Click for next page ( 32


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 31
3 Innovation Policy Landscape Comparative Analysis F undamental physical limitations in semiconductor scaling have slowed future expectations of improved, single-processor computing performance on which all sectors of society in the United States--and around the world--have relied. Many of by government investment in basic academic research. In contrast, the European Union and emerging economies such as China, Korea, and Taiwan rely much more on the government to articulate strategic objectives and key parameters. these technological challenges at the frontiers of device In the United States, there is a widespread design, computer architecture, and parallel programming expectation that government-centered innovation methodologies were described in Chapter 1. While many systems will "naturally" converge to a U.S.-style market- short-term technological fixes have led to recent led system. However, comparative research on national computing performance improvements, no silver bullets innovation policies suggests that this convergence is have emerged to reclaim the steady exponential limited.1 In addition, innovation policies change over computing performance gains once achieved by time, even within the same country. As Charles Vest successive generations of single microprocessor emphasizes, the American innovation system has a long computer systems. tradition of highly decentralized, market-driven How, then, will the next generation of innovation networks, where the government historically semiconductor, computer architecture and programming played a role primarily at the local level.2 However, as breakthroughs come about? What types of policies and ubiquitous globalization disaggregates manufacturing, institutions, whether public, private or partnerships, will product development, and research, it is not yet clear be required to bring about the technological innovation which policies will best support future innovation in the necessary for next-generation hardware devices, system United States.3 architectures, and programming systems? To address To understand how requirements for innovation these questions, it is important to understand the role policy differ across industries and technologies, it is innovation policies have played in supporting U.S. innovation and, in the context of this report, the policies 1 See R. R. Nelson, 1993, ed., National Innovation Systems. A that strengthen, sustain, and/or erode the innovative Comparative Analysis, Oxford University Press, New York. For capabilities of the computer and semiconductor an analysis of the persistent diversity of China's and America's industries. innovation and standards policies, see D. Ernst, 2011, Indigenous Innovation policies differ across countries, Innovation and Globalization: The Challenge for China's Stand- industries, and technologies. Countries differ in their ardization Strategy, University of California Institute on Global levels of development and in their economic institutions, Conflict and Cooperation, La Jolla, CA, and East-West Center, Honolulu, HI, 123 pp. and hence pursue quite different approaches to 2 C. Vest, 2011, "Universities and the U.S. Innovation System," innovation policy. The United States model, for instance, in C. W. Wessner, ed., Building the 21st Century. U.S.-China Co- has largely been based on a belief that market forces operation on Science, Technology, and Innovation, Washington, (which include government and defense as consumers D.C.: The National Academies Press, pp. 70-73. 3 that demand leading-edge technology) and the private D. C. Mowery, 2009, Plus ca change: Industrial R&D in the sector should play a primary role in innovation, backed "third industrial revolution," Industrial and Corporate Change, (18):1, pp. 1-50. 31

OCR for page 31
32 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING useful to consider how innovations differ--in the programs. Another variant of innovation policy can be complexity of the infrastructure and capabilities required found in the European Union's recent push toward new to foster and implement them. Furthermore, the demands forms of cross-border coordination of innovation of innovation policy differ, depending on the nature and markets and infrastructures. the intensity of innovation barriers that constrain the Section 3.1 provides a history of the U.S. deployment of new ideas, inventions, and discoveries semiconductor industry and examines how America's into commercially successful products, services, and decentralized market-driven innovation system has led to business models. where the United States is today. Section 3.2 looks at Today, the effects of globalization extend across all China's indigenous innovation policy, especially its stages of the value chain, including engineering, product recent Strategic Emerging Industries (SEI) Program. development, and applied and basic research. This has Section 3.3 examines the evolving role of Taiwanese resulted in an increase in the organizational and innovation policies to support low-cost and fast geographic mobility of knowledge.4 However, the new innovation through domestic and global innovation geography of knowledge is not a flatter world where networks. Section 3.4 looks at Korea's coevolution of technical change and liberalization spread the benefits of international and domestic knowledge linkages. Section globalization rapidly and equally. Instead, even mature 3.5 examines the European Union's recent efforts to and established technology and manufacturing leaders develop an integrated innovation strategy and its recent now face competition from a handful of new--yet very Key Enabling Technologies (KETs) Program. Section diverse and intensely competitive--manufacturing and 3.6 provides concluding remarks and policy research and development hubs around the world.5 implications. Therefore, the United States can learn a great deal by looking at the strengths and weaknesses of alternative 3.1 Development of the U.S. Computer and information technology (IT) innovation policies in other Semiconductor Industry nations.6 An analysis of these diverse approaches to innovation policy is shaped by issues such as: the range 3.1.1 Historical Context of policy options that have been pursued, how policy approaches differ, how these differences affect Several factors influence the range and type of innovation capacities, and how innovation policies policy options available to nations to promote and pursued elsewhere affect the global supply chain. manage development and competitiveness in the This chapter examines the strengths and weaknesses semiconductor, computer architecture, and software of different innovation strategies, policy tools, and programming arenas. Among those factors historically institutional arrangements implemented in countries that dominating U.S. policy considerations are are potentially important players in the development of computing devices, technologies, and products. While The economic importance of semiconductors and U.S. innovation strategies have primarily relied on computing in the U.S. national economy; market forces and the private sector, it is important to The economic importance of closely related U.S. understand the varied and complex factors that drive the industries (e.g., telecommunications, consumer evolution of different national innovation ecosystems. electronics, military and aerospace); For example, countries such as China and Taiwan have The outlook on the U.S. federal budget, the climate relied on top-down government leadership to define for public and private investment, the employment strategic objectives and key parameters of innovation picture, and predictions on economic growth; Political perceptions about the health of these industries relative to others; 4 D. Ernst, 2005, "The New Mobility of Knowledge: Digital In- Public perceptions about the United State's formation Systems and Global Flagship Networks," in R. Latham competitive commercial position, as well as and S. Sassen (eds.), Digital Formations: IT and New Architec- leadership of the United States vs. other nations, in tures in the Global Realm, Princeton University Press, Princeton these industries; and Oxford. 5 D. Ernst, A New Geography of Knowledge in the Electronics Both real and perceived dependence of U.S. Industry? Asia's Role in Global Innovation Networks, Policy intelligence and national security on leadership in Studies, No. 54, August 2009, East-West Center, Honolulu, HI, 65 these industries, and U.S. reliance on foreign pp. technologies and assistance in areas related to 6 This is in line with Jacques Gansler's argument for a "global intelligence and national security; and strategy" made for the U.S. defense industry (J. Gansler, 2011, Democracy's Arsenal: Creating a Twenty-First-Century Defense Prevailing political philosophies regarding industrial Industry, The MIT Press, Cambridge, MA). policy.

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 33 The last two decades have exposed significant economic pressure.9 The perception of favoring certain conflicts among these traditional influences on policy, industries, "picking winners," by government pressures largely brought about by three important changes: (1) the and incentives, rather than allowing for natural market end of the cold war, (2) the general stagnation of the forces and laissez-faire investment, has been politically Japanese economy, and (3) the globalization of the toxic. On the other hand, rescuing at least some computer and semiconductor industries into well- foundering industries, or attempting to regain lost ground established7 and mutually dependent supply chains and in critical ones, has been generally politically rewarding. markets. The Asian competitor nations, for example, China and Federal funding of electronic development, from the Japan, traditionally have both subsidized and protected launch of Sputnik in 1957 almost until the fall of the (by legal and covert subsidies and tariffs) those Berlin Wall, was driven by perceived military industries that they choose to target. requirements, which had significant noneconomic In contrast, it is important to recognize that U.S. motivation. During this period, significant federal R&D industries, and information technology in particular, do investment was made in innovative semiconductor not tend to receive attention or assistance from federal technology for military application. After some cost sources simply because they are slowing down in growth reduction and normal technology adoption delay, the or maturing; there typically must be a specific adversary. same technology and technology roadmap steadily For example, once U.S. superiority in electronics and appeared in the commercial market, including advanced computation (e.g., for guidance systems) over the Soviet compound semiconductors and dramatically new Union became assured, the focus of government policy manufacturing equipment, which also found strong switched to the rising Japanese dominance in electronics, commercial adoption, for example, in lithography. especially including memories. By the mid-1990s this pattern had reversed; that is, While countries such as Japan began forming R&D the incredible acceleration of the personal computer (PC) consortia as early as 1956, the practice was illegal in the and server industries meant that commercial technology United States until Congress passed the National was leading rather than lagging behind military Cooperative Research Act in 1984.10 Two years later, technology. This shift led to an increasing focus on the concerns of a U.S. decline in semiconductor market use and adaptation of commercial off-the-shelf share prompted a call by the Semiconductor Research technology in federal procurement and contributed to the Corporation (SRC) and Semiconductor Industry steady decline in federal funding for R&D,8 given the Association (SIA)11 for increased cooperation to provide U.S. preeminence in the area and the already high levels the U.S. semiconductor industry with the capability of of research investments by the U.S. computer and regaining world leadership in semiconductor semiconductor industry. manufacturing. As a result of this effort, SEMATECH Today, cutting-edge R&D in semiconductors, the (Semiconductor Manufacturing Technology) was created historical engine of computer performance growth, has in 1987 as a partnership of 14 U.S. semiconductor become unmanageably expensive for the usual U.S. companies with the Defense Advanced Research federal agencies. At the same time, it is extremely Projects Agency (DARPA), which contributed U.S. $500 difficult for industry to invest in long-term R&D, given the near-term expectations of the financial markets. One 9 consequence has been limited commercial R&D U.S. innovation policy can be thought of as "market conform- ing" in its intent to address problems that economists have deemed investment in hardware and software technologies whose weaknesses for technological advancements. In particular, these economic return is not realized rather quickly. were externality problems that required collective R&D funding In the United States, industrial policy has typically and that funding took specific paths because of appropriation pro- not been viewed as an offensive tool for economic cesses in Congress. 10 competition or a means to create new industries or D. V. Gibson, and E. M. Rogers, 1994, R&D Collaborations accelerate successful ones, but rather as a defensive tool on Trial, Harvard Business School Press, Cambridge, MA. 11 "Founded in 1977 by five microelectronics pioneers, SIA to protect or restore existing industries under competitive unites over 60 companies that account for 80 percent of the semi- conductor production of this country." (see www.sia-online.org) SIA, along with the European Semiconductor Industry Association (ESIA), the Japan Electronics and Information Technology Indus- 7 While well established and interdependent, these value chains tries Association (JEITA), the Korea Semiconductor Industry As- can be highly vulnerable to sudden disruptions from natural dis- sociation (KSIA) and the Taiwan Semiconductor Industry Associ- asters, geopolitical conflicts, and so on. Some of these are dis- ation (TSIA), sponsors the International Technology Roadmap for cussed in greater detail in Chapter 4. Semiconductors, a15-year assessment of the semiconductor in- 8 J. Gansler, 2011, Democracy's Arsenal: Creating a Twenty- dustry's future technology requirements (see www.public.itrs.net). First-Century Defense Industry, The MIT Press, Cambridge, MA. Last accessed on June 30, 2012.

OCR for page 31
34 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING million over 5 years, to solve common manufacturing memories based on technology from IBM, to avoid problems and to regain U.S. competitiveness in the dependence on Japanese vendors. However, by early semiconductor industry that had been lost to Japanese 1990 the consortium members had proven unwilling to industry in the mid-1980s12. In the committee's view, make the necessary investments, and major memory SEMATECH played a strong role in early efforts to users like Apple, HP, and Sun did not participate, so the reclaim U.S. semiconductor manufacturing leadership project was canceled. Thus far, consortia that include IT and has been a successful example of a U.S. consortium competitors but that do not have government leadership demonstrating the value of federal funds and federal have fared poorly, due to a combination of mutual participation. This position is reiterated by a 2002 suspicion, lack of focus, and no real sense of urgency. National Research Council report, Government-Industry In summary, U.S. federal support and investment Partnerships for the Development of New Technologies, has historically relied upon a perception of military which found that the SEMATECH partnership directly threat, economic decline, industry crisis, and/or loss of contributed to the global competitiveness of U.S. competitive position; and in the United States, electronic industry, specifically the resurgence of the U.S. and computer consortia without both federal R&D semiconductor industry.13 support and federal direction have not generally Today, the SRC also continues to play a significant succeeded. Thus, centralist technology policies that may role in advancing the semiconductor industry though work in nations and cultures that accept such direction synergetic industry and university research programs and readily are a poor match to the U.S. free-market model. support initiatives around the world, such as the Global Further, innovation policy has to reflect each country's Research Collaboration Program, Nanoelectronics unique economic institutions, industry structure, and Research Initiative, Focus Center Research Program, and growth model. Semiconductor Research Corporation Education Alliance. The National Nanofabrication Infrastructure 3.1.2 Global Semiconductor Competition Network (NNIN) also provides a successful example of U.S. government (National Science Foundation) support While it could be proposed that some U.S. computer of university semiconductor research. By paying for vendors "failed to innovate," or "gave up the fight" to some expensive semiconductor research equipment at foreign competition, it is important to recognize the universities, the NNIN enables leading-edge research, paired advantages and shortcomings of a free-market which indirectly supports the U.S. semiconductor industrial economy, and the capacity it provides for industry with research results and science and innovation, not only in technology, but in the creation engineering graduates. (and destruction) of whole economic sectors. U.S. capital In contrast, two other industry-only consortia, market investors are often quick to spot and to capitalize started near the same time and for similar reasons, both on transformative shifts in a business paradigm, and, failed. The Microelectronics and Computer Technology consequently, to move their investments in a way that Consortium (MCC) was formed in the early 1980s as a often accelerates the change. Capital markets tend to response to Japan's Fifth Generation Computer Systems value short-term quarterly profits and tend to reward or (FGCS) project.14 Entirely funded by corporate partners, punish a company accordingly, which manifests in MCC worked on a wide range of technology and changes in its stock price. This has advantages and software projects, with early sponsorship particularly disadvantages. On the one hand, it discourages waste and from mainframe computer companies. By 2000 the encourages competitiveness. On the other hand, a short- Board of Directors had decided to dissolve the term focus often discourages long-term thinking and organization. Another industry-only consortium, U.S. R&D investment, particularly during difficult economic Memories was organized in 1989 to manufacture times. Federal R&D programs and public-private consortia play a crucial role in coping with this tension. 12 In the late 1980s, when it appeared that focused See www.sematech.org/corporate history; www.sematech.org/ corporate/timeline; NRC, 2003, Securing the Future: Support to the government programs in Japan, as well as unfair or Semiconductor Industry, Washington, D.C.: The National Acad- unreasonable trade practices, might overtake U.S. emies Press (available online at http://www.nap.edu/catalog.php? competitiveness, DARPA investments, especially record_id=10677#toc). SEMATECH, drove the necessary R&D efforts in 13 NRC, 2002, Government-Industry Partnerships for the process and equipment to sustain Moore's Law and to Development of New Technologies, Washington, D.C.: The maintain the confidence of capital markets. National Academies Press (available online at http://www.nap.edu /catalog.php?record_id=10584). Concurrently, IBM began to accelerate its investment in 14 Kazuhiro Fuchi, 1984, Revisiting Original Philosophy of Fifth very high-performance semiconductor technology and to Generation Computer Systems Project, FGCS, pp. 1-2. form joint innovation partnerships with numerous (non-

