Competing Programs: Government Support for Microelectronics
Thomas R. Howell*
Partner, Dewey Ballantine LLP, Washington, D.C.
EXECUTIVE SUMMARY
Government promotional policies have played an important role in the development of the semiconductor industry in every country in which such an industry has emerged.† In the United States, the proper relationship between government and industry has long been controversial, and the measures taken by the U.S. government in microelectronics have proven no exception. However, the collaborative government-industry-university programs that have been implemented in the U.S. semiconductor sector since the 1980s—of which the SEMATECH consortium is emblematic—are widely admired abroad and are being emulated on a large scale in Japan and the European Union.
At a moment when U.S. government support for the semiconductor industry is significantly reduced, the level and scope of government involvement in the industry outside the United States is increasing substantially, with various elements of the U.S. economic system and industry-government relationship frequently cited as models for foreign promotional programs. The divergence between U.S. and foreign government policies toward the semiconductor industry occurs at a time when the U.S. industry enjoys a position of undisputed world leadership but also as it confronts unprecedented technological challenges with “no known solutions” and no clear plan for mustering the resources needed to surmount those challenges.
Foreign government measures to support the semiconductor industry are larger in scale and broader in scope than anything currently under way in the United States. In Japan and the European Union, as in the United States, the principal form of government support for microelectronics is the provision of funding and infrastructure for industry-government research and development projects. However, the Japanese and European programs are funded at a much higher level and place a greater emphasis on R&D with immediate commercial applications, in contrast to U.S. programs, which are normally limited to pre-competitive R&D. In addition, in the European Union substantial funding is being provided by national and regional governments to individual companies for investment in semiconductor manufacturing facilities, and in both the European Union and Japan, direct government funding is being used to stimulate new “venture” businesses in the microelectronics field.
In the growing number of newly industrializing countries promoting an indigenous capability in microelectronics—Taiwan, Korea, Singapore, China, and Malaysia—government policies emphasize the acquisition and diffusion of advanced semiconductor technology from the industrialized countries rather than pursuit of leading-edge R&D. The principal forms of government support in these countries are direct provision of capital to domestic firms (including funding of small- and medium-size venture companies); funding of research institutes that assist in technology diffusion; technology acquisition and transfer to industry; tax holidays; programs to train personnel and attract scientists and engineers from other countries; establishment of industrial zones with incentives for firms locating in these zones; and outright creation of new semiconductor enterprises.
While it is difficult to predict the precise effects that the various government programs in this sector will have over the long run, government measures are clearly contributing to several of the most significant observable trends in this industry.
-
Japan is pursuing a national revival of its competitive position in microelectronics, driven by an array of large new government-sponsored R&D projects.
-
The European Union has reversed its declining position in the semiconductor sector and improved its relative competitive standing, a development sub
-
stantially attributed to European Union- and national government-supported R&D projects, most notably JESSI and MEDEA.
-
Both Japan and the European Union are now pursuing comprehensive strategies designed to challenge U.S. leadership in microelectronics by leveraging their present and anticipated advantages in mobile communications and digital home appliances.
-
Taiwan has emerged as a major production base for semiconductors and— reflecting a sustained government promotional effort and with initial capitalization from the government—has pioneered a business model, the dedicated foundry, that many believe will revolutionize the industry.
-
China is emerging as a potentially significant competitor, reflecting its government’s efforts to attract inward foreign investment and foreign technology and measures to promote indigenous producers. The dramatic ongoing movement of Taiwanese information technology manufacturing functions to Mainland China—encouraged for differing reasons by governments on both sides of the Fujian Straits—will almost certainly accelerate the maturation of the Chinese semiconductor industry.
U.S. economic thinking places a high value on independent entrepreneurship and emphasizes the need to circumscribe carefully the government’s role in the market. While U.S. federal, state, and local governments have frequently worked in close collaboration with the private sector in a diverse array of industries— agriculture, aerospace, biotechnology, and others—government intervention in specific sectors or on behalf of individual enterprises is controversial and therefore usually limited.
The United States, however, has generally not proven successful in dissuading other governments from intervening heavily on behalf of strategic sectors like semiconductors, where governments often have the explicit objective of promoting the commercial success of individual indigenous firms. Reflecting that fact, since the early 1980s the U.S. government has been drawn into a series of limited market interventions to counter the adverse effects of foreign government measures on the U.S. semiconductor industry, most notably the Semiconductor Trade Agreement and federal funding of the SEMATECH consortium. These and other, similar measures were improvisations devised by the U.S. government and industry working together in response to challenges that arose out of foreign industrial policies. These measures, while sometimes the subject of criticism, did not represent a fundamental departure from U.S. economic values and succeeded in addressing the problems that made the measures necessary in the first place.
