半导体行业的发展趋势 2010 普华永道

中国半导体行业历史及现状

Technology, Media & Telecommunications

A change of pace for the semiconductor industry?

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

Edited by PricewaterhouseCoopers

By Werner Ballhaus, Dr Alessandro Pagella and Constantin Vogel

All rights reserved. Reproductions, microfilming, storage and processing in electronic media are not

permitted without the publisher’s approval.

Typesetting

Nina Irmer, Digitale Gestaltung & Medienproduktion, Frankfurt am Main

Published by

Kohlhammer und Wallishauser GmbH, Druckerei und Verlag, Hechingen

Printed in Germany

© November 2009

PricewaterhouseCoopers refers to the German firm PricewaterhouseCoopers AG Wirtschafts-

pruefungsgesellschaft and the other member firms of PricewaterhouseCoopers International

Limited, each of which is a separate and independent legal entity.

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

Preface

Innovation drives the semiconductor industry, and in turn its chips form the

heart of modern industrial societies. Chips are at the essential core of mobile

telephones, computers, flat-screen monitors and television sets, a wide range of

medical procedures including CT scans, ultrasound and X-rays, and they play

an enormous role in today’s sophisticated cars and aircraft. The number of

installed semiconductor components expands constantly, so it would be easy to

assume that the semiconductor industry is experiencing a golden era. However,

despite the increasing demand for chips, hardly any other sector has been hit

so hard by the current economic crisis.

How will the market for chips develop in the next few years? What business

models will prove to be robust during the crisis and beyond? Where are the

current opportunities? What are the critical factors of success? This study

considers those and other essential questions.

Apart from an analysis of the production process and the subsequent value

chain, this study comprises a benchmark analysis of the main segments of the

semiconductor industry. Experts who offer their view of technological and

economic developments in interviews include Professor Doris Schmitt-

Landsiedel, Technische Universität München; Professor Jürgen Becker and

Professor Michael Siegel, Karlsruher Institut für Technologie; Peter Wennink,

ASML; Andrea Cuomo, STMicroelectronics; and Herbert Halamek, Continental.

In addition to the interviews, we had fruitful discussions with Carla Sinanian,

NXP Semiconductors; Dr Peter Hardt, Infineon Technologies; and Dr Helmut

Lagger and Ulrich Schoen, Nokia Siemens Networks. We appreciate all these

contributions and thank the participants for their cooperation and wisdom. We

particularly thank Peter Bauer, Infineon Technologies, who has enhanced our

study with a guest article.

The aim of this study is to identify issues which will be the driving force behind

the market in the next few years and to use this analysis to recommend action

for semiconductor companies (some of which have already acted and are now

well poised to meet the challenges of the future). Which products, which

components and which regions have to be considered is the subject of our

sales forecast. Our study focuses particularly on the European perspective.

Are there indications of a change of pace for the semiconductor industry? Join

our journey into the complex world of the semiconductor industry.

If you would like further information, or to discuss any of the findings in our

report and how they might impact your business, please do not hesitate to

contact either of us (raman.chitkara@ or

werner.ballhaus@) or any member of our technology team listed in

the back of this document.

Düsseldorf, November 2009

Raman Chitkara Werner Ballhaus

Global Technology Leader German Technology, Media and

Telecommunications Leader

Preface

3

中国半导体行业历史及现状

Contents

4 A change of pace for the semiconductor industry? Contents Preface.................................................................................................................3 Figures.................................................................................................................6 Tables...................................................................................................................7 Abbreviations........................................................................................................8 A Summary......................................................................................................10 1 The main results – an overview...................................................................10 2 Structure of the study...................................................................................11 Guest article by Peter Bauer, CEO, Infineon Technologies AG........................12 B The semiconductor industry.........................................................................14 1 Market structure...........................................................................................14 2 Semiconductors in the crisis........................................................................19 Related discussion: the semiconductor value chain..........................................23 3 Analysis of the competition..........................................................................27 4 Summary of results and outlook..................................................................31 C Technology and trends.................................................................................33 1 Production capacity......................................................................................33 Related discussion: complementary metal-oxide semiconductors (CMOS)......34 2 Feature sizes................................................................................................34 3 Enhanced functionality.................................................................................38 4 Wafer sizes..................................................................................................40 D Sales forecasts.............................................................................................42 1 The semiconductor market according to applications..................................42 2 The semiconductor market according to installed components...................46 3 The semiconductor market: a regional comparison.....................................49 E Conclusion and outlook................................................................................52

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

F Methodology.................................................................................................54

1 Sales forecast..............................................................................................54

2 Interviews with experts................................................................................54

3 Analysis of the competition..........................................................................54

4 Peers in the analysis of competition............................................................55

Worldwide contacts for technology....................................................................56

About PricewaterhouseCoopers........................................................................59

Authors...............................................................................................................60 Contents

5

中国半导体行业历史及现状

Figures

6 A change of pace for the semiconductor industry? Figures Fig. 1 The semiconductor industry and its environment............................15 Fig. 2 Product families of integrated circuits..............................................16 Fig. 3 Comparison of standard and customer-specific products, as well as typical factors of success.....................................................17 Fig. 4 Semiconductor sales........................................................................19 Fig. 5 Value chain according to Porter.......................................................23 Fig. 6 Front-end production processes......................................................25 Fig. 7 Back-end production processes.......................................................26 Fig. 8 Breakdown of sales according to semiconductor business models..............................................................................................28 Fig. 9 EBITDA margin broken down according to business models..........28 Fig. 10 EBIT margin broken down according to business models...............29 Fig. 11 Research and development spending broken down according to business models...........................................................................29 Fig. 12 Selling and administrative expenses broken down according to business models...........................................................................30 Fig. 13 Consolidation of the results of the analysis of competition..............31 Fig. 14 Example of technology cycle for the changeover to new technologies in semiconductor production.......................................36 Fig. 15 Breakdown of the installed worldwide production capacity by wafer size.........................................................................................41

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

Tables

Tab. 1 Comparison of profit margins in 2007 and 2008 for different

business models..............................................................................31

Tab. 2 Distribution of the installed worldwide CMOS production

capacity according to smallest possible feature size.......................35

Tab. 3 Development of feature sizes attainable by lithography broken

down according to product...............................................................36

Tab. 4 Semiconductor sales, according to applications for 2006–2012

(in billion USD).................................................................................43

Tab. 5 Percentage of applications in relation to the total

semiconductor market for 2006–2012.............................................43

Tab. 6 Semiconductor sales according to components (in billion US

dollars).............................................................................................47

Tab. 7 Percentage of components in relation to the total

semiconductor market for 2006–2012.............................................47

Tab. 8 Semiconductor sales broken down according to regions for

2006–2012.......................................................................................49

Tab. 9 Sales growth broken down according to regions for 2006–

2012.................................................................................................49

Tables

7

中国半导体行业历史及现状

Abbreviations A change of pace for the semiconductor industry?

Abbreviations

ABS

AG Anti-lock braking system Aktiengesellschaft (joint stock company)

ASIC Application-specific integrated circuit

8 ASSP Application-specific standard product CAGR CFO CMOS CPU DDI DRAM DSP Digital EBIT EBITDA EPROM EEPROM Electrically ESP Electronic Fab A FERAM IC Integrated IC IDM IDM IP Intellectual ITRS KIT LED Light-emitting LTE MCU Microcontroller

Compound annual growth rate Chief financial officer Complementary metal-oxide semiconductor Central processing unit Display driver integrated circuit Dynamic random access memory signal processor Earnings before interest and taxes Earnings before interest, taxes, depreciation and amortisation Erasable programmable read-only memory erasable programmable read-only memory Stability Programme production facility for semiconductors Ferroelectric random access memory circuit Integrated device manufacturer, mainly manufacturing logic chips Integrated device manufacturer property International Technology Roadmap for Semiconductors Karlsruher Institut für Technologie diode Long Term Evolution unit

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

Memory IDM Integrated device manufacturer, manufacturing memory

chips

mm Millimetre

MPU Microprocessor unit

MRAM Magnetoresistive random access memory

nm Nanometre

OEM Original equipment manufacturer

OLED Organic light-emitting diode

PC Personal computer

R&D Research and development

SG&A Selling, general and administrative

SIA Semiconductor Industry Association

SRAM Static random access memory

SoC System-on-a-chip

SiP System-in-package

Abbreviations

9

中国半导体行业历史及现状

Summary

10 A change of pace for the semiconductor industry? A Summary 1 The main results – an overview The semiconductor industry reported constant growth from the time it came into being around 40 years ago until the current economic crisis began. The industry was able to cope well with short-term corrections, such as the bursting of the Internet bubble in 2001, but the current crisis brought major problems. For 2009, we expect to see a decline in revenue of about 20% from the previous year. But the industry will probably soon overcome these problems: for 2010, we anticipate considerable growth and expect that worldwide sales in 2011will roughly match the level of 2008, and grow further in 2012. Annual average growth is expected to be more than 10% between 2009 and 2012, in line with figures seen in the past. In this growth, only minor shifts in sales distribution are expected. Applications of data processing will continue to account for the highest percentage of overall sales, followed by consumer electronics and mobile communications. Nor are major changes anticipated in the product mix: logic chips will remain the main segment. They will be followed by microprocessors and microcontrollers, with memory chips in third place. The pace of innovation in the semiconductor industry is high. Moore’s law postulates that the number of transistors that can be placed on a single computer chip doubles every 18 to 24 months, and over four decades the law has held up. And the process of reducing feature size will continue for years to come. However, the limits of the existing technology (CMOS, for complementary metal-oxide semiconductor) are already evident: in order to produce extremely small features in a cost-effective manner while avoiding functional restrictions, new materials and technologies must be brought into play. And the functionalities are becoming more and more important: previously independent modules are increasingly being integrated in individual chips. Moreover, larger wafer diameters promise to permit more efficient chip production in semiconductor manufacturing. The changeover to diameters of 450 mm awaits high levels of investment, and in view of the financial crisis, is probably several years off. We do not expect it before 2012. The need for capital and the high pace of innovation mean that companies are increasingly specialising in individual elements in the value chain – such as fabless companies (those without production facilities) that design chips and foundries that specialise in semiconductor production. Integrated device manufacturers (IDM), which cover the entire value chain, are increasingly adopting a ‘fab-light’ strategy, in which some areas of production are outsourced to foundries. Integrated device manufacturers generate the highest average sales in the sector; fabless companies, because they specialise in product development, report the highest spending on research and development as a percentage of overall sales. The lowest percentage spending on research and development is reported at the foundries, which do not have any chip development operations. Selling and administrative expenses follow a similar pattern: fabless companies report the highest percentage of costs, foundries the lowest. The economic crisis has meant that most semiconductor companies were not able to generate positive results in 2008. This trend will also continue in 2009, despite cost savings and worldwide capacity reductions. We expect to see positive results in 2010 as a result of the growth in sales.

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

2 Structure of the study

This study consists of four parts. The first part (chapter B) starts with an article

by Peter Bauer, CEO of Infineon Technologies AG, on prospects in the semi-

conductor industry. This is followed by a market overview, which analyses the

value chain, the business models and the competitive environment. In the

analysis of the competition, we use selected parameters to detail special

features and mechanisms of the individual business models.

Chapter C deals with technology and technology trends. The technology drivers

production, feature size, functionality and wafer diameter are used to examine

the current state of technology and discuss expected developments.

Chapter D shows our market forecasts until 2012, broken down according to

applications, components and regions.

Chapter E, the fourth and final part, offers our conclusion and provides an

overview beyond the period covered by the study. Summary

11

中国半导体行业历史及现状

Summary

Peter Bauer

CEO

Infineon Technologies AG

12 A change of pace for the semiconductor industry? Guest article by Peter Bauer, CEO, Infineon Technologies AG Growing with energy efficiency, communications, and security The year 2009 will enter the history of the semiconductor industry as a ‘black year’. Following many years of growth, the worldwide economic and financial crisis resulted in a huge downturn in sales in the industry. Although it will take some time for the market to return to the level of 2008, the semiconductor industry is and will continue to be a growth industry. In future, it will continue to be the case that many innovations will only be possible with the aid of semiconductor products. This is applicable particularly to three key challenges of our modern society: energy efficiency, communications and security. Energy efficiency – green technologies on the advance The increasing global demand for energy, the limited availability of natural resources, rising energy prices and the threat of climate change require solutions for enabling energy to be handled more efficiently. In order to meet the requirements of climate policy, for instance for reducing CO2 emissions, it is necessary to increase efficiency throughout the entire chain of utilisable energy – that is, for the production, transmission and consumption of energy. Innovations from the semiconductor industry are playing a key role with regard to implementing these objectives. The requirement for more energy efficiency will have a positive impact particularly on demand for power semiconductors in the course of the next few years. This is applicable specifically to renewable energies, as well as for example to motor drives in industrial applications and in household products. With regard to power semiconductors for renewable energies, market researchers are assuming average annual growth rates of 18% in the course of the next years.1 Solar and wind power will continue to be two of the main growth drivers. Power semiconductors are the core of rectifiers in photovoltaic and wind power installations, and are a key component for efficiently supplying power to the network. In the automotive industry, the requirement for low-consumption and low-emission cars will also result in stronger demand for semiconductors in the long term, particularly for the drive train. With expected annual growth of around 9%, this is the area of application with the second highest growth rate for automotive semiconductors behind the safety segment.2 This is due to the higher demand for powerful semiconductors for improved regulation of engines, and also the increased production of hybrid drives, which have a much higher semiconductor content than conventional drives. Additional opportunities are developing in the field of electric vehicles. In order to meet the requirements with regard to limiting pollutant emissions, carmakers around the globe are preparing for the mass production of hybrid and electric cars. Overall, despite the sales crisis affecting the car makers, the market for automotive semiconductors is expected to achieve annual average growth of 8% until the year 2013.3 As the long-standing market leader for power semi-conductors and the worldwide number two in the overall market for automotive semi-conductors, Infineon will benefit from these trends. Communications – transmission of data, everywhere and at all times More and more people are wanting to have full access everywhere to the telephone, e-mails and the Internet at all times, with high transmission speeds. The mobile phone industry is working on setting up data networks with even higher speeds. Although mobile phone sales this year will decline for the first time since 2001, the manufacturers are developing new devices in order to be in a position to meet the expected long-term growth. Ultra-low-cost mobile phones and smartphones are two segments which are reporting considerable growth rates even this year despite the recession. With anticipated annual growth of approximately 20%, these fields will continue to be the segments which will report the strongest growth in the worldwide mobile phone market in the course of the next few years.4 1 IMS Research, August 2009, CAGR 2008-13 2 Strategy Analytics, July 2009. CAGR 2008-13 3 Strategy Analytics, July 2009, CAGR 2008-13 4 iSuppli, July 2009, CAGR 2008-13