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 35 Intel) semiconductor fabrication companies, creating a leadership in information technology generally. business counterpoint to Intel. More recently, the rise of Although federal support to long-term R&D has been mobile computing devices has created new competition, indispensible, particularly in bad economic both among existing companies and new ones formed in environments, such achievements would almost certainly response to emerging market economies. not have been possible under a centrally managed policy regime. For instance, in a more managed environment, 3.1.3 Creation of the U.S. (and Global) Software the policy impulse might have been to save legacy Industry15 computer companies; instead, market forces coupled with a noninterventionist approach and (to a lesser Continuous technical innovation that sustained extent) government antitrust efforts helped ensure the Moore's Law (and exponential growth in computing "creative destruction" that has transformed computing performance) led also to the creation of the commercial and the role of the United States in it. The United States software industry as a meaningful force in the U.S. has been rewarded by the emergence of very strong economy. This took place in two ways. First, the falling semiconductor design and software industry cost and wide availability of powerful microprocessors leadership--in exchange for the loss of some greatly increased the number of computers in use, and semiconductor fabrication and the assembly and testing successful software products could be sold in enormous of commodity products to foreign vendors, for example, numbers at modest prices. the off-shore assembly by contract manufacturers of Second, the fact that a small number of instruction even the strongest U.S. computer brands, based upon set architectures (ISAs) dominated the PC and server cost. marketplace16 meant there was a larger and consolidated However, staying strictly on any technical path software market that would benefit from steady involves bypassing others, and sacrificing progress in improvements in cost and performance, while seldom some areas to sustain others. It is certainly worth requiring any significant changes to the programs. examining some of the approaches delayed or abandoned Vendors rarely prefer to use new instructions sets until by the course taken by the IT industry. they have been in the market for many years and are In the 1960s and 1970s, computers were expensive, available on a significant fraction of deployed machines. resources were limited, and programmers were scarce. This allowed larger software investments to be made, in There was great emphasis on creating clever algorithms products that would surely perform better over time, that required the least number of instructions or smallest courtesy of Moore's Law. amount of memory, or both. Elegant, parsimonious U.S. firms dominate this 30-year-old PC and server program design was celebrated, and improvements to industry, although a few European firms (e.g., SAP) are compilers for denser code and new languages for of significant size and share of the market. There is early programmer productivity were high priorities in evidence that this market dominance may extend to the academia and industry alike. High-productivity new world of smartphones and tablets, as well as cloud programming languages help programmers produce services, though global competition in this space is new working programmers faster, as compared with high- and intense. performance programming languages, which help programmers extract as much performance as possible 3.1.4 Consequences of the U.S. Free-Market Approach by exposing machine details to them. From high- productivity languages and the relentless hardware These three phenomena--the enormous growth of performance improvements enabled by Moore's Law, a the semiconductor industry, the commoditization of the new and much larger pool of programmers emerged. computer industry, and the emergence of a huge software These programmers applied application-specific industry--are mutually dependent, and have created a knowledge, for example, machine learning, graphics, virtuous economic framework. They also afforded the animation, accounting, government functions, and so on, United States the opportunity to achieve and maintain its driving an explosion in software capabilities in the era of ever-faster and cheaper central processing units (CPUs) and memory. In addition, programming emphasis moved 15 Section 3.1.3 and part of Section 3.1.4 rely heavily on David from performance productivity to getting new products Liddle, "The Wider Impact of Moore's Law," IEEE SSCS News- out faster. letter, September 2006. Available at http://ieeexplore.ieee.org/ Another area of technology research and innovation stamp/stamp.jsp?arnumber=4785858. Last accessed August 21, 2012. affected by Moore's Law was parallel computing. In the 16 The history of the mobile and embedded computing space is 1970s, very large scientific computers with parallelism much more varied, with a diversity of ISAs and vendors. among several arithmetic units were just beginning to

OCR for page 31
36 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING work well. Equally important, software researchers were in these areas should carefully consider the opportunities beginning to make real progress on the problem of and consequences of alternative innovation strategies, programming systems for parallel machines. Moore's some of which have been tried elsewhere.18 Law advances relegated most work in parallel computing to business servers and scientific and technical 3.2 China Strengthening Indigenous Innovation computing.17 A revival of parallel computing research and development in the 1990s yielded several new Over the last several decades, China has made approaches and companies, but the early promise was significant efforts to align its science, technology, and not realized, for many reasons. Because the size of the innovation policies to support indigenous innovation. technical computing market was small relative to the PC These trends towards technonationalism19 were market, the research and product development took a prompted by political concerns within China that it both different path, focusing on performance maximization at lacks indigenous technology and depends on foreign reduced costs rather than ease of use and programmer technology, as well as from several lessons learned over productivity. the past several decades. For example, after its Only now that the limits of growth in CPU clock relationship with the former Soviet Union ruptured in the frequency are in sight for consumer devices has serious 1950s, China shed its reliance on Soviet technology and focus on parallel processing reentered the mainstream. developed a national strategic weapons program, Earlier research from the 1980s and 1990s showed the developing its own nuclear weapons, missiles, and difficulty of developing tools that can easily convert satellites. legacy sequential codes into scalable parallel code that Then, in the late 1970s, China embraced run well on current generation or next-generation globalization by opening its huge market to multinational machines. This research experience suggests that future corporations for the exchange of advanced technology. work should emphasize simplicity and programmability In turn, the rapid growth of China's semiconductor of heterogeneous multicore devices to address consumption primarily reflects its emergence as the mainstream product needs. In addition, such research dominant global factory for IT equipment. Between 2004 should be driven by an awareness of the demonstrated and 2009, alone, China's share of global electronic limitations of automatic parallelization and recognition equipment production increased from 17 percent to 31 that the intrinsic parallelism in application problems differs markedly. Another effect of Moore's Law over the years has been that the capital markets, given the visible, vast 18 investment in scalable complementary-symmetry metal- When examining innovation strategies elsewhere, it is oxide-semiconductors (CMOS) (except for extremely important to recognize that most are being applied to those whose attempts to enter the advanced computing market were low-volume exotic noncommercial uses) have not late relative to the United States. Latecomers have disad- encouraged R&D on new post-silicon materials. Even vantages and advantages that need to be assessed and taken small deviations from the CMOS path, like silicon- into account when considering policy options and what les- germanium or silicon carbide have aroused skepticism, sons might be learned. For instance, latecomers to advanced let alone compound IIIV semiconductors. computing need to overcome very substantial entry barriers Typically, a free-market approach will continue on a (disadvantages), as well as to exploit new opportunities that result from beginning with less-complicated technology and profitable path until it begins to reach diminishing having fewer legacy constraints on technology development, returns; unfortunately, that point is sometimes strategy, and organization (advantages). Economies of scale recognized too late. The uniprocessor CMOS clock- may be a barrier to market entry requiring nonprice means of frequency race has already ended, leaving the United market penetration, that is, through product differentiation States ill prepared with either semiconductor or software and the creation of new markets and distribution channels. succession plans. This type of situation has traditionally On the other hand, latecomers who become fast followers of established technology roadmaps are able to set clear targets been one in which U.S. government participation with for product development and related research, as well as to academia and industry has been effective. Neither the compare and learn from the experience and failures of in- U.S. federal government nor the U.S. computer industry cumbent leaders. Latecomers are also not locked into sup- has come to consensus on a strategy that ensures U.S. porting and maintaining legacy technologies or infrastruc- leadership in the next generation of computing tures. 19 technologies. Thus, those developing future U.S. policy Policy orientation towards autonomy and independence from other states (see B. Naughton and A. Segal, 2001, "Technology Development in the New Millenium [sic]: China in Search of a 17 It is worth noting that parallel computing work continued in Workable Model," MIT Japan Program, Working Paper Series the high-performance computing (HPC) and server segments. 01.03., May 28).

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 37 percent.20 This marked increase suggests a reshaping of basic software, extra-large-scale integrated circuit (IC) the IT equipment manufacturing landscape. manufacturing and techniques, and new-generation However, as the global financial and economic crisis broadband wireless mobile telecommunications (mega- continues, exports from China have slowed, placing engineering programs).22 pressure on China's export-oriented economic development model. In pursuit of new growth engines Strategic Emerging Industries (SEI) Program for its economy, several government policies and initiatives have facilitated the strengthening Chinese Strategic emerging industries (SEIs) refer to indigenous innovation. industries23 associated with the development of technologies (e.g., information, biotechnology, medical, 3.2.1 Government Policies and Initiatives to Strengthen new energy, environment, marine, and space) that have Indigenous Innovation strategic importance to China; many are similar to the frontier technologies prioritized in the MLP. These SEIs Medium and Long-Term Plan (20062020) have been said to represent the future direction of industrial development in China and will play a critical In early 2006, China released its Medium- and role in its continuous and sustainable economic growth, Long-Term Plan for the Development of Science and particularly in national economic and social development Technology (MLP) (20062020). This plan set the tone and optimization and upgrading of industrial structure. for strengthening China's indigenous innovation Launched in October 2010,24 the SEI Program was capability by addressing four problems in China's highlighted as an important component of the 12th Five- scientific and technological development: (1) lack of Year Plan for National Economic and Social innovation in commercial technologies and dependence Development (20112015). As selected SEIs are science on foreign technology; (2) increasingly unfriendly and technology based, the SEI Program is expected to international environment for acquisition of foreign decrease China's dependency upon external technology technologies; (3) technological failure to meet critical and boost indigenous innovation capabilities, ultimately energy, water and resource utilization generally, and spurring economic growth and the formation of a new environmental protection and public health needs; and industrial cycle. It is expected that the government will (4) mounting technological challenges for meeting also work out financial and taxation policies to support, national defense needs.21 While not specifying what guide, and encourage capital investment, and establish indigenous innovation means, the MLP highlights three special funds for the development of SEIs. channels through which indigenous innovation Foreign-invested design subsidiaries of leading capabilities may be strengthened: (1) genuine "original foreign semiconductor companies and global original innovation," (2) "integrated innovation" (fusing together equipment manufacturers (OEMs) play an important role existing technologies in new ways), and (3) in China's chip design industry. Of the 472 design "reinnovation" (assimilation and improvement of enterprises reported in China at the end of 2009, imported technologies). approximately 100 were the design units or activities of The MLP singles out 16 engineering mega- foreign-invested or subsidiary multinational programs, as well as identifies 11 key areas, 8 frontier technologies, and 4 science mega-programs, to support in the next 15 years (see Appendix I). Many of these programs and focus areas are directly relevant to 22 See Alan Wm. Wolff, "China's Drive Toward Innovation," advanced computing, including the IT industry and Issues Online, Spring 2007. Available at http://www.issues.org/ modern services (key areas); information technology and 23.3/index.html. Last accessed August 21, 2012. 23 new and advanced materials (frontier technologies); and SEIs include the following: Energy-saving and environmental core electronic components, high-end generic chips, protection, new generation of IT, biotechnology, and high-end equipment manufacturing industries; new energy, new materials, and new energy automobile industries. State Council of China, 20 PwC, 2011, Continued growth: China's impact on the semi- 2010, Decisions of State Council on Accelerating the Cultivation conductor industry 2011 update, p 13. Available at www.pwc. and Development of Emerging Strategic Industries, G.F. No.32, com/gx/en/technology/assets/china-semiconductor-report-2011. October 29 [USITO Draft Translation]. 24 pdf. Last accessed January 27, 2012. On October 18, 2010, the State Council issued a "Decision on 21 C. Cao, R. P. Suttmeier, D. F. Simon, 2009, "Success in State the Acceleration of Nurturing and Developing Strategic Emerging Directed Innovation? Perspectives on China's Plan for the Devel- Industries," formally launching the SEIs. The 12th Five-Year Plan opment of Science and Technology," in G. Parayil and A. P. for National Economic and Social Development (20112015), D'Costa (eds.), The New Asian Innovation Dynamics: China and released in 2011, includes SEIs as one of its important compo- India in Perspective, Palgrave Macmillan, London, pp. 247264. nents.