Given the challenges confronting the U.S. industry in the coming decade, which reflect both difficult technological obstacles and dramatic increases in foreign government support measures, it is likely that additional industry-government improvisations—and continued cooperation—will be required if the U.S. industry is to sustain its current position of leadership.
INTRODUCTION
This paper surveys government policy measures that are influencing the international competitive environment in the semiconductor industry. The role that governments should play in this industry has always been controversial, and the impact that current programs will ultimately have is difficult to predict with much precision. Looking back it is clear that government policies have played a major role in the evolution of the global semiconductor industry. Indeed, to date no country in the world, including the United States, has established and sustained a world-class semiconductor industry in the absence of a very substantial government promotional effort.1
Many of the most significant developments of the past 15 years in this industry are substantially attributable to government policy actions; these include the rebound of the U.S. semiconductor industry from the crisis of the mid-1980s; the erosion of Japanese leadership by new competitors arising in Taiwan and Korea; the resurrection of the European position in microelectronics; and the advent of China as a potentially significant competitor. Government programs to support the semiconductor industry have shaped the competitive environment and will continue to do so for the foreseeable future.
The most striking aspect of the current pattern of government support for microelectronics worldwide is the level at which such support is declining in the United States while increasing substantially in the key semiconductor-producing regions outside the United States. Ironically, at the moment the United States appears to be curtailing direct government support for this industry, promotional programs abroad are attempting to replicate what is seen as a highly successful example of government-industry collaboration in the United States. Some argue that U.S. industry leads the world today precisely because market-based competition has shaped its evolution to a far greater degree than has been the case anywhere else, and that the steep decline in U.S. government financial support and other forms of involvement is therefore a positive trend.
But, U.S. leadership in microelectronics reflects the success of a system of innovation based not only on the activities of private firms but on government institutions, universities, and research associations and consortia in which the government plays a role. Significantly, during the next 10 years this system will confront fundamental technological challenges that simply have no precedent and
will require a level of human, financial, and infrastructural resources that the industry, standing by itself, will be hard pressed to achieve.
At present the U.S. semiconductor industry stands at the pinnacle of success. It holds by far the largest share of the world market and is the undisputed world leader in many key areas of semiconductor technology.2 It has fought off a major competitive challenge from Japan and has seen its Japanese rivals’ market share progressively decline for a decade. Its productivity has consistently grown at a rate that far outstrips most other sectors of the U.S. economy.3 Its methods and business culture are widely studied in academia and increasingly emulated abroad. It continues to attract extraordinarily talented people from every part of the world.4 Its revolutionary contributions to the economy and to the society as a whole are universally acknowledged and acclaimed.
Challenges Facing the U.S. Semiconductor Industry
The present sanguine state of the U.S. semiconductor industry masks trends that could jeopardize the U.S. position in the coming decade. These challenges are both structural and technological in nature.
Structural and Technical Challenges
Despite continued high rates of growth in long-term demand the industry remains sharply cyclical—a dynamic that results in erratic levels of funding for R&D. The industry faces a growing shortage of trained scientists and engineers, reflecting such trends as a decline in U.S. graduate electronic engineering degrees and an increase in the number of foreign students who return home after graduat
ing from U.S. universities. The capital investments required for semiconductor manufacturing have become so large that a significant and growing proportion of U.S. production is outsourced to “foundries,” by far the most advanced of which are operated by foreign companies based in East Asia.
The U.S. industry also confronts a formidable array of technological hurdles as it pushes miniaturization to the molecular and atomic level—in the next 5 years it will encounter problems “for which there are no known solutions.”5 These physical limits may herald the end of further miniaturization using known silicon-based technology and require a radical leap to some other form of technology to achieve further advances; but research and development on the scale necessary to develop such a replacement technology is not taking place.
Competition for New Markets: PCs to Wireless
Finally, the U.S. industry faces competition from abroad, a factor that has receded as a perceived challenge but which now should be receiving heightened attention. The U.S. industry’s present dominant position is based in significant part on its success in developing products for PCs and PC equipment. European and Japanese industry leaders believe that they can dominate what they see as the main semiconductor growth markets of twