able thoughtful consideration of power
with new capability to "infer" from new to process large, diverse and highly usage, latency and cost.
data sets to reach a conclusion. parallel datasets. Therefore, this
Figure: The two-phases of deep learning

DEEP LEARNING
TRAINING INFERENCE
Learning a new capability from existing data Applying this capability to new data
TRAINING NEW App or Service

Untrained Neural Deep Learning DATASET Trained Model DATA
Network Model New Capability Featuring Capability
Framework
“Cat”
“？”
Trained Model
Optimized for
Performance
“dog” “Cat”
“Cat”
Source: nVidia
Innovation in AI chips has just AI chips to enjoy explosive growth

begun, with vendors taking different The AI chip market is expected to
approaches to chip acceleration. account for over 12% of the total
Google, for example, has taken the AI market by 2022, with a CAGR of
ASIC path, whereas Microsoft has 54%. The Americas is expected to
demonstrated that comparable, lead the market, followed by EMEA
and sometimes even better results and then APAC. The Americas overall
can be achieved using an FPGA. will dominate the market throughout
Meanwhile, Xilinx, Baidu and Amazon 2022.
are all working to lower the traditional
barriers to FPGA adoption.
32
Figure: The global AI and AI chip market (2022)
120
120 100% 16 120%
14.616
97.1 14
100 100%
80%
12 10.998
80 74.5 80%
60% 10
7.445
60 51.9 8 60%
40% 6
40 4.798 40%
29.3
4
20% 2.388
20 20%
2
0 0% 0 0%
2016 2017 2018 2019E2 2020E 2016 2017 2018 2019E2 2020E
Global AI market size (billion dollars) YOY Global AI chip market size (billion dollars) YOY
Source: CDIC
Cloud-based AI chips the most promising segment

The AI chip market can be split into two categories based on deployment
methods: cloud-based and network edge.
Deployment Application Market Potential (2022) CAGR (2018-2022)
Foundry 1
Training ~17billion ~55%
Cloud-based
Infer ~7 billion ~85%
AI Chip
2 Security ~2 billion ~40%
Network
edge Automotive ~5 billion ~43%
IP Smart Phones ~4 billion ~60%
Source: CICC, Deloitte analysis
33
The cloud is the biggest market Inference AI chips in smartphones are

for AI chips, as their adoption in now part of a three-way race between
data centers continues to increase, smartphone manufactures like Apple,
enhancing efficiency, reducing Samsung and Huawei; independent
operational cost and improving chip providers such as Qualcomm and
infrastructure management. In MediaTek; and IP license providers
particular, the AI training market including ARM and Synopsys. AI chips
will reach almost USD17 billion, from smartphone manufactures
with inference chips on the cloud are typically optimized for their own
accounting for USD7 billion. By phones with the goal of improving
product type, GPUs have become the performance and user experience.
mainstream choice for AI chips and However, independent chip providers
have the largest share, accounting for could come out with better chip
over 30% of the market. specifications to compete for the rest
of the market.
Network edge AI chips are
emerging Autonomous driving the ideal
AI chip deployment is not limited to application for AI chips
the cloud, but can also be seen in a Autonomous driving is not only a
wide variety of network edge devices complex application scenario for
such as smartphones, autonomous artificial intelligence but also an
vehicles and security cameras. Most AI important one. It is expected be a
chips at the edge are inference chips, major driver for AI inference chips,
and they are becoming increasingly with an estimated market value of
specialized. The AI inference chip USD5 billion and a CAGR of 40%.
market is expected to grow at a CAGR
of 40% and reach USD2 billion by Sensing, modelling and decision
2022. making are the three-step process
normally required for autonomous
AI chips drive smartphone ASP driving, and inference chips
AI chip vendors could benefit from are involved every step of the
the rising cost of their products. For way. Whether they are used for
example, Apple's A11 chip cost has environment sensing or obstacle
risen to USD27.50. The rise in cost of avoidance, autonomous driving
AI chips will in turn increase prices of has high requirements for the
smartphones, providing smartphone computational power of AI chips.
manufactures with increased revenue.
AI chips have also moved down
from high-end phones to mid-range
ones, potentially bringing even more
revenue to smartphone vendors.
34
Owning to limitations in latency, The addition of AI can be seen as

computation in autonomous driving the "third wave" of transformation.
should ideally be done on the edge AI inference chips are now added to
rather than in the cloud because cameras at the edge, allowing real-
decisions need to be made in near time processing of video data. This
real-time. According to Toyota, saves storage space on the cloud
computational power of 12 tera and increases the performance of
operations per second (TOPS) is surveillance systems as massive
required to realize L5 autonomous amounts of data are generated on the
driving, but most chips today are edge each day.
only capable of supporting 2-3 TOPS.
There is a clear urgency for AI chips to China now the hotspot for AI chips
migrate to the edge instead of doing AI chip funding activity in China has
the heavy lifting in the cloud. been active and M&A is also on the
rise. One example is the acquisition
OEMs are testing chips from of DeePhi, a start-up with industry-
vendors to identify the most suitable leading capabilities in machine
candidates. Large OEMs have prefer learning, deep compression, pruning,
constructing their own autonomous and system-level optimization for
driving platforms and purchasing neural networks, by global giant Xilinx.
AI chips separately, whereas newer Tech giants are also getting entering
OEMs normally prefer buying the field of play, led by Alibaba, Baidu
complete autonomous driving and Huawei. In particular, Huawei
platforms. Over time, the number of has lit up AI chip competition in the
AI applications that can benefit from smartphone sector. Some Bitcoin
local processing is likely to grow, such mining equipment manufacturers are
as Apple's Face ID. also getting into the AI optimization
game.
Smart surveillance systems have
high demand AI companies in China are generally
Surveillance systems are getting swift at identifying viable business
smarter thanks to AI. The surveillance applications for AI, especially
system industry has gone through business model innovation and rapid
three phase of transformation in the implementation. However, original
past 10 years. The first wave was AI model creation capabilities are a
the "high definition" phase, where rarity among Chinese players, with
a system is capable of recording most domestic research in AI focused
ultra-clear videos. The second wave on tweaking and perfecting existing
was the "networking" phase, where models, rather than creating an
networking and interconnection original and systematic AI framework.
capabilities are added to the system. In addition, training devoted to AI is
also limited in China compared to
countries such as the US.
35
Navigating the AI scene neural networks stuck in the cloud.