OCR for page 31
38 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING companies.25 These foreign-invested design subsidiaries both in China and in the United States, will have to be engage in a variety of activities that range from the iteratively redefined to meet these challenges. This is simple to the complex, including adapting parent also the case should Chinese-designed computing company product standards for the China market, systems begin to have any commercial success inside or providing lower-cost capacity for standardized back-end outside of China. design functions that are integrated in the parent It is worth noting that China does not have any company's design flow, but they also include integrated representation among the largest semiconductor design projects for system-on-a-chip (SoC) designs. companies, nor does it have any mass commercial In addition to multinational partnerships, long-term processors or architectures. China does, however, have a investments in the IT and advanced computing industries seat at the six-seat World Semiconductor Council. This are now beginning to yield competitive, indigenous seat may reflect both China's growing potential as a microprocessors, designed wholly in China. Over the last semiconductor market and its increasing capabilities in decade, the Chinese Academy of Sciences (CAS) has semiconductor design. China's increasing share of been developing and prototyping their line of Loongson worldwide patents focused on semiconductor and Godson processors.26,27 During this time, each technology, from 13 percent to 22 percent in 2009, also generation of Loongson or Godson processor has suggests improvements in China's semiconductor become more capable and is now rivaling leading-edge innovative capacity. More significantly, China's share of microprocessors in performance. Similarly, the ShenWei semiconductor patents that are first issued in China has series of microprocessors,28 developed by the Jinan grown from 0 percent in 2005 to ~24 percent in 2009. Institute of Computing Technology (affiliated with the People's Liberation Army) since 2006, has found its way 3.2.2 Impacts of Government Policy Efforts into the first home-grown supercomputer in China, called BlueLight. As of November 2011, BlueLight It is unclear to what degree the MLP will enhance includes 8,704 ShenWei chips and achieves Linpack China's indigenous innovation capability. However, performance of 795 TFlops,29 while being one of the international technical and business communities have most energy-efficient computers in the world. already expressed concern over Chinese efforts to While neither of these programs has shown any support the indigenous innovation efforts. For example, commercial success either inside or outside of China, the government policies, which gave preference for investment period is long and ongoing. One would procuring domestic technologies and products in the expect that such programs aspire to serve the Chinese name of supporting indigenous innovation, were domestic market, at least for high-end technical abolished, at least temporarily, due to extreme pressure computing. As these, and other, systems diverge from from foreign governments and companies. In addition, the x86 and ARM ecosystems, they will have different some studies30 have characterized China's innovation sets of innovation and performance optimization policies as a threat to global intellectual property rights; challenges. It is likely that national innovation policies, a recent report by the U.S. Chamber of Commerce has even claimed that Chinese innovation policy is "a blueprint for technology theft on a scale the world has 25 PwC, 2011, Continued growth: China's impact on the never seen before."31 semiconductor industry 2011 update. Available at www.pwc. Lastly, the committee believes, growth in China's com/gx/en/technology/assets/china-semiconductor-report-2011. pdf. Last accessed January 27, 2012. homegrown industrial capacity, plus China's massive 26 Godson 3B is an 8-core processor with vector extensions im- urbanization, has nurtured an increasingly large domestic plemented in a 65 nm technology. It employs a MIPS instruction market in different manufacturing--and increasingly set (originally developed in the United States), runs at 1 GHz, and R&D--sectors. It is not clear to what degree other key has a peak performance of 128 GF. 27 The Loongson and Godson processors were developed under 30 the leadership of U.S.-trained computer scientist Li Guojie at the U.S. International Trade Commission, 2010, China: Intellec- Chinese Academy of Sciences' Institute of Computing Technol- tual Property Infringement, Indigenous Innovation Policies, and ogy, CAS. Frameworks for Measuring the Effects on the U.S. Economy, No- 28 The ShenWei 3 chip contains 16 cores, is implemented in vember. 31 65nm and runs at 1.1GHz. James McGregor, 2010, China's Drive for "Indigenous 29 There are two other Chinese HPC installations above Innovation": A Web of Industrial Policies, Global Intellectual BlueLight (no. 14) in the November 2011 TOP500 List. However, Property Center and Global Regulatory Cooperation Project under both of these are built from Intel CPUs and NVIDIA graphics- the U.S. Chamber of Commerce, and APCO Worldwide, processing units, delivering 2.566 PFlops (no. 2) and 1.271 PFlops Washington, D.C. Available at http://www.apcoworldwide.com/ (no. 4). The BlueLight cluster is ranked no. 39 on the Green500 content/PDFs/Chinas_Drive_for_Indigenous_Innovation.pdf. Last List of the most efficient supercomputer clusters. accessed on August 7, 2011.

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 39 industries will follow similar trends, in particular, those characteristic of Taiwan's IT industry has been its deep SEIs related to the advanced computing technologies and integration into diverse and global corporate networks of products. In overall growth, the added value of SEIs has production and innovation. In addition to facilitating the been estimated to reach U.S. ~$682 billion in 2015 and role of Taiwanese firms as fast followers, this network U.S. ~$1.8 trillion in 2020, with projected annual growth integration also encouraged IT firms to focus their R&D rates of 24.1 percent between 2011 and 2015 and 21.3 efforts on incremental innovation. Today, there is a percent between 2016 and 2020.32 growing recognition that Taiwanese firms must now increase and broaden R&D in order to avoid diminishing 3.2.3 Transition Toward Economic Outcomes-driven returns of network integration. Thus, new policies are S&T Programs needed to spur domestic capabilities for low-cost innovation in IT. Unlike most previous government-led science and While Taiwan's new innovation strategy is still a technology (S&T) programs, where the Ministry of "work in progress," some major policy building blocks, Science and Technology (MOST) was the only or which combine market-led innovation and public policy biggest stakeholder, SEI program development efforts coordination of multiple layers of private and public have primarily been led by the National Development innovation stakeholders, are taking shape. Due to its and Reform Commission and Ministry of Industry and pragmatism and openness to new forms of public policy Information Technology (MIIT), two superministries and private-public partnerships, Taiwan's innovation with strong economic missions in China's bureaucracy.33 policy may in fact shed new light on the opportunities In contrast to MOST, these agencies are expected to and challenges for strengthening America's innovation increase industry participation in the program (as capabilities in advanced computing. opposed to primary participation by universities and research institutes). As such, it is likely that future S&T 3.3.1 Taiwan's "Global Factory" Innovation Model35 programs will require strong economic components and targets and cannot be assessed by publications and Early on, Taiwan's IT industry depended heavily on patents alone. The fact that economic instead of science international markets and access to foreign technology, agencies are deeply involved in the SEI Program reveals tools, and ideas to overcome substantial entry barriers to a long-standing issue--the separation between research network participation, namely, a lack of domestic market and the economy--that China has tried to solve when it and limited resources and infrastructure. From the started to reform its S&T management system in the beginning, the key to Taiwan's success has been an early mid-1980s. Regardless, the successful transformation of integration into diverse and constantly evolving network the Chinese economy will depend upon the successful arrangements that include both formal corporate and coordination among different government ministries, informal knowledge networks. Formal corporate each with a unique mission. Achieving this will be production networks link Taiwanese firms to large global difficult, if not impossible. brand leaders (the customers), investors, technology suppliers, and strategic partners through foreign direct 3.3 Taiwan Low-cost and Fast Innovation investment as well as through venture capital, private equity investment, and contract-based alliances. Equally In the last several decades, Taiwan has experienced important are informal global knowledge networks that incredible economic growth--particularly in its IT link Taiwan to more developed overseas innovation industry, which now accounts for 70 percent of Taiwan's systems and knowledge communities, primarily in the total manufacturing R&D.34 To date, a defining United States, through the international circulation of students and knowledge workers.36 Finally, domestic 32 interorganizational linkages with large Taiwanese Zhou Zhizue, chief economist on MIIT projections. Available at http://tech.sina.com.cn/it/2011-08-04/07185880063.shtml. Last accessed on November 7, 2011. 33 Barry Naughton, 2009, "China's Emergence from Economic 35 Crisis," China Leadership Monitor, No. 29. Available at Sections 2.3.1- 2.3.4 rely heavily on Dieter Ernst, "Upgrading http://media.hoover.org/sites/default/files/documents/CLM29BN. through innovation in a small network economy: insights from pdf. Last accessed on November 7, 2011. Taiwan's IT industry", Economics of Innovation and New Tech- 34 Between 2001 and 2006, almost 90 percent of the R&D nology, June, 2010 Volume 19, No. 4pp. 295-324. Last accessed investment of Taiwan's private sector was concentrated on two August 21, 2012. 36 sectors, electronics components (56 percent) and computers and Between 1987 and 2003, this small island has been the fifth electronic and optoelectronic products (32 percent). See largest nation of origin of international students in the United http://eng.stat.gov.tw/ct.asp?xItem=6503&CtNode=2202&mp=5. States. (Guo, 2005: 142).