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

The ultra-low-cost segment is being driven by the strong demand of first-time buyers in

countries such as China and India. India alone is reporting around 10 million new users

every month. This development is benefiting semiconductor manufacturers which are

strong in the field of single-chip integration. This lowers the costs of manufacturing mobile

phones by more than 30%. The market researchers are assuming that the revenues

generated by single-chip products will achieve growth of more than 70% per annum until

2013.5 Wireless high-speed data transfer is a key factor of success in the market for

smartphones. Semiconductor manufacturers which provide a complete platform with a

high-speed reliable modem are able to benefit in this respect.

Security – optimum protection of data, information and objects

The increasing extent to which our globalised world is networked requires solutions which

take account of the more stringent security requirements – in communications, financial

transactions, the identification of persons and objects as well as the protection of data and

networks. This will have a positive impact on the demand for security controllers, security

memory products and other system solutions. The identification of persons is one of the

most rapidly expanding segments. The transition to electronic identity documents through-

out the world – including passports, ID cards and driving licences – is the key growth

driver in this respect. In the course of the next few years, the strongest growth rates are

expected for electronic ID cards. This is a mandatory document in many countries, and the

transition to an ID card in a chip card format will take place gradually. In Germany, the

implementation process will commence in the year 2010 and all citizens of Germany are

expected to have a chip-based ID card by no later than the year 2020. Secure payment

using contactless cards is a further growth area for contactless security controllers. So-

called micropayment is becoming rapidly established particularly in the major cities of the

emerging economies of Asia.

Into the future with optimism

The semiconductor industry will benefit considerably from the growth potential resulting in

the long term from the need for solutions for the efficient handling of energy, modern

communication technologies and individual high-security solutions in a globalised world.

Infineon will use its core skills in the fields of radio-frequency electronics, system integration,

analogue or analogue-digital circuits as well as power semiconductors in order to participate

in the future growth generated in these three key areas, and also to further expand its

leading position in automotive and in industrial electronics, wireless communications and

with chip cards and security solutions. Infineon has focused its activities on these four

target markets and, for many years, has been one of the market leaders in these particular

fields thanks to its high technological competence and innovative products.

5 Strategy Analytics, December 2008, CAGR 2009-13 Summary

13

中国半导体行业历史及现状

The semiconductor industry

Semiconductors are extremely

important in virtually all areas of life.

They are vital for many consumer

products.

Demand for semiconductors is driven

mainly by the end markets in

communications, data processing,

consumer electronics, the automotive

industry and the industrial sector.

14 A change of pace for the semiconductor industry? B The semiconductor industry The semiconductor industry has changed society in uncountable ways. Semi-conductors are installed in virtually all technical equipment, ranging from dish-washers, microwave ovens and flat screens to machine tools. The use of semi-conductors in cars, trains, aircraft and ships is constantly expanding. Mobile telephones consist mainly of chips. PCs, servers and pocket calculators owe their existence to the development of semiconductors. The world would be different without semiconductors and would surely be slower. And semi-conductors have also laid the foundation for worldwide networking through the Internet. The development process continues to move ahead. Gordon Moore, a co-founder of Intel, formulated what came to be known as Moore’s law in the late 1960s: the number of transistors that can be placed on a standard processor will double every 18 to 24 months. This law remains valid today and helps explain the tremendous pace of innovation. However, progress is expensive, particularly in the semiconductor industry. Manufacturing processes and product development are based on the latest technologies, which require substantial levels of investment. Continuous research and development work is necessary, and the capital needs of the industry are accordingly high. 1 Market structure The semiconductor industry forms part of a complex interaction among various industrial sectors. In general, demand for semiconductors does not emanate directly from end users, but rather is determined by the related end-customer markets. Most sales (around 38%) are generated by data processing, followed by communications (26%), consumer electronics (18%), industrial accounts (approximately 10%) and the automotive industry (approximately 8%).1 Suppliers to the semiconductor manufacturers provide the raw silicon that is processed into wafers (see the related discussion) and develop installations for manufacturing semiconductors. The semiconductor manufacturers can have various business models and can be classified as integrated device manufacturers (IDM), fabless, licensing, foundry and back-end processes (assembly and test, packaging). The business models differ in their value creation. Figure 1 shows the way in which the various sectors interact. 1 PwC analysis, figures for 2008

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

Fig. 1 The semiconductor industry and its environment

● Integrated device manufacturers (IDM) are companies which operate along

the entire value chain in semiconductor manufacturing. The considerable

levels of investment for a new semiconductor production facility have resulted

in many IDMs changing over to what is known as a fab-light strategy. This

means that existing production capacities are retained and that newly

developed semiconductors, which require more modern manufacturing

procedures – for instance, as a result of very small feature sizes – are

manufactured by partner companies.

● On the other hand, fabless companies focus exclusively on research and

development, as well as sales of products. They do not have their own

production facilities; they use semiconductors manufactured by other

companies. Not only are no costs incurred for establishing production

facilities, no fixed costs are incurred in conjunction with these factories.

● Some companies devote themselves exclusively to licensing (intellectual

property or IP companies). They specialise in the design of certain modules

and license the resulting intellectual property to their customers. Unlike

fabless companies, IP companies do not have sales operations and license

their design and development services exclusively to third parties. There are

also companies which focus on electronic design automation (EDA).

Compared with the other business models, the volume of sales generated by

IP and EDA companies is a small part of the overall market.

● Foundries do not do product development; they manufacture semiconductors

in their own facilities for other market participants, such as fabless

companies. Foundries mostly operate large and modern production facilities,

which they attempt always to operate at high levels of capacity utilisation, by

managing commissioned production. The semiconductor industryThe main business models in the semiconductor industry are integrated device manufacturers, fabless, foundry, and assembly and test. Suppliers such as manufacturers of installations and equipment also have an important role to play.

15

中国半导体行业历史及现状

The semiconductor industry

16 A change of pace for the semiconductor industry? ● The process of increasingly specialising on specific tasks has led to the emergence of companies which focus on back-end processes such as testing and packaging (assembly and test). Open-market assembly and test service providers are based mainly in South-East Asia, particularly in Taiwan, Singapore and Malaysia, drawn by the low labour costs. The degree of automation in these back-end processes is not as high as in other areas of semiconductor added value. Semiconductors are broken down into discrete semiconductor elements, such as diodes, thyristors and transistors, and integrated circuits that connect numerous discrete semiconductors and thus provide a high degree of functionality. Integrated circuits include memory, micro and logic families (see Figure 2). Most sales are attributable to memory modules (approximately 18% of the worldwide semiconductor market), microprocessors and microcontrollers (approximately 21%) and logic semiconductors (approximately 29%).2 Digital integrated circuits accordingly account for virtually 70% of the worldwide market volume. The standard products consist mainly of memory products, which face fierce competitive and price pressure. With application and customer-specific products, price pressure is less pronounced because of the focus on specific application aspects or specific customers.

Fig. 2 Product families of integrated circuits Effective entry barriers to competitors are customer contracts, earned through a design tender procedure, and intellectual property rights for certain application characteristics. Product development and chip design in line with the needs of customers are extremely important to be successful in design tender procedures; they accordingly constitute factors of success. 2 Source: PwC, figures for 2008

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

In the case of standard products, efficient mastery of the production process is

a critical factor, in view of the typically high production scales. The ‘yield’, a

parameter which provides information concerning the number of non-defective

chips from each wafer processed, is a key metric. Moreover, effective capacity

and its utilization are important for the production of standard products, because

the modern production facilities used, for instance, in memory production have

a very high capacity and require very high levels of investment to set up.

Figure 3 compares products and certain characteristics, as well as factors of

success of standard and specific products.

Fig. 3 Comparison of standard and customer-specific products, as well as typical factors of

success

Mr Cuomo, in the aftermath of the crisis, do you expect consolidation in the industry?

Consolidation in the industry is happening now. It happens in two ways: one is that

companies go out of certain businesses; the other is through transactions. We have

already seen some M&A, but now we see companies that are starting to lose substantial

market shares in selected applications. And this trend is going to continue, especially on

the platform side. Consolidation will happen whether through M&A or not. And we have

seen M&A for the sake of M&A – at times the last step before collapse.

How do you think that business models within the industry will change? Are we going to

see the fabless trend continue?

It depends very much on the winning chip architectures. We are at the crossroads: we

either will see an evolution of today’s business, or we will see processor-centric

architectures in many applications, such as netbooks and smartphones. If the processor-

centric business will expand to market segments beyond the PC, then we shall experience

very important structural changes in the competitive scenario. Just to mention one, it is

very difficult for semiconductor companies to keep the design function separate from the

production process, because in the processor world the relation between chip design and

manufacturing process is extremely tight. As a consequence, there will be reduced space

for foundries, and the need to own a modern fab will trigger a very fast industry

consolidation process. If, on the other side, this business model does not succeed and

System-on-a-chip (SoC) prospers, then most companies will go fabless and we will have a

longer time horizon for consolidation. The semiconductor industryAndrea Cuomo, Executive Vice President, General Manager, Sales & Marketing, Europe, Middle East and Africa, STMicroelectronics

17

中国半导体行业历史及现状

The semiconductor industry

“We are at the crossroads: we either

will see an evolution of today’s

business, or we will see processor-

centric architectures in many

applications such as netbooks and

smartphones.” 18 A change of pace for the semiconductor industry? Which philosophy with respect to architecture will have an edge in the medium term? It is going to be a battle between the SoC and the processor-centric view, and I am not sure who will win. For certain applications, processor-centric architecture may have advantages, such as faster time to market and software flexibility. But the negative aspects are higher costs, higher power consumption, the commoditisation of hardware and the consequent progressive reduction of the role of System OEMs. On the other hand, SoC architecture provides lower power consumption, higher efficiency and is a means of differentiation. A key aspect in this discussion is the software architecture on silicon. The amount of software running on our devices is huge, and people will not be able to redesign their software with each new device. Therefore, basic choices about software architecture will dictate the future of the industry. Which business model wins remains to be seen. And often life is about different shades of grey rather than a straight choice between black and white. What do you view as the key growth areas for the semiconductor industry? We see two major trends. First, data is going mobile. The movement toward netbooks will be as big as the run on mobile phones some years ago. We had voice going mobile years ago, and now we see data going mobile. We are going to see cheap wireless connections, and everything will be interconnected by wireless. The second big trend is the penetration of silicon into all aspects of daily life. The first examples that come to mind are silicon in medical devices, micropayment systems, food chain control and personal identification. Do you expect that the e-mobility trend – the development of electronic cars – will have a big impact on the industry? Of course. But I would not only say e-mobility. Rather, I see ‘going green’ at large as a major trend gaining more and more importance. What are the key drivers for profitability in the industry? First of all, of course, Moore’s law and being on the leading edge of technology. Second, it is definitely the available production capacity and the usual overcapacity/shortage cycles. A third, and at times forgotten one, is product innovation. The first player in the market enjoys good margins, but for followers to attract customers they have to cut prices. When doing business forecasts, such events are often underestimated, as they tend to occur earlier than expected. Of course, as technology becomes increasingly complex and application lifetime is shortened, we may see more single source designs, and therefore some of these dynamics will change. Do you think that Moore’s law will remain the industry standard? And for how long? A long time! Die shrinking is not always pushed by market needs. It is often driven by competitive behaviour. If you look at the foundry side, some players are willing to push the technology as much as they can just to make it harder for others to follow. We are coming to a point where the complexity, the cost of technology, market dimensions and market positioning are determining if consolidation will happen, simply because some players will not be able to compete on that level any more. In other words, as the game gets tougher, market leaders may be tempted to push their technology advantage to the point where they can marginalise their competitors.

中国半导体行业历史及现状

A change of pace for the semiconductor industry? The semiconductor industry

2 Semiconductors in the crisis

The central role of semiconductors and the constantly rising demand for them

are reflected in the continuous trend of growth in the sector. Industry sales have

more than quadrupled over the past 20 years and have achieved annual

average growth of 9% (CAGR) between 1988 and 2008. Fig. 4 Semiconductor sales

However, this growth should not disguise the semiconductor industry’s

exposure to considerable volatility, with specific and defined cycles that closely

correlate to economic cycles (see Figure 4, which shows the development of

worldwide semiconductor sales between 1998 and 2008). In growth years,

boosted by strong demand and profits, the semiconductor industry built up

significant production capacities, because this was the only way to finance the

high costs of setting up new installations. The surplus capacities which existed

after the years of economic growth (e.g., in 1995, 2000 and 2007/8) exerted

pressure on prices, with corresponding consequences for profitability and

growth. Production capacity declined, semiconductors were affected by a

shortage of supply, and prices stabilised. This, in turn, resulted in market

growth, and the process started anew (e.g., in 1998 and 2002). The correlation

between the cyclical nature, production capacities and supply and demand is

not the only factor with a major impact on sales. Seasonality also influences

supply and demand patterns. For instance, consumer electronic products

(mobile telephones, MP3 players and computers) are mainly purchased in the

run-up to Christmas, and demand for chips reflects that seasonality.

However, the current crisis is significantly different than previous cycles, with

two factors that exacerbate each other. On the one hand, in mid-2008, the

industry was in a market that was already declining, with excess capacities and

falling prices, such as those affecting DRAM memory modules. On the other

hand, demand dropped significantly in the industry’s main markets. Automotive

sales declined by 4% in 2008, a slowdown was reported by PC manufacturers,

and demand for mobile devices and consumer electronic products has been

declining since the end of 2008. This trend has continued in 2009.

The DRAM manufacturers, which together with the processor manufacturers

constitute the main market group in the semiconductor industry, are suffering

the most from low sales prices and comparatively low sales volume. Indeed,

even before the crisis, the prices for some DRAM memory modules were lower Between 1988 and 2008, global semiconductor sales increased annually, on average, by 9%.

The current crisis affecting the semiconductor industry is mainly attributable to lower sales on user markets. The situation is being exacerbated by global excess production capacities.

19

中国半导体行业历史及现状

The semiconductor industry

Peter Wennink, CFO

ASML Holding N.V.

“2010 will be a transition year for the

semiconductor industry, while we

might observe full recovery in 2011

and 2012. That is the kind of time

period you have to think of.”

20 A change of pace for the semiconductor industry? than their production costs. This was due not only to the excess capacities which had built up, but also to state subsidies, which, nowadays, are necessary to keep the manufacturers in business. Japan, for instance, has intervened at Elpida; Taiwan is assessing the possibility of consolidating several chip manu-facturers into one large Taiwan Memory Corp.; South Korea has again provided considerable aid to Hynix, following its support at the beginning of the decade that enabled Hynix to grow into the second-largest DRAM manufacturer. In these Asian countries, the semiconductor industry, with its high percentage of direct and indirect jobs, forms an important part of the economy. DRAMs also form a major element within the value chain of computers, sales of which will also increase. Accordingly, various governments are attempting to retain these key industries. However, these subsidies counteract the reduction of worldwide capacities that was initiated by market mechanisms through price erosion. Accordingly, the situation for DRAMs may not improve in the foreseeable future, and more and more state aid will be necessary to keep DRAM manufacturers in business. Worldwide sales of computers and servers, the most important customers for microchips, have also declined appreciably amid the economic crisis. Moreover, particularly in consumer business, signs are evident of a considerable shift in sales in the direction of smaller, less expensive computers such as netbooks. This is reflected in lower average sales revenues and has had a negative impact on the market for microprocessors. Logic semiconductors and analogue semiconductors also have to contend with lower sales volumes. However, this is less likely to produce lower prices than, for instance, as is the case with DRAMs, because the products are not traded on spot markets. Nevertheless, if customers are aware of their increased power in terms of prices, they are likely able to force through lower supplier prices. The following chapter considers the positioning of the companies in the various semiconductor segments.

When do you expect the current crisis in the semiconductor industry to end? Historically, the trend line in the semiconductor industry has been 8 to 9% unit growth over the last 20 years. Without doubt, we are currently lower than the trend line, but the industry is rebounding from the bottom. After stopping production and selling inventory, we have seen an increase in sales of some major electronic products, like PCs. Restocking of inventory is taking place, and semiconductor companies are able to report better than expected earnings. Additionally, the important and necessary investments in technology transitions which were on hold for about nine months are resuming again. But for demand to gain historical growth rates, we need to see some economic recovery. The year 2010 will be a transition year for the semiconductor industry, while we might observe full recovery in 2011 or 2012. That is the kind of time period you have to think of. You mentioned the important investments in technology transitions. When do you think the technological transition to 45 nanometres is going to happen? It is happening this year. The transition to the advanced 45 nm will come first in logic. At this stage, the leading foundries are ramping up production of the 45 nm process, and the others will follow next year. In DRAM, the full transition to 50 nm started in the second half of 2009 and is continuing. The technological leaders in the DRAM industry are even going to pilot production of 40 nm by the end of the year.

中国半导体行业历史及现状

A change of pace for the semiconductor industry?

And will there also be a transition to 450 mm wafers?

The transition to a bigger wafer is driven by cost. The most effective way to reduce cost is

by applying Moore’s law, which describes the regular trend to shrink the size of transistor

on a chip. There are discussions about increasing the size of wafers, but a major issue is

the big investment required by the equipment industry for the development of 450 mm

tools. Without the commitment from the customer base, the equipment makers will not

foster the transition to 450 mm wafers.

Do you think market consolidation will play a bigger role in the future? To what extent will

this have an impact on your business?

We are already seeing market consolidation. For example, in the flash business there are

effectively only four players left, with two of them having close to 80% market share.

Market consolidation in the semiconductor industry is an inevitable trend which actually

started at the end of the last decade and will continue. The main driver for consolidation is

the need for economies of scale. R&D costs are ever-increasing, and only by achieving

production economies of scale can firms cover their R&D expenses. We saw a lot of logic

producers that admitted they did not have enough market share to fill one of the huge

300 mm fabs, and eventually they consolidated. The need for economies of scale also

explains the emergence of many joint-venture structures. The impact on our business is in

principle the same. We see that the technology requirements and R&D expenses are

continually increasing and that the drive for scale gains in importance. I think we will see

more consolidation in the future.

We talked about the developments in the market, but these days, many companies are

concerned about access to capital markets as well. What are your medium- to long-term

expectations concerning access to capital, especially for risky businesses like semiconductors?

In recent years, money has never been that cheap and I think clearly that will not be the

case in the future. Access to capital will still be there. In fact, profitable companies will

always have access to capital and will be able to refinance. Consequently, all elements

which drive profitability, such as achieving economies of scale, are of major importance for

the semiconductor industry. In the second quarter of 2009 the capital markets opened up

a bit, and our customers were able to finance transactions. So I think access to capital will

be there, but it will be for a selective group and it will more expensive than in the past.

Are companies currently more focused on liquidity management than they were in the

recent years? How does that affect you when your customers are focusing more on

liquidity and working capital management?

Without doubt, liquidity management has become an important issue. In times of the credit

crises when credit lines are limited, companies focus on cash and usually find it in their

working capital and in their fixed assets. The semiconductor business is very capital

intensive, and liquidity is everything. Capital markets opened up a bit, which eased the

situation, but liquidity management is still of great concern. Our customers are demanding

more favourable payment terms and are tightening their capital expenditures. That will

remain the case until there is a more relaxed financial situation in the industry.

Do you expect that financial investors such as private equity firms will play an increasing

role in the future, because they have access to capital and might find relatively cheap

companies in the market?

Companies might be relatively cheap, but you always have to ask yourself why they are

cheap. But, in general, I think the current appetite of private equity firms is not as high as it

used to be. Money was very cheap, which it isn’t today. Banks and investors are not

willing to lend their money at the same cost as one or two years ago. Consequently,

private equity firms have to put more of their own stake into the venture. We all know that

the semiconductor industry is a very cyclical business, and I think that an element of

conservative financing is strongly required.

What about R&D spending in the current crisis? Do you expect that R&D spending will be

cut in the next few years?

R&D is absolutely essential in the semiconductor industry. Every transition drives down

costs eventually. It is just a matter of time until companies which are not improving their

technology are driven out of the market. In this sense, R&D is almost a prerequisite for

existence. I don’t think that the willingness to spend on R&D is going down, but R&D

spending will be limited to those companies which can actually afford it. R&D spending is

not going to be lower as a result of less access to capital. Instead, the industry will use the

R&D budgets more efficiently by consolidating and cooperating. The semiconductor industry

21

本文来源：https://www.bwwdw.com/article/8mt4.html