The rise of artificial intelligence has Efforts are being made to create a
certainly created new opportunities new architecture that allows more
in semiconductor devices, particularly tightly coupled processors and
for AI chips. Semiconductor memory, which improve performance
companies already in or planning to and energy efficiency. The idea is
enter AI systems should adhere to to add new functions to memory
the following key observations to stay that will make devices smarter
competitive. without replacing the processor.
The semiconductor industry should
Specialization is key to AI chips: In embrace such designs so AI can be
future, AI chip companies should no brought out of the cloud and onto the
longer be content just to be hardware edge.
companies, but possess a deep
understanding of the requirement Select the appropriate
of their customers to create suitable semiconductor process
products. Customers today are not technology: Unlike CPUs, which
just looking for a general-purpose need the most advanced processing
chip with some added AI capability; technology available according to
they are looking to see whether it can Moore's law, it is not always necessary
satisfy their business requirements for AI chips to be designed on the
at a reasonable cost. They should latest process technology due to
balance power, performance and cost. AI's use of parallel processing. For
Computational density (computational example, 40nm and 28nm processes
power per unit of power consumed) are sufficient to provide 1 TOPS of
will become a core competitiveness computational power. In addition,
for AI chip vendors. older-generation processes have
access to mature toolsets and
Moving from the cloud to the building blocks. Many large foundries
edge: Opportunities have begun offer a wide range of advanced
to pop up at the network edge process technologies, ranging from
and many large players are moving 28 nm all the way to 7 nm, depending
from the cloud to the edge to offer on their mix of performance and
comprehensive AI solutions that power capability. For semiconductor
span the spectrum, from training to vendors, selection of the appropriate
inferencing workloads. It is worth semiconductor process technology
noting that most AI systems today should ideally be based on criteria
run on von Neumann architecture, such as computational capability and
in which processing and memory power consumption, as well as form
are separate. This has led to AI factor.
becoming power hungry and left
36
Semiconductors – the Next Wave | Muted tone for M&A
Software tool support is Look for opportunities beyond

essential: The amount of support AI chips: Artificial intelligence does
semiconductor companies have not rely on AI chips alone to achieve
for standard open-source software its processing power. Memory is
frameworks is key to winning the also an important component in
AI race, particularly for challengers the advancement of AI, where high
trying to catch up with leaders that throughput parallel processing
already support literally all deep places multiple strains on data
learning software and tools for bandwidth in memory systems. The
their semiconductor chips. To be demand for AI system memory will
a viable contender in this market, create opportunities for memory
semiconductor vendors should at vendors. In addition, as AI systems
a minimum support major open scale up, the performance of
source software frameworks such as interconnects between subsystems
TensorFlow, Caffe2, Theano, CNTK, and devices will become a bottleneck.
MXNet and Torch. Tools to help Thus, opportunities also exist for
developers build applications are also semiconductor vendors to create
required. Semiconductor vendors high speed interconnects to meet the
will need to invest in software and demand of high volume data flowing
collaborate with software developers between systems. Further, AI chips
to access their AI device architecture. today can contain multiple processors
The number of software frameworks to achieve maximum parallelism,
for processing neural networks is resulting in a very large die size.
growing. Many will be developed and This presents a great challenge for
released over the next few years, so thermal and high voltage power
there is still room for new entrants to management in which custom cooling
grow. solution may be required. This creates
opportunities for packaging vendors
to come up with products that have
thinner form factor and less thermal
dissipation for a more cost-effective
solution.
37
4
Muted tone for M&A

M&A activity in semiconductors is market. Regardless, semiconductor
already past its peak and specialized conglomerates are always on the
verticals such as automotive, AI and lookout for potential targets in various
network/data centers are becoming verticals that have high market share
the most sought after emerging and profit.
verticals. Japan and Korea have
been actively seeking to revive M&A enters plateau phase
their semiconductor industries Global semiconductor M&A
via acquisitions of mid-sized transactions peaked at over USD120
companies in the US and Europe, billion in 2016. In 2017, the value of
as well as cooperation with China. M&A in the semiconductor industry
At the same time, disputes over fell sharply. In addition to fewer M&A
intellectual property and national targets due to previous transactions,
security issues will dampen China's tightened regulatory reviews by
ambition to go global. The tightening Europe and the US were an important
of Chinese outbound investment in factor. M&A transaction value rose
high-tech companies from the US again in 2018 due to value per deal. For
will become the new norm, shrinking example, US-based Broadcom acquired
the overall size of the global M&A CA Technology for USD17.99 billion.
38
Figure: Global semiconductor M&A transactions (2014-2018)
140 500
120
400
100
300
80
60 200
40
100
20
0 0
2014 2015 2016 2017 2018
Value ($B, LHS) Volume (RHS)
Figure: The Top 10 Global semiconductor M&A transactions (2018)
Districted District of Value

Date Bidder Target Motivations
bidder target ($B)
2018.07.11 Broadcom Inc. US CA Technologies US 17.987 Expand industry supply chain
Microchip
Microsemi
2018.03.01 Technology US US 9.836 Expand industry supply chain
corporation
Incorporated
Renesas Integrated Search for cutting-edge
2018.09.11 Japan US 7.004
Electronic Device Technology application
Tsinghua Mainland Linxens

2018.07.25 France 2.632 Acquire advanced technology
Unigroup China Group
Will Mainland Beijing OminiVision Mainland Strengthen market position and

2018.08.15 2.178
Semiconductor China Technologies China expand market share
Beijing Mainland Beijing Yandong Mainland Strengthen market position and

2018.05.25 0.626
Electronics China Micro-electronics China expand market share
Strengthen market position and

2018.09.14 UKC Holdings Japan Vitec Holdings Japan 0.529
expand market share
United Mie Fujitsu Strengthen market position and

2018.06.29 Taiwan Japan 0.519
Microelectronics Semiconductor expand market share
Ingenic Mainland Mainland

2018.11.10 Beijing Xicheng 0.38 Expand industry supply chain
Semiconductor China China
GigaDevice Mainland Mainland Strengthen market position and

2018.01.31 Semiconductor 0.268
Semiconductor China China expand market share
Source: Mergermarket, Deloitte
39
In East Asia (China, Japan, South Korea expansion, the pace stalled somewhat
and Taiwan), M&A volume showed in 2017 and 2018. In 2017, the volume
rapid growth between 2014 and 2015, of semiconductor M&A in East Asia
amounting to over USD22 billion. declined 1%, while value grew a
However, after several years of rapid modest of 2%.
Figure: East Asia (China, Korea, Japan and Taiwan) M&A transactions (2014-2018)
Expansion Phase Plateau Phase
30 140
25 120
100
20
80
15
60
10
40
5 20
0 0
2014 2015 2016 2017 2018
Value ($B, LHS) Volume (RHS)
Figure: M&A transaction value - China, Japan, South Korea and Taiwan (2014-2018)
25
20
15
10
0
2014 2015 2016 2017 2018
Mainland China Japan South Korea Taiwan District
40
Figure：Top 10 M&A deals in China, Japan, South Korea and Taiwan (2018)
1
Districted District of Value
Date Bidder Target Motivations
bidder target ($B)
Renesas Integrated Device Search for cutting-edge

2018.09.11 Japan US 7.004
Electronics Technology application
Strengthen market
United Mie Fujitsu
2018.06.29 Taiwan Japan 0.519 position and expand
Microelectronics Semiconductor
market share
Advanced Strengthen market

Suzhou ASEN Mainland
2018.03.27 Semiconductor Taiwan 0.127 position and expand
Semiconductor China
Engineering market share
Strengthen market
Hoejin Mainland
2018.07.10 SK Hynix System South Korea 0.075 position and expand
Semiconductor China
market share
Strengthen market
Taiwan
2018.06.26 Taiwan ON Semiconductor US 0.006 position and expand
Semiconductor
market share
Strengthen market
2018.02.09 NEC Corporation Japan XON Holding South Africa 0.023 position and expand
market share
Strengthen market
Shanghai Fortune
2018.01.11 Mainland China Upstar Technology Hong Kong 0.022 position and expand
Techgroup
market share
Huada Salantro Acquire advanced

2018.09.07 Mainland China Canada N/A
Semiconductor Semiconductor technology
Strengthen market
2018.05.09 TDK Japan Faraday Semi US N/A position and expand
market share
Acquire advanced
2018.02.28 TDK Japan Chirp Microsystems US N/A
technology
China leading the way in domestic for semiconductor M&A. From 2014
M&A to 2018, the volume of M&A in China's
Regardless of the volume or value semiconductor industry ballooned
of transactions, Mainland China is from 48% to 72% in 2018 at a
undoubtedly the most active region compound annual growth rate of 18%.
41
Figure: Domestic M&A transaction volume growth – East Asia (2014-2018)
2018
2018
11 2017 2016
10 11
2015 2014
12 15
10 10 11
89 2 73 51 6 36 8 6 30 11 6
2
Mainland China Japan South Korea Taiwan
Government policy has been the Outbound M&A mixed Figure: Motivations for M&A in the
largest contributor to the rapid Overall, volumes of cross border M&A semiconductor industry
development in China over the from East Asia declined after 2016,
past five years. China today is the particularly as the US strengthened its
largest importer of semiconductor investigations of Chinese companies Acquire Strengthen
advanced market
chips globally and the government's seeking leading-edge technologies. technology position and
overall strategy is to reduce reliance In 2017, the White House published expand
on foreign imports and build up the report, Ensuring Long-Term US market share
its domestic semiconductor base, Leadership in Semiconductors, which

which has fueled Chinese companies' sets out the perceived threat posed
entrance into the industry and by China's semiconductor policy to
acquisition of advanced technologies. the United States and suggests the Search for Expand
US government take steps to prohibit cutting-edge industry
application supply chain
In domestic M&A, Mainland China is or heavily restrict Chinese acquisition,
undoubtedly the most active region as well as tighten rules that restrict
in East Asia. Its M&A volume had a the flow of critical semiconductor
compound growth rate of 24% from intellectual property. Yet, despite Source: Deloitte analysis
2014 to 2018. For example, in 2018, increasing government reviews of

Alibaba acquired Hangzhou C·SKY. M&A, North America and Europe are
Before this, Alibaba had invested still the main destinations for East
in five chip companies: Cambrian, Asian semiconductor companies.
Barefoot Networks, Deephi Tech,
Kneron and ASR. The motivations for M&A are clear
Semiconductor companies today
M&A activity in Japan, South Korea, are motivated to conduct domestic
and Taiwan was relatively quiet and cross-border M&A primarily for
compared to China. The main four reasons: to acquire advanced
motivations for deals there are technology; strengthen market
strengthening market position and position and expand market share;
expanding market share, as well as search for cutting-edge applications;
searching for emerging applications. and expand their industry supply
chains.
42
Acquiring advanced technology: Strengthening market position: Accessing emerging sectors: The
The semiconductor industry in By acquiring competitors, enterprises rapid development of emerging
China relies heavily on imports. In play a synergistic role to increase technologies such as artificial
2018, there have been some urgent market share and enhance intelligence and autonomous driving
moves from Chinese tech players. profitability. In 2018, Shanghai Will has greatly stimulated demand
For instance, Chinese semiconductor Semiconductor acquired a 85% stake for semiconductor chips. Chip
cleaning equipment company NAURA in Beijing SuperPix Micro Technology companies are now expanding
acquired Akrion, a semiconductor and a 96% stake in Beijing OmniVision their business areas to emerging
wafer cleaning equipment company in Technologies for USD2.18bn to sectors through M&A. For instance,
Pennsylvania, to expand its business take advantage of the high-end Samsung Electronics has acquired
into silicon wafer fabrication, MEMS technics of OmniVision and cost Haman, one of the largest suppliers
and packaging, becoming the first control capabilities of SuperPix. M&A of connected, intelligent automotive
case approved by the Committee also paves a path to quick market component globally. In 2018, Japan-
on Foreign Investment in the United entry, particularly overseas. South based developer and manufacturer of
States (CFIUS) since the Trump Korean foundry SK Hynix acquired semiconductors Renesasa Electronics
administration took office. a 50% stake in Chinese foundry acquired US-based company
company Haejin Semiconductor Integrated Device Technology (IDT) to
Despite possessing more advanced (Wuxi) by investing USD75m as enhance its competence in self-driving
semiconductor technology, Japan, a strategy to expand its foundry car technology.
South Korea and Taiwan are still business. China-based company
seeking semiconductor-related Huada Semiconductor has acquired Expanding industry supply chain:
technologies through M&A. Canadian Solantro Semiconductor, The semiconductor industry chain
Taiwan-based company United which designs integrated circuits, to includes design, manufacturing,
Microelectronics acquired a 84% establish and grow its presence in packaging and testing from upstream
stake in Japan-based wafer foundry Ottawa. to downstream. By entering other
company Mie Fujitsu Semiconductor parts of the value chain, incumbents
for USD519m in June 2018 to enable not only create new revenue streams,
UMC to leverage its IC production but also create synergies. In 2018,
experience. Meanwhile Taiwan Ingenic Semiconductor, the leading
Semiconductor Company acquired embedded CPU chip and solutions
US-based ON Semiconductor for provider in China, acquired a stake in
USD6M to expand its semiconductor Beijing Xicheng Semiconductor, which
application portfolio and strengthen will enable Ingenic to expand into the
its core business. high-end memory chip business.
43
Look out for M&A risks Risks Related to Financial Risks Related to Overseas M&A
Despite the many benefits of M&A, Information and Due Diligence and Integration
a host of issues can arise pre and Compared with domestic M&A There are many challenges for cross-
post- merger, including mistakes in projects, overseas M&A projects border mergers and acquisitions
judging the target, failure to conduct have a different quality of financial in the process of transaction and
detailed due diligence and poor information, interpretation, supply integration, including the complexity
implementation. method and verification. Accordingly, of restructuring, lack of local
due diligence procedures should be integration resources and teams,
Risks Related to M&A Target reasonably arranged to deal with management of external stakeholders
Screening and Valuation these differences. and loss of talent, differences in
The Chinese technology industry management information systems,
faces various political and legal risks M&A involving foreign semiconductor and differences in culture, salary and
in overseas M&A. Governments in companies needs to focus on the risks welfare systems. Specifically, these
the US and Europe have very tight in shareholders' backgrounds: include:
restrictions on foreign acquisitions
or investments, especially in 1) Small and medium-sized Differences in culture and subsidy
semiconductors as a high-tech enterprises managed by founders: benefits systems: Different
industry. It will be difficult for Chinese these companies often do not pay performance appraisal systems
companies to acquire high-tech enough attention to daily accounting, will have an impact on employee
companies with the advantage of have relatively confused financial performance. Generally, the higher
high technology and high commercial data, and do not employ well-known the degree of company localization,
value, which can increase the difficulty accounting firms to audit their the lower the basic wage and
of selecting M&A targets. In addition, financials. It is difficult to obtain the higher the sales commission;
although Chinese enterprises can financial data to conduct business whereas the basic wages of foreign
obtain intangible assets (such as analysis through written materials. enterprises are generally higher and
technology and brand) and advance sales commission relatively low; these
their industry level by acquiring 2) Private-equity fund management differences may cause inequalities in
foreign enterprises, the premium enterprises: Private-equity funds wages and subsidies in the integration
paid is generally high, which brings tend to plan business expansion in of two enterprises, affecting the
substantial operational and financial the early stage, and sell at better enthusiasm of employees.
risks to Chinese enterprises. High performance points. Therefore,
financial leverage is the most financial data is detailed and
significant characteristic of overseas comprehensive, corresponding
M&A for Chinese enterprises. And written materials are available and
high leverage inevitably brings high financial data is highly authentic.
risks. If M&A fails or companies fail However, good historical financial
to integrate, leading to losses, M&A performance can often be caused
companies will face great financial by short-term incentives. Although
risks. current profits are high, the
development prospects may be
limited.
44
Semiconductors – the Next Wave | Accessing the Chinese market
Brain drain and management: Low-efficiency in external Differences in management

M&A can result in the loss of core stakeholder management: information systems: Most Chinese
managerial personnel, thus affecting customers and suppliers may think companies adopt a native enterprise
the normal operations of a company's restructuring will bring uncertainty resource planning (ERP) system, such
business. Core personnel include: to a company's business, and lose as UFI, which will hinder information
people who master key technologies/ confidence in cooperation or confuse sharing with companies using
processes, those who master future partnerships; disruption of Oracle/SAP ERP systems to a certain
government and customer resources. supply flow and customer losses may extent, resulting in some delays in
Senior managers with multiple be difficult to reverse; competitors information reception and operation.
roles will have greater resistance to could seize the opportunity to
corporate restructuring. A company compete for passenger flow during
is likely to lay off employees after the restructuring run-in period.
reorganization and its leadership
responsibilities can change.
Complexity and chain reaction in

restructuring: Foreign labor laws
highly favor the employees' side;
if restructuring could cause social
instability, local governments might
interfere with factory relocation/
restructuring plans.
Lack of local project teams with

the function of supervising the
reorganization process: M&A
and reorganization projects are
typically headquarters-led, but a
lack of local team participation can
lead to local problems that cannot
be submitted to a preparatory
committee for resolution; a lack of
local team leadership will reduce the
effectiveness of the implementation
of a reorganization plan. Language
and cultural barriers can hinder
project teams' ability to communicate
effectively with local staff.
45
5
Accessing the
Chinese market
Semiconductor cash cow for MNCs It is no surprise then that China is a
China is expected to maintain its source of income for many of the top
position as the world's largest global semiconductor companies,
semiconductor consumption market. several of which generate over half
In 2018, China's semiconductor their revenue from China. Qualcomm,
consumption accounted for 41% of for example, gets 65% of its revenue
the global total. Commercialization from China.
of artificial intelligence, adoption
of the Internet of Things and 5G
will further drive growth. By 2024,
China will account for 57% of global
semiconductor consumption.
46
No one-size-fits-all solution for single correct method, but in general, weak in the high-end design of
market entry MNC's technology status as well as semiconductors and manufacturing,
Multinationals trying to access the China's domestic technology status and competition is not high, so MNCs
Chinese market should consider a play a crucial role. When an MNC typically set up regional offices or
multitude of factors such as policies, has the upper hand in technology, a wholly foreign-owned enterprise
technologies, marketing, logistics and it has higher bargaining power and (WFOE) to access the market. In areas
global strategies. It is also important will have less incentive to share where China is relatively strong such
for multinationals to realize the intellectual property. However, MNCs as packaging, testing and low-end
position they are in before entering might altogether avoid cases where design, MNCs may opt to setup a JV or
China to come up with the best entry domestic players are already strong. avoid the market altogether.
strategy. Obviously, there is no one For instance, China is relatively
Figure: Semiconductor MNC revenue from China (2017)
90%
83%
80%
70% 65%
62%
60%
54%
50%
39%
40%
30%
24%
20%
20%
10%
0%
Skyworks Qualcomm Qorvo Broadcom Acacia Intel Nvidle
47
Figure: MNC's position/involvement in the Chinese semiconductor industry
Strong
Low competition Moderate competition High competition
Design – High end
• MNCs typically setup Manufacturing

regional offices only
• MNCs typically setup

WOFE Packaging & Testing
MNC’s relative
technical capability
• MNCs typically setup JV
Design – Low end
• MNCs typically avoids
Weak
Weak China’s relative technical capability Strong
Regional office: This model is

Capacity
generally applicable to sectors where Commissioning
Company Headquarter Location (thousand
technology is completely monopolized date
/month)
by MNCs. There is little incentive to
share. For example, Qualcomm has TSMC Taiwan Nanjing 20 2018
established regional offices in Beijing,
Shanghai, Shenzhen and Xi'an.
Powerchip Taiwan Hefei 10 2018
WFOE: Leading semiconductor MNCs

are keen to set up wholly owned AOS US Chongqing 20 2018
plants to meet the enormous demand
for semiconductors in China. Overseas Intel US Xi’an 100 -
foundries from Taiwan, South Korea,
and the United States have set up new
plants to increase production capacity SK hynix South Korea Wuxi 168 2018
in China. For example, TSMC will build
a 12-inch wafer plant in Nanjing, and Sumsang South Korea Xi’an - 2014
its production capacity will reach 20
thousand/month. SK Hynix also plans Source: SEMI, company websites, Deloitte analysis
to build a production plant in Xi'an,
with a capacity of 168 thousand/
month.
48
Joint ventures: China, Taiwan and Figure: Top 10 packaging and testing companies worldwide (2017)
the United States hold the leading
positions worldwide in semiconductor 12.00%
packaging and testing. From 2010
10.10%
to 2016, the output value of China's 10.00%
local packaging and testing industry
7.90%
increased from RMB62.9 billion to 8.00%
RMB156.4 billion at a compound 6.20%
annual growth rate of 20%, higher 6.00%
5.20%
than the global average. As of 2017,
3.70%
there are three domestic companies 4.00%
in the top 10 globally, namely CJCET,
2.00% 1.80%
Huatian and Tongfu. 2.00% 1.30% 1.30% 1.20%
Given local semiconductor companies 0.00%

ASE Amkor JCET SPIL PTI Huatian Tongfu KYEC UTAC ChipMOS
have become among the best in Technology
the packaging and testing sectors,
MNCs can further enhance their Note: Taiwan based SPIL was acquired by ASE in 2018.
technology via cooperation with local Source: Founder Securities, Deloitte
companies. In 2016, AMD and Chinese

advanced technology company
Nanjong Fujitsu Microelectronics Co. minimal incremental investment is
established a joint venture enterprise, required. On the other hand, if the
which absorbed the R&D team and market is already competitive and the
advanced equipment assets of AMD strategic value of the market is still
in Suzhou and Penang, Malaysia to high for China, aggressive localization
become the world's top packaging and opportunities might be needed to
testing company. maximize value. In the worst case,
where an MNC's capability and
In essence, an MNC has to consider strategic value are both low, the MNC
its competitiveness and the strategic should exit the market.
value of the Chinese market. Unlike
domestic players, MNCs' China For example, some MNCs gave
initiatives are heavily influenced up medium to low-end cell phone
by their global strategies, which business to focus on advanced, high-
determine their next steps. The end cell phones. Meanwhile, other
majority of MNCs' participation comes MNCs cooperate with local IT giants
in areas where their competitiveness to localize hardware and software
and the strategic value of China technologies, in order to avoid
are both high. But, there are other regulatory restraints. Meanwhile,
options. For instance, if an MNC many MNCs have exited China's
possesses a strong ability to compete mobile phone market and withdrawn
even though the strategic value of their investments in JVs.
China is relatively low, it can take an
"opportunistic harvest" approach to
select favorable businesses in which
49
Figure: MNC's China initiatives
High Note: Strategic Value:

• Direct financial value contribution amongst its
Opportunistic Harvest
global portfolio;
Select favorable MNC Initiative • Indirect value (e.g. cost saving by local
Capability Competitiveness
business which need production, PR building with government for

minimum incremental other product sales)
investment
Source: Deloitte Analysis
Aggressive Localization
Aggressively
Withdraw
seeking localization
opportunities to
maximize value
Low
Low Strategic Value Of China market High
Navigating risks and challenges differences in culture, laws and

Multinational companies from other regulations. By differentiating the
countries or regions will encounter two dimensions of impact and
a series of challenges when entering controllability, we have identified four
Mainland China, due to various major issues MNCs can encounter.
Figure: Challenges foreign MNCs can encounter:
large Notes: Impact denotes the degree to which

Localization a certain risk will influence a deal;
Cultural Controllability measures how well
Differences companies can control specific risks.
Law and Source: Deloitte

regulations
Impact Security review
Protectionist trend
Small
Low Controllability High
50
Cultural differences can hinder Compared to localization, the trend

multinational companies from towards protectionism means security
obtaining the expected returns. reviews are less controllable and
When a foreign company completely have a greater impact. Since the
replicates its business model in China, ZTE event, the Chinese government
its corporate culture and working and enterprises have fully realized
methods will also be brought in, the importance to national security
which will create cultural conflicts and economic development of
and thus become a risk factor for developing a complete semiconductor
the company's business activities. industry with advanced technology,
Completely copying rather than and have begun to invest heavily
understanding the business model in the semiconductor industry to
of Chinese companies will also achieve semiconductor localization.
bring certain risks. Moreover, with Once Chinese companies make
the expansion of their businesses, breakthroughs, local products should
overseas companies also need to quickly occupy almost the entire
consider recruiting talent that is market, which would spell bad news
familiar with foreign corporate culture for MNCs.
and language, and at the same
time proficient in the business. It is
therefore a challenge to build a talent
system that is in line with the business
development and corporate culture
in Mainland China. At the same time,
given laws and regulations in Mainland
China are different from those in
other countries, so foreign companies
entering the Chinese semiconductor
market face compliance challenges.
51
6
Digitalization is key
to competitiveness
Semiconductor companies tody Digital transformation has become the
must be more rapid and agile main strategy for many enterprises
than ever to remain competitive. to meet the challenges they face. For
The commercialization of new example, the digital transformation
technologies such as AI and big data is of the retail industry has penetrated
promoting the digital transformation all areas of the value chainInvolving
of enterprises and the realization consumer-oriented demand
of intelligent production, intelligent forecasting, personalized marketing,
management and intelligent sales. purchasing experience and intelligent
By investing in digital infrastructure customer service. The main aim is to
to increase productivity and develop continuously improve efficiency and
new business channels, companies effectively attract consumers.
will have the opportunity to The foundations of digitalization are
overcome development barriers and sound, as chip processing capacity,
digitalization will provide new impetus cloud service popularization, sensor
for their development. and other hardware prices have
52
Semiconductors – the Next Wave | Digitalization is key to competitiveness
declined and computing power has Many semiconductor companies are

improved a great deal. From the already utilizing digital tools to gain
application perspective, technology competitive advantage throughout
companies have provided the market the product life cycle. For example,
with various data analysis tools to the application of AI and analytical
improve data utilization rate, improve tool applications in the semiconductor
operational efficiency and reduce industry extends from design,
production costs. manufacturing, packaging and testing
all the way to management.
Figure: Digital tools utilized by semiconductor companies
Current situations Improvement
Design • Companies have to correct errors in physical • Correct design errors using virtual tools
designs (R&D) • Obtain data from device and tool to find out the
• Hard to find errors based on past factor creating mistakes and instruct next design
experience (Upgrade)
Manufacturing • Data generated from each devices cannot be • Connect data from all equipment and create data
connected set
• Manual maintenance are required and • Analyzing data set to predict the equipment failure
cannot predict the machine failure in advance
Packaging and • Consume a lot of time during testing • Build plat platform for data sharing between testing
testing companies and component suppliers
• Reduce the time required to achieve acceptable
shipping quality
management • Pricing based on subjective opinions • Data-based analysis could provide more accurate
• Difficult to obtain data from other players information for pricing
along the supply chain • Create system to track all information related with
• Hard to spot potential customer and provide chip and make it easier to find mistake
service for them • Spot potential customers that could contribute
more on revenue
Source: Public information, Deloitte
53
Artificial intelligence paves way Some leading Japanese semiconductor firms have
for efficiency
AI technology will play an already used AI extensively in the manufacturing
indispensable role in all aspects process. For example, in the manufacturing and
of semiconductor production and transportation process, a large amount of images and
company operations. Massive
amount of data is generated in vibration data are collected and analyzed, and the effect
the semiconductor manufacturing of improving productivity and yield is already shown. The
process and traditional data analysis
adoption for AI at the moment is mostly concentrated in
methods can only utilize parts
of the structured data for post- manufacturing, but it will be relevant in various fields in
mortem analysis. But intelligent the future such as quality control and demand forecast.
analytics based on AI can perform
comprehensive real-time analysis of
data sets to improve production and
management efficiency.
Design: Artificial intelligence can alter Figure: The benefits AI brings to semiconductor manufacturing
the process of design. Each step in
the semiconductor design produces
a large number of parameters. Unlike
Huge amount of Real-time
traditional analysis tools, new analysis data including analyze
technology can help semiconductor unstructured
designers comprehensively analyze data
all the data they obtain, and learn
from experience, as well as analyze
past data and extract relationships
from the data and results. Whether it Increase efficiency
is high frequency or low- to medium-
frequency data, data combination
can be used to spot potential errors
Make Precise
and increase yield, providing insights prediction control
into newly generated data to make
decisions or correct errors by simply
changing a certain parameter. In
addition, using data as a basis
for decision-making can avoid Source: Deloitte analysis
communication barriers between
design teams and process teams. .
54
Manufacturing: In the Management: An AI customer information on changes in enterprises'

manufacturing process, data service system starts with semantic daily income and the corresponding
generated by each process are understanding and problem sales data. Combined with customer
shared, as well as directly analyze identification derived from customer information and external auxiliary
and report errors, thereby reducing questions, searches for big data in data, analytical tools can provide
manual inspections that may contain the identified problems, analyzes the strategies and suggestions for the
errors, and improving efficiency. An meaning of the problem, produces a pricing of semiconductor products
AI system can check data thousands knowledge map and performs answer and form a smart pricing plan. A
of times in a minute, providing nearly matching and decision-making. An large amount of data is generated in
600 times the efficiency of staff. AI AI customer service system realizes interactions with customers. Through
detection and maintenance system 24-hour customer service online, the continuous accumulation of
connections cover the entire process answers questions at any time, customer demand data, a company
of production data, and can build real- improves customer satisfaction and can analyze and understand
time predictions about equipment saves the labor costs of semiconductor preferences and needs in depth, as
failures, reducing losses caused by enterprises, as well as frees staff from well as support precision marketing
production interruptions. boring and high-pressure work to do and large-scale personalized
more valuable work. recommendations and services.
Packaging and testing: The full
use of data shortens testing time Analytical tools to know Master data management (MDM)
and accelerates time to market. customers better is important
Data integration and connectivity Analytical tools are mainly being used Master data management involves
can greatly improve data utilization in management, including decision- data collection, classification,
efficiency. On one hand, building a making and customer development. management, cleaning and other
data exchange platform between Data-driven decision support has areas within different organizations.
semiconductor testing companies undoubtedly become an important It is the basis of AI data analysis and
and component suppliers can enable strategy to help semiconductor various data analysis tools.
the causes of errors to be found in time, companies gain insights into potential
and on the other hand, it can reduce the customers, thereby making more
number of malfunctioning chips. These rational decisions and reaching out
platforms have reduced the number of to more customers. The application
malfunctioning chips by 50%. of data analysis tools includes
55
Figure: Master data management
Data Preparation: Information Lifecycle Management:

Automated aggregation, classification, and Consistent management of information from
filtering of data at the edge limits the “noise” creation to final disposition (strategy, process,
and technology)
• Edge Computing • Self-Service Enterprise
Data Preparation Monitor and
Access and
Adapt to new data
Classify Data
requirements
Manage Data Dispose Data
ERP
Advanced Master Data Management:
Integrate various systems and maintain the
CRM
Video Sensors consistency and visibility of Master Data across
the organization
The digitization of semiconductor In the semiconductor design and

companies is based on the use manufacturing process, MDM can
of data, as well as more refined alsoprovide intelligent device data-to-
management of enterprises' data analysis. In the packaging and
production and operations. Can the testing process, it can link the data of
data generated by various sensors packaging and testing enterprises and
and smart devices be stored in the component suppliers. In management
data set in time? How is the data processes, it can deal with business
classified? How can structured and operations and financial data.
unstructured data be managed and
cleaned? How is the data from each
area shared? MDM can solve all of
these issues.
56
Semiconductors – the Next Wave | Contacts
Contacts
William Chou Leo Chen
Deloitte China Technology, Media Deloitte China Semiconductor
and Telecommunications Industry Sector Leader
Leader Tel: +86 21 61411911
Technology Sector Leader Email: leoschen@deloitte.com.cn
Tel: +86 10 85207102
Email: wilchou@deloitte.com.cn Andrew Chen
Deloitte China Technology Sector
Jennifer Shao Consulting Leader
Deloitte China Semiconductor Tel: + 852 22387678
Sector Partner Email: andrewclchen@deloitte.com.hk
Tel: + 86 21 23166586
Email: jennifershao@deloitte.com.cn Lisa Zhou
Deloitte China Technology, Media
Roger Chung and Telecommunications Industry
Deloitte China Technology, Media Manager
and Telecommunications Industry Tel: + 86 10 85125909
Research Senior Manager Email: liyzhou@deloitte.com.cn
Tel: + 86 21 23166657
Email: rochung@deloitte.com.cn
Acknowledgements
The authors would like to thank Deloitte Taiwan TMT Industry Leader Gordon
Chen, TMT Industry Manager Betty Fang, Deloitte Korea Technology Sector
Leader In Kyoo Baek, Financial Services Industry Leader Hyung In Lee, and
Deloitte Japan Technology Sector Leader Yuki Shuto for their contributions to
this report.
57
Office locations
Beijing Harbin Shenyang
12/F China Life Financial Center Room 1618, Development Zone Mansion Unit 3605-3606, Forum 66 Office Tower 1
No. 23 Zhenzhi Road 368 Changjiang Road No. 1-1 Qingnian Avenue
Chaoyang District Nangang District Shenhe District
Beijing 100026, PRC Harbin 150090, PRC Shenyang 110063, PRC
Tel: +86 10 8520 7788 Tel: +86 451 8586 0060 Tel: +86 24 6785 4068
Fax: +86 10 6508 8781 Fax: +86 451 8586 0056 Fax: +86 24 6785 4067
Changsha Hefei Shenzhen

20/F Tower 3, HC International Plaza Room 1201 Tower A 13/F China Resources Building
No. 109 Furong Road North Hua Bang ICC Building 5001 Shennan Road East
Kaifu District No.190 Qian Shan Road Shenzhen 518010, PRC
Changsha 410008, PRC Government and Cultural Tel: +86 755 8246 3255
Tel: +86 731 8522 8790 New Development District Fax: +86 755 8246 3186
Fax: +86 731 8522 8230 Hefei 230601, PRC
Tel: +86 551 6585 5927 Suzhou
Chengdu Fax: +86 551 6585 5687 23/F Building 1
17/F China Overseas Global Wealth Square
International Center Block F Hong Kong 88 Su Hui Road, Industrial Park
No.365 Jiaozi Avenue 35/F One Pacific Place Suzhou 215021, PRC
Chengdu 610041, PRC 88 Queensway Tel: +86 512 6289 1238
Tel: +86 28 6789 8188 Hong Kong Fax: +86 512 6762 3338 / 3318
Fax: +86 28 6317 3500 Tel: +852 2852 1600
Fax: +852 2541 1911 Tianjin
Chongqing 45/F Metropolitan Tower
43/F World Financial Center Jinan 183 Nanjing Road
188 Minzu Road Units 2802-2804, 28/F Heping District
Yuzhong District China Overseas Plaza Office Tianjin 300051, PRC
Chongqing 400010, PRC No. 6636, 2nd Ring South Road Tel: +86 22 2320 6688
Tel: +86 23 8823 1888 Shizhong District Fax: +86 22 8312 6099
Fax: +86 23 8857 0978 Jinan 250000, PRC
Tel: +86 531 8973 5800 Wuhan
Dalian Fax: +86 531 8973 5811 Unit 1, 49/F
15/F Senmao Building New World International Trade Tower
147 Zhongshan Road Macau 568 Jianshe Avenue
Dalian 116011, PRC 19/F The Macau Square Apartment H-N Wuhan 430000, PRC
Tel: +86 411 8371 2888 43-53A Av. do Infante D. Henrique Tel: +86 27 8526 6618
Fax: +86 411 8360 3297 Macau Fax: +86 27 8526 7032
Tel: +853 2871 2998
Guangzhou Fax: +853 2871 3033 Xiamen
26/F Yuexiu Financial Tower Unit E, 26/F International Plaza
28 Pearl River East Road Mongolia 8 Lujiang Road, Siming District
Guangzhou 510623, PRC 15/F, ICC Tower, Jamiyan-Gun Street Xiamen 361001, PRC
Tel: +86 20 8396 9228 1st Khoroo, Sukhbaatar District, 14240- Tel: +86 592 2107 298
Fax: +86 20 3888 0121 0025 Ulaanbaatar, Mongolia Fax: +86 592 2107 259
Tel: +976 7010 0450
Hangzhou Fax: +976 7013 0450 Xi’an
Room 1206-1210 Room 5104A, 51F Block A
East Building, Central Plaza Nanjing Greenland Center
No.9 Feiyunjiang Road 6/F Asia Pacific Tower 9 Jinye Road, High-tech Zone
Shangcheng District 2 Hanzhong Road Xi'an 710065, PRC
Hangzhou 310008, PRC Xinjiekou Square Tel: +86 29 8114 0201
Tel: +86 571 8972 7688 Nanjing 210005, PRC Fax: +86 29 8114 0205
Fax: +86 571 8779 7915 / 8779 7916 Tel: +86 25 5790 8880
Fax: +86 25 8691 8776
Shanghai
30/F Bund Center
222 Yan An Road East
Shanghai 200002, PRC
Tel: +86 21 6141 8888
Fax: +86 21 6335 0003
58
About Deloitte
Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited
(“DTTL”), its global network of member firms, and their related entities.
DTTL (also referred to as “Deloitte Global”) and each of its member firms
and their affiliated entities are legally separate and independent entities.
DTTL does not provide services to clients. Please see www.deloitte.com/
about to learn more.
Deloitte Asia Pacific Limited is a company limited by guarantee and a

member firm of DTTL. Members of Deloitte Asia Pacific Limited and their
related entities provide services in Australia, Brunei Darussalam, Cambodia,
East Timor, Federated States of Micronesia, Guam, Indonesia, Japan, Laos,
Malaysia, Mongolia, Myanmar, New Zealand, Palau, Papua New Guinea,
Singapore, Thailand, The Marshall Islands, The Northern Mariana Islands,
The People’s Republic of China (incl. Hong Kong SAR and Macau SAR), The
Philippines and Vietnam, in each of which operations are conducted by
separate and independent legal entities.
The Deloitte brand entered the China market in 1917 with the opening
of an office in Shanghai. Today, Deloitte China delivers a comprehensive
range of audit & assurance, consulting, financial advisory, risk advisory
and tax services to local, multinational and growth enterprise clients in
China. Deloitte China has also made—and continues to make—substantial
contributions to the development of China's accounting standards, taxation
system and professional expertise. To learn more about how Deloitte makes
an Impact that Matters in China, please connect with our social media
platforms at www2.deloitte.com/cn/en/social-media.
This communication contains general information only, and none of Deloitte

Touche Tohmatsu Limited, its member firms, or their related entities
(collectively the “Deloitte Network”) is by means of this communication,
rendering professional advice or services. Before making any decision or
taking any action that may affect your finances or your business, you should
consult a qualified professional adviser. No entity in the Deloitte Network
shall be responsible for any loss whatsoever sustained by any person who
relies on this communication.
©2019. For information, contact Deloitte China.

BJ-011SC-19
59

TW Semiconductor Report en

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

TW Semiconductor Report en

Uploaded by

Copyright:

Available Formats

Semiconductors – the Next Wave

Opportunities and winning strategies for

The ever-evolving semiconductor scene 5

Future of automotive semiconductors 12

Race is on for AI chips 30

Muted tone for M&A 38

Accessing the Chinese market 46

Digitalization is key to competitiveness 52

East Asia (Mainland China, Japan, South Korea and Taiwan) is

M&A activity in the semiconductor industry has already peaked

The semiconductor sector's growth trajectory will ﬂatten

In the automotive sector, the adoption of safety-related

The AI semiconductor scene has seen a race not just at the

Finally, China has become a source of income for top global

Figure: Global semiconductor sales revenue (2016-2022, billion USD)

542.64 billion (2022)

Source: Gartner, Deloitte analysis

Automotive and industrial the semiconductor industry, with revenue

Figure: Semiconductor revenue growth by electronic equipment type (2017- 2022)

Head Mounted Display：34.9%

ADAS：23.6% Automotive Electronics

Source: Gartner, Deloitte analysis

North America 19.60%

Figure: Major players in the Chinese semiconductor industry

National Team Local Team

Computers & electronics $167.3

Electrical equipment $39.9

150 Machinery $25.7

0 Plastics & rubber $15.6

Source: Deloitte analysis

Figure: Factory-installed electronics

Windshield Digital Turn

Adaptive Front EV/HEV

100% 100% 100%

100% 100% 100%

Electronic systems as % of total car cost

Advanced driver assist Active-passive safety

2000 22% Airbag ABS/ESP

1980 10% Electronic fuel injection

0% 10% 20% 30% 40% 50% 60%

Source: Deloitte analysis

The cost of semiconductor content, from a surge in demand for various

Figure: Cost contribution of automotive electronics and semiconductor

700 Semiconductor price 50%

Source: IHS, Deloitte Analysis

Semiconductor vendors play a entire ecosystem for years to come.

Figure: Role of semiconductors in the automotive ecosystem

~10% <10% ~40%-70% <10% ~20% Cost for

R&D & Material Parts Parts Marketing

Material Component Distributor &

IP Vendors Bosch, Delphi,

$400M $30B $120B $2T

Source: Cadence, IHS, Deloitte analysis

Key words: Automation, electrification, connectivity

Figure: Mega automotive semiconductor trends

Trends Players Technology Transfomation

Source: Deloitte analysis

Figure: Number of sensors for different levels of autonomous driving

AC C: Active Cruise Control Level 5 (2040)

Source: Deloitte analysis

2. Electrification 3. Digital Connectivity especially have an advantage in this

At the same time, EV/HEV advances

from 2019). Many global automakers