OCR for page 31
40 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING business groups complement these international However, over the last decade, the globalization of linkages.37 S&T and of the economy has placed pressures on Progressive integration into these diverse Taiwan's IT industry. Investments necessary for radical production, knowledge, and innovation networks has innovations that can compete with global technology enabled Taiwanese firms to combine the speed and leaders are beyond the reach of many Taiwanese IT flexibility of smaller firms with the advantages of scale companies. Even TSMC (Taiwan Semiconductor and scope that normally only large firms can reap, as Manufacturing Company), the world's leading IC well as to tap into the world's leading markets, foundry, has had to stretch its resources to the limit to especially in the United States. Network participation sustain its leadership position. In addition, Taiwanese IT has also multiplied conduits for knowledge transfers to firms are establishing low-cost supply bases in China Taiwanese IT firms, broadening their scope for learning and Southeast Asia to reduce production costs. To and capability development. This, in turn, has created expand their position as network suppliers, Taiwanese new opportunities, pressures, and incentives for firms are also moving beyond the provision of Taiwanese IT firms to upgrade their technological and manufacturing services, and developing integrated management capabilities and the skill levels of workers. service packages that include logistics and product Today, Taiwan has established itself as an important development. However, the downturn in the global "global high-tech factory" for PC-related products, electronics industry since late 2000 has exposed several handsets, wireless equipment, integrated circuits, and flat challenges for Taiwan's innovation model. panel displays.38 For global IT industry leaders, Taiwanese firms have become preferred OEM and ODM 3.3.2 Negative Effects of Network Integration (original design manufacturing) suppliers.39 In addition, Taiwanese firms have made considerable progress in Taiwan's focus on the provision of OEM and ODM product development, especially in electronic design. services has severely constrained the capacity of Beginning in the 1980s, Taiwan's leading PC firms Taiwanese firms to invest in "upgrading through low- established R&D laboratories in Silicon Valley to gain cost innovation" strategies.41 This problem is early access to the product and technology road maps of exacerbated by relentless pressure from global brand the global industry leaders and to improve their product- marketers to reduce cost and time-to-market for development capabilities. By the mid-1980s, Taiwan's commodity-type products with low profit margins that semiconductor firms became involved in board-level and are apt to penetrate mass markets. As a result, Taiwanese application-specific integrated circuit design,40 giving firms are stuck in a "commodity price trap," with rise to a broad portfolio of design implementation insufficient profit margins to support investment in new capabilities. This enabled Taiwanese semiconductor R&D, intellectual property creation and branding. firms to compete on the speed, cost, flexibility, and Furthermore, as specialized OEM and ODM suppliers, quality of providing these services. Taiwanese firms typically concentrate on incremental innovations within existing product architectures that are predefined by global brand leaders charging hefty patent- 37 D. Ernst, 2001, "Small Firms Competing in Globalized High licensing fees. This structure constrains their capacity to Tech Industries: The Co-Evolution of Domestic and International develop new products and to shape technology road Knowledge Linkages in Taiwan's Computer Industry," in P. maps and standards. Guerrieri, S. Lammarino, and C. Pietrobelli (eds.), The Global Challenge to Industrial Districts: Small and Medium-Sized Since 2003, many Taiwanese handset makers have Enterprises in Italy and Taiwan, Edward Elgar, Aldershot, U.K. attempted to increase profits by increasing their branded 38 D. Ernst, Upgrading through innovation in a small network handset sales relative to their OEM or ODM business. economy: insights from Taiwan's IT industry, Economics of Inno- However, with the possible exception of ASUS and vation and New Technology, Vol. 19, No. 4, June 2010, pp. 295 HTC,42 most of these attempts seem to have failed, 324. causing Taiwanese handset makers to switch back to the 39 An OEM contract refers to arrangements between a brand- name company (the customer) and the contractor (the supplier), OEM-ODM model. The most spectacular failure has where the customer provides detailed technical blueprints and 41 most of the components to allow the contractors to produce D. Ernst, Upgrading through innovation in a small network according to specifications. In ODM arrangements, the contractor economy: insights from Taiwan's IT industry, Economics of Inno- is responsible for design and most of the component procurement, vation and New Technology, Vol. 19, No. 4, June 2010, pp. 295 with the brand-name company retaining exclusive control over 324. 42 marketing. HTC has successfully developed own-brand touch-screen 40 D. Ernst and D. O'Connor, 1992, Competing in the Electron- smartphones, initially based on Microsoft's Windows Mobile ics Industry. The Experience of Newly Industrialising Economies, operating system, but now also on Google's open-source Android Development Centre Studies, OECD, Paris, 303 pp. platform.

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 41 been the attempt by the BenQ Group (a spin-off of the Taiwan's IC design industry provides a telling Acer Group) to accelerate its global branding strategy by example of the substantial challenges of developing acquiring the mobile handset business of Siemens and its indigenous intellectual property. As specialized suppliers intellectual property.43 to global semiconductor and system companies, Taiwanese chip design firms have limited resources and 3.3.3 Constraints to Developing Indigenous Intellectual incentives to close the technology gap relative to Property industry leaders--and as a result, they are typically not active at the leading edge of process technology and IC While Taiwan's patent filings at the U.S. Patent and complexity.48 In addition, Taiwanese design houses have Trademark Office (USPTO) have grown rapidly (in "all not been able to develop in-house complete solution patents" per million of its population and in "utility packages. For instance, in the important cellular chipset patents"44), its patent counts are highly concentrated, market, only one Taiwanese design house (MediaTek) both in terms of products (technology classes) and patent offers a complete cellular chipset solution. All other holders (assignees). In 2010 the largest number of Taiwanese companies competing in this market, such as Taiwan's U.S. patents45 were in semiconductor Sunplus and Airoha, have focused on specific building manufacturing, and these patents were dominated by blocks and niche markets. Given the rapid change and TSMC (Taiwan's third-largest patent filer, with 405 unpredictability of these markets, such a focused patents), followed by MediaTek (no. 5), Macronix (no. approach is a high-risk strategy. 6), United Microelectronics Corporation (no. 7), and Via Technologies (no. 9, with 108 patents). Hon Hai 3.3.4 Hollowing-out Through Offshoring to China (Taiwan's largest patent filer, with 572 patents), the world's largest maker of electronic components, has To retain its position as OEM and ODM suppliers to pursued an aggressive strategy to file protective patents global brand marketers, Taiwan has established low-cost (especially for its connector technology), primarily supply bases--and more recently, R&D centers--in against China.46 In addition, Taiwan's patent quality China and Southeast Asia. The increasingly common remains low (in patent citation, "science linkages," and practice of Taiwanese IT manufacturers receiving orders technological capabilities),47 and its most influential in Taiwan and shipping manufactured goods from China patents are highly concentrated with TSMC. has given rise to "a new cross-Strait division of labor along the lines of pilot run vs. mass production."; 43 Less than 1 year after the acquisition, the German subsidiary, however, this offshore outsourcing now imposes severe BenQ MobileGmbH & Co OHG, was closed amid continuing pressures on Taiwan's IT industry, as reflected by a huge losses at the subsidiary. BenQ's share of the Taiwan handset declining domestic value-added ratio that is much lower market now languishes around 8 percent. BenQ now outsources than for the United States and Japan. 49,50 handset production to Taiwanese contract manufacturers. 44 A utility patent protects any new invention or functional Another important concern is the continuing improvements on existing inventions (such as going from light- relocation of wafer fabrication capacity. Although emitting diode [LED] technology to organic LED technology), China's current wafer fabrication capacity represents while a design patent protects the ornamental design, configura- about 9.4 percent of worldwide wafer fabrication tion, improved decorative appearance, or shape of an invention. capability,51 as of May 2011, 28 new wafer fabrication China's utility model patents protect any new technical solution facilities were under construction in Greater China. Once relating to the shape and/or structure of a product, which is fit for practical use. Utility patents, which offer the same protection these begin production, it is estimated that China and (albeit for a shorter time span) as invention patents, are quicker and cheaper to obtain, because they only receive a preliminary examination rather than the full substantive examination of an linkages, Taiwan's patents, even for semiconductors, are less fre- invention application. quently cited in scientific journals than Korea's patents. 45 48 See http://www.uspto.gov/web/offices/ac/ido/oeip/taf/asgstc Joy Teng, 2006, IC Design House Survey 2006: Taiwan, /twx_ror.htm. Last accessed on January 12, 2012. courtesy of Electronic Engineering Times Taiwan (available at 46 Hon Hai has been expanding its USPTO patent portfolio, www.eettaiwan.com). 49 more than doubling its USPTO patent filings between 2006 and This hollowing-out effect, and the resultant job displacements, 2010. Since 1995, 61 percent of Hon Hai's patents were filed in may have been reduced by the growth of Taiwanese exports to China, against less than 18 percent in the United States. Asia (especially China) of increasingly sophisticated production 47 Xin-Wu Lin, 2005, An Analysis of Taiwan's Technological equipment. 50 Innovation On the Basis of USPTO Patent Data Analysis, slide S.-H. Chen, M.-C. Liu, and K.-H. Lin, 2005, "Industrial presentation, Taiwan Institute of Economic Research, Taipei, July Development Models and Economic Outputs: A Reflection on the 27. For instance, Taiwan's patents are less "original" and have less `High Tech, High Value-Added' Proposition", manuscript, "impact" than Korea's, that is, they are less frequently cited within Chung-Hua Institution for Economic Research, Taipei: p. 25. 51 a technology class and in other technology classes. As for science SEMI Wafer Fab Watch, May 2010.

OCR for page 31
42 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING Taiwan together will have a 29 percent share of total In addition to providing aggressive tax incentives,54 worldwide wafer fabrication capacity. Taiwan's innovation policy seeks to strengthen the lead As Taiwanese offshoring extends beyond role of the private sector by generating new public- manufacturing into product development, the private partnerships and by coordinating their competitive advantages previously afforded by Taiwan's interactions.55 In particular, government initiatives, such high-tech cluster (i.e., combination of flexibility, low as Taiwan's Technology Development Programs, cost, and timely service) have begun to erode. For Hsinchu Science Park, and Industrial Technology example, as production of computer, communications, Research Institute (ITRI), are intended to foster and consumer products moves to China, Taiwan's IC industrial upgrading through low-cost and fast design houses are forced to follow suit to sustain close innovation. Today, Hsinchu Science Park is the world's interaction with their customers. Once in China, leading cluster for semiconductor manufacturing. ITRI Taiwanese design houses face intense competition from also continues to play a significant role in Taiwan's IT lower-cost Chinese competitors, and they lose their most and semiconductor industries. ITRI's recent Cloud fundamental competitive advantage: access to a pool of Computing Center for Mobile Application (CCCMA) highly trained and experienced lower-cost engineers and seeks to promote Internet-based, on-demand computing managers from diverse sources. Even worse, Chinese IC (cloud computing) as a catalyst for strengthening design firms can now draw on Chinese returnees who Taiwan's software capabilities, building on Taiwan's have studied and worked in the United States, as well as strengths in lower-cost hardware, such as memory, recruit former employees of Taiwanese companies to chipsets, server, and storage network equipment. train China's growing pool of local engineering graduates. 3.3.6 U.S.-Taiwan-China Linkages 3.3.5 Government Policies to Support Low-Cost and Since its inception, Taiwan's IT industry has greatly Fast Innovation benefited from its deep integration with America's innovation system, especially Silicon Valley. As a There is a growing consensus in Taiwan that an byproduct, the United States and Taiwan have developed exclusive focus on hardware manufacturing is no longer a strong mutual dependence on each other's IT and sufficient to guarantee sustainable growth. Taiwan's new semiconductor industries. U.S. IT companies remain the innovation strategies now seek to build on its capacity most important buyers of Taiwanese ODM and OEM for low-cost and fast manufacturing by complementing services, and Taiwan's silicon foundries are a critical its contract manufacturing and component production supplier of process technology as well as manufacturing excellence with knowledge-intensive support services and design services to U.S. fabless design companies. In and a capacity to provide "integrated solutions." In addition, Taiwan exploits a first-tier supplier advantage addition, Taiwan has a long-term objective to strengthen due to the establishment of leading U.S. R&D centers in its software capabilities, especially for the design of Taiwan and to the acceleration of its "upgrading through complex system software and for cloud-computing innovation" strategy. applications.52 To implement this strategy, Taiwan's However, these relationships have been complicated innovation policies seek to strengthen further the by the emergence of China as both a partner and linkages and interactions among industry, academia, and competitor with Taiwan. In the last decade, China has public and private R&D organizations. become not only the most important production site for A defining characteristic of Taiwan's innovation Taiwan's IT companies, but also a major growth market. policy is its openness to foreign strategic advice and Not only are Taiwan's foundries, IC design houses, and knowledge sharing, distinguishing it from Japan, Korea, ODM suppliers well placed to exploit China's rapid- and China53 with their much more closed systems of innovation policy. innovation networks (see Run of the Red Queen: Government, Innovation, Globalization, and Economic Growth in China by D. 52 Interview with Dr. Tzi-cker Chiueh, General Director, ITRI- Breznitz and M. Murphree, 2001). 54 CCCMA, April 25, 2011. As is typical for Taiwan's leading inno- Taiwan's Statute for Industrial Innovation has lowered the vation actors, Dr. Chiueh's education and employment history business tax from 25 percent to 17 percent in 2010 (which com- shows strong links with the United States. See also Ministry of pares to China's 25 percent rate, Korea's 22 percent rate, and Economic Affairs (MOEA), 2011, Taiwan's ICT industry devel- Singapore's 17 percent rate). 55 opment and outlook, as reported in DigiTimes, August 29. H.-S. Chu, 2007, "The Taiwanese Model: Cooperation and 53 While China has, to some degree, followed the Taiwanese Growth," in C. W. Wessner, ed., Innovation Policies for the 21st low-cost and fast innovation model, the Chinese model differs in Century. Report of a Symposium, The National Academies Press, that it has not leveraged domestic and, to a lesser extent, global Washington, D.C., p. 120.

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 43 demand growth for IT products and services, but through domestic and global innovation networks. Taiwan's government is convinced that China is Recent policies suggest that China is following suit with gradually becoming a regional technology leader. This Taiwan's innovation model and will focus in the future reliance has resulted in new initiatives for cross-strait on low-cost mass adoption of new technologies and cooperation in industrial standards, for broader bilateral innovation. economic cooperation, especially through the Economic Cooperation Framework Agreement (ECFA),56 and for 3.4 Korea Coevolution of International and deregulation of Chinese investment in Taiwan. On the Domestic Knowledge Linkages59 other hand, continuous penetration of the Chinese market will require that Taiwanese firms also redeploy new Countries with emerging economies must rely product development and research to China. By primarily on foreign sources of knowledge as the main providing critical inputs (through training, technology vehicle of learning and capability formation. transfer, and joint product development) to Chinese International linkages are needed to pave the way for an firms, Taiwan accelerates China's ability to catch up. effective exploitation of latecomer advantages. Empirical As Taiwan's IT industry becomes increasingly research has shown that, as a developing country integrated with China's economy and its innovation progresses in its industrial transformation, its reliance on system, it is unclear how and to what degree Taiwan will international technology sourcing and knowledge strike a balance between cooperation with China and linkages substantially increases.60 The Korean cooperation with the United States. If the sheer weight of innovation system in the electronics industry is China forces Taiwanese firms to give priority to their emblematic for a heavy reliance on international links with China, how will this affect America's access linkages, combined with the development of to the semiconductor global value chain? It is too early complementary domestic linkages.61 for a conclusive answer to these questions. So far, Early on, as a part of its innovation strategy, the however, Taiwan's economic diplomacy related to the IT Korean "government encouraged some of the leading industry remains closely aligned with the U.S. position.57 chaebol62 to focus on learning and knowledge accumulation through a variety of links with foreign 3.3.7 Summary equipment and component suppliers, technology licensing partners, OEM clients, and minority joint- If Taiwan is to survive intensifying technology- venture partners."63 In addition, much of Korea's success based global competition, it must move beyond its lay in its firms' abilities to develop the knowledge and traditional "global factory" innovation model, which will require quick access to radical innovations, especially in 59 generic technologies. While Taiwan has significant This section heavily relies on Dieter Ernst, "Global Production Networks and the Changing Geography of Innovation Systems: policy initiatives in each of the above areas,58 the risk of Implications for developing Countries." East-West Center failure remains high, implying that an exclusive focus on Working Papers, No. 9, November 2000. Available at technology leadership strategies is unlikely to support a http://scholarspace.manoa.hawaii.edu/bitstream/handle/10125/607 broad-based upgrading through innovation strategy. 4/ECONwp009.pdf?sequence=1. Last accessed August 21, 2012. 60 These risks explain why Taiwan's new innovation For instance, Ernst, Ganiatsos, and Mytelka (eds.), 1998, strategy emphasizes low-cost and fast innovation Technological Capabilities and Export Success - Lessons from East Asia, Routledge, London and New York. 61 D. Ernst, 2000, "Catching-Up and Post-Crisis Industrial Up- 56 ECFA is a special free-trade agreement between Taiwan and grading. Searching for New Sources of Growth in Korea's Elec- China, which was concluded in September 2010. tronics Industry," in F. Deyo, R. Doner, and E Hershberg (eds.), 57 For instance, during the November 2011 Asia-Pacific Eco- Economic Governance and Flexible Production in East Asia, nomic Cooperation meeting in Honolulu, Taiwan supported U.S. Rowman and Littlefield Publishers. Taiwan provides another, proposals to extend the Information Technology Agreement and to albeit very different, approach to the development of network establish an Environmental Goods and Services Program. integration services through international linkages. 58 62 On SoC design, the government has initiated a National SoC Chaebol refers to South Korean business conglomerates that Research Program. On nanotechnology R&D, the government has are global multinationals owning numerous international enter- committed substantial funds, while ITRI and the National Science prises. 63 Council have signed an agreement to conduct joint research with D. Ernst, 2000, "Catching-Up and Post-Crisis Industrial Up- the National Research Council of Canada. And Sha et al. ("ITRI's grading. Searching for New Sources of Growth in Korea's Elec- Role in Developing the Access Network Industry in Taiwan" in H. tronics Industry," in F. Deyo, R. Doner, and E Hershberg (eds.), S. Rowen, M. G. Hancock, and W. F. Miller (eds.), 2008, Greater Economic Governance and Flexible Production in East Asia, China's Quest for Innovation, Shorenstein APARC, Stanford, CA) Rowman and Littlefield Publishers. Taiwan provides another, describe ITRI's role in the industry-level upgrading of Taiwan's albeit very different, approach to the development of network access network industry. integration services through international linkages.

OCR for page 31
44 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING skills necessary to monitor, unpackage, absorb, and access to funding is restricted to organizations based in upgrade foreign technology. Equally important was a the European Union. This restrictive approach to capacity to mobilize the substantial funds for paying international cooperation in science and technology is technology licensing fees and for importing best-practice further emphasized in the European Commission's (EC) production equipment and leading-edge components. 2010 policy document Innovation Union to "ensure that Most Korean producers arguably would have hesitated to leading academics, researchers and innovators reside and pursue such high-cost, high-risk strategies had they not work in Europe and to attract a sufficient number of been induced to do so by a variety of selective policy highly skilled third country nationals to stay in interventions by the Korean state. By providing critical Europe."66 externalities such as information, training, maintenance The 2012 FP7 Work Program is the EC's largest and other support services, and finance, the Korean funding package (about 7 billion) under the FP7 so far government has fostered the growth of firms large and will provide funding to EU-based universities, enough to overcome high entry barriers. research organizations, and industries, with special It is this coevolution of international and domestic attention given to small and medium enterprises. In knowledge linkages that explains Korea's extraordinary addition, it is expected to create around 174,000 jobs in success. It has enabled Korean firms to reverse the the short term and nearly 450,000 jobs and 80 billion sequence of technological capability formation. Rather growth in gross domestic product (GDP) over 15 years. than proceeding from innovation to investment to Since the initiation of the FP7 Program, investment in production, they focused on the ability to operate industrial R&D by the European Union's top 1,000 production facilities according to competitive cost and companies has grown by ~10 percent. Between 2010 and quality standards.64 2011, industrial R&D in the European Union grew by Through reverse engineering and other forms of almost 6 percent, compared with higher growth reported copying and imitating foreign technology, as well as for the United States (~10 percent), Taiwan (~18 integrating into the increasingly complex global percent), Korea (~21 percent), Hong Kong (~29 percent), production networks of American, Japanese, and some and China (~30 percent), and lower reported growth in European global flagship corporations, Korean firms Japan (-10 percent). These industrial R&D investments were able to avoid the huge cost burdens and risks can also be broken down by industry classification--of involved in R&D and in developing international specific interest to this study are the R&D contributions distribution channels. from the telecommunications, software, computer For Korea, international linkages provided an science, semiconductors, and electronics industry. These important initial catalyst for the development of a sectors make up ~23 percent of the European Union's67 sufficiently broad portfolio of domestic capabilities that and of Japan's industrial R&D investments, compared are needed to reap potential benefits of latecomer with ~35 percent for Hong Kong, ~39 percent for China, advantages. ~41 percent for the United States, ~41 percent for India,68 ~70 percent for Korea, ~77 percent for 3.5 Europe Integrated EU-wide Innovation Policy Singapore, and a staggering ~94 percent for Taiwan.69,70 Coordination In R&D intensity, however, the European Union continues to lag behind Japan and the United States. At 3.5.1 The Seventh Framework Program for Research and Technological Development (FP7) 66 EC, 2010, Europe 2020 Flagship Initiative Innovation Union. Communication from the Commission to the European Parlia- The 20072013 Seventh Framework Program (FP7) ment, the Council, the European Economic and Social Committee for research and technological development is the and the Committee of the Regions. Available at http://ec.europa.eu European Union's main instrument for funding research /research/innovation-union/pdf/innovation-union-communication_ en.pdf, page 27. Last accessed on January 7, 2012. in Europe.65 With a total budget of 53.2 billion, the FP7 67 Among the European Union member states, Germany, France, aims to increase Europe's growth, competitiveness, and and Finland made the largest R&D investments in the employment through initiatives and existing programs telecommunications, software, computer science, semiconductors, that finance grants to research actors all over Europe, and electronics industry sectors (16.1 percent, 22.4 percent, and usually through cofinancing research, technological 82.8 percent, respectively). 68 development, and demonstration projects. However, India's R&D investment sectors consist primarily software and computer services. 69 See http://iri.jrc.es/research/scoreboard_2011.htm. Last ac- 64 Ibid. cessed on January 7, 2012. 65 70 See http://ec.europa.eu/research/fp7/index_en.cfm. Last ac- See http://iri.jrc.ec.europa.eu/research/scoreboard_2008.htm. cessed on January 7, 2012. Last accessed on January 7, 2012.

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 45 1.6 percent, the European Union's 2010 share of R&D slow standardization and ineffective use of public expenditure in GDP trails both Japan and the United procurement; and (3) too much fragmentation and costly States by a considerable margin, with 3.3 percent and 2.7 duplication.76 percent shares, respectively.71 Among the member states, Germany dominates--at 2.5 percent, its share of R&D 3.5.3 The European Union's Key Enabling Technologies expenditures in GDP is much larger than the European (KET) program Union share. More importantly, however, Germany was deemed to have the highest propensity to innovate.72,73 An interesting attempt to operationalize Europe's Hence, it is important to emphasize that national integrated innovation strategy is the European Union's innovation policies differ quite substantially across Key Enabling Technologies (KET) Program.77 The EC's Europe, both in their overall strategic vision, and in their six KETs--nanotechnology, micro- and nanoelectronics, effectiveness. advanced materials, photonics, industrial biotechnology, and advanced manufacturing systems78--were selected 3.5.2. Toward an Integrated EU-wide Innovation based on their economic potential, their value-adding Strategy and enabling role, and their technology and capital intensity with R&D and initial investment costs. KETs Germany's move toward an integrated innovation are defined as "knowledge and capital-intensive strategy74 is emblematic for a growing trend within the technologies associated with high research and European Union to adopt a much more centralized development (R&D) intensity, rapid and integrated approach to innovation. In 2000 the European Union innovation cycles, high capital expenditure and highly- established the European Research Area (ERA) to skilled employment."79 KETs are also embedded in promote a "single innovation market." One of its main advanced products and they underpin innovation chains. objectives was to optimize and open European, national, Advanced computing products, such as multicore and regional research programs to support the best processors and parallel software developments, are research throughout Europe and to coordinate these examples of technologies that are consistent with the programs to address major challenges together.75 KET definition. Like other KETs, advanced computing In December 2008 the Competitiveness Council technologies provide potential first-mover advantages, adopted a 2020 ERA vision, which seeks to increase the and enable the owner of relevant intellectual property Europe-wide mobility of innovation capabilities by rights to create new lead markets as new technologies promoting the free circulation of researchers, knowledge, replace old technologies with few or no other players. and technology. In 2010 the EC developed an integrated One of the key goals of the European Union's KET innovation strategy entitled "Innovation Union" to tackle Program is to reduce the deeply ingrained barriers to three main challenges for EU innovation policy: (1) industrial innovation. In other words: Why are underinvestment in knowledge foundation (e.g., the breakthrough ideas, inventions, and discoveries (that United States and Japan are out-investing Europe and China is rapidly catching up); (2) unsatisfactory 76 framework conditions, ranging from poor access to EC, 2010, Europe 2020 Flagship Initiative Innovation Union. finance, high costs of intellectual property rights (IPR) to Communication from the Commission to the European Parlia- ment, the Council, the European Economic and Social Committee and the Committee of the Regions, Commission Communication 71 Battelle, 2010, "Global R&D Funding Forecast" in R&D (COM(2010)546). Available at http://ec.europa. Magazine, December 2009. Available at http://www.rdmag.com/ eu/research/innovation-union/pdf/innovation-union-communica- uploadedFiles/RD/Featured_Articles/2009/12/GFF2010_ads_small tion_en.pdf, page 27. Last accessed on January 7, 2012. 77 .pdf. Last accessed on August 11, 2012. EC, 2011, High-Level Expert Group on Key Enabling 72 An innovation, here, is defined as a new or significantly im- Technologies. Final Report, June. Available at http://ec.europa.eu/ proved product (good or service) introduced to the market or the enterprise/sectors/ict/files/kets/hlg_report_final_en.pdf. Last ac- introduction within an enterprise of a new or significantly im- cessed on August 15, 2012. 78 proved process. EC, 2009, Preparing for our future: Developing a common 73 Eurostat, 2010 Yearbook, p. 606. Available at strategy for key enabling technologies in the EU, Commission http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-CD-10- Communication (COM(2009)512). Available at http://ec.europa. 220/EN/KS-CD-10-220-EN.PDF. Last accessed on August 15, eu/enterprise/sectors/ict/files/communication_key_enabling_techn 2012. ologies_en.pdf. Last accessed on August 15, 2012. 74 79 Federal Ministry of Education and Research (BMBF) (2011) EC, 2010, Current situation of key enabling technologies in High-Tech Strategy. Available at http://www.hightech-strategie. Europe, Commission Staff Working Document (SEC(2009)1257). de/en/350.php. Last accessed on January 12, 2012. Available at http://ec.europa.eu/enterprise/sectors/ict/files/staff_ 75 See http://ec.europa.eu/research/era/index_en.htm. Last ac- working_document_sec512_key_enabling_technologies_en.pdf. cessed on January 12, 2012. Last accessed on August 15, 2012.

OCR for page 31
46 THE GLOBAL ECOSYSTEM IN ADVANCED COMPUTING were developed with public R&D funds) not transformed 3.6 Conclusions and Policy Implications into commercially successful innovations within reasonably short time frames? The diversity of economic and IT innovation policies across the United States, China, Taiwan, and 3.5.4 Policy Options Europe reflect their differing cultures and history, economic status and technical capabilities. The U.S. To overcome the above deeply entrenched approach rests on government support for basic innovation barriers, the European Union's KET Program academic research and a vibrant capital market and proposes a broad range of coordinated support policies private enterprise ecosystem for product innovation. The that cover the following stages of the "innovation chain," other countries and regions blend elements of private from the transformation of fundamental research into enterprise and central planning. Each is unique and not globally competitive technologies, through product directly transferrable to another region. Nevertheless, development to make innovative and cost-effective there are some general principles that can be gleaned product development and prototyping, to globally from this survey of policies, coupled with technical competitive manufacturing. insights regarding semiconductor device fabrication, Specifically, the EC KET Program identifies the chip architecture, and software. following five priority areas for Europe's evolving EU- Some of the largest computing companies in the wide innovation strategy: (1) sustain a critical mass in United States have internal multidimensional knowledge and funding through effective use of technological capabilities in chip design, process economies of scale and scope; (2) increase market focus development, wafer manufacturing, and software and of R&D projects; (3) invest in large-scale demonstrators have demonstrated success tapping into foreign talent and pilot test facilities; (4) provide post-R&D pools and markets. However, IT talent, capabilities, and commercialization support; and (5) practice trade facilities are increasingly distributed globally. Although diplomacy, that is, reduce unfair subsidies and protect research prowess is correlated with industry success, domestic companies from unfair trade practices.80 This information flows globally via many sources. The lesson last policy priority is of particular concern from a U.S. of basic research, both in industry and academia, has perspective. In fact, the European Union's KET Program been that the discoverers are not always those who culminates in a fairly "techno-nationalist" notion of IPR convert the ideas into economically successful products. protection and states that "the EU should clearly promote Oftentimes, the likelihood that an idea can be an `in Europe first' IP policy" and that proposals require successfully commercialized and implemented depends clear IP plans for "first exploitation of IP" and rules that on a nation's or region's innovation policies and "favour the EU exploitation of the results of projects."81 entrepreneurial climate. Second, the cost of semiconductor fabrication 3.5.5 Summary facilities is rising exponentially, placing their construction beyond the economic reach of small- and The European Union has experienced a fundamental mid-sized companies. Only the largest multinational change in its innovation policy from government- companies and nation-states can fund their construction centered national strategies to attempts to combine and operation. This suggests that the United States must market-led innovation and public policy coordination be mindful of its global dependence on fabrication across Europe. While government initiatives, such as the supply chains and that it develop realistic models that KET Program, attempted to bridge the perennial gaps balance the need for the latest process technology versus that stymie Europe's industrial innovation ecosystem, multiaxis innovation and that combine reliability and significant challenges remain. To a large degree, resilience, programmability, and functionality. Although however, this transformation is still a work in progress, financial investment in fabrication facilities by a small as European IT innovation and commercialization number of U.S. companies, primarily by Intel, provides continue to lag. some domestic sourcing, most of the chips contained in In addition, there are signs that Europe's fiscal crisis devices sold in the United States are fabricated offshore. and increasingly severe austerity policies might slow IBM does produce some chips in the United States, both down Europe's move towards greater openness and for U.S. defense needs and its own products, but the internationalization of its innovation system. volume is relatively small. Third, there is no assurance that historical U.S. dominance in computing will transfer to new and 80 EC, 2011 KET, p. 33. emerging domains. The need for architectural and 81 EC, 2011 KET, p. 37. software innovation to deliver new features and greater

OCR for page 31
INNOVATION POLICY LANDSCAPE COMPARATIVE ANALYSIS 47 performance via parallelism creates opportunities for local and regional benefit. They are investing in the new ecosystems to emerge and evolve. With licensable future, hoping to position their region for success. Which components and global access to fabrication facilities, it of the myriad approaches being pursued will be most is possible for this innovation to occur almost anywhere. successful is difficult to predict. In addition to performance as measured by computing Today is an inflection point, when the virtuous cycle speed (clock speed, bandwidth, interconnect, and so on), of faster sequential processors has broken down and it may be that other measures--such as reliability and when new devices and services are emerging to reshape resilience, programmability, security, and efficiency-- the computing landscape. An intense global competition become equally, or potentially more, important. For for IT hegemony is under way. No company, country, or example, efforts to improve programmability and region will reap all of the economic benefits, as the efficiency of base processors might yield significant global value chain is too intertwined for that. However, improvements in software quality, software development there will be economic winners and losers, just as there times, and (ultimately) application performance. always are whenever technology shifts occur. U.S. Fourth, global policy makers see information policy makers would be wise to think carefully and technology in general and consumer computing in deeply about the shifts under way and their implications particular as major economic forces to be harnessed for for economic competitiveness and national security.

OCR for page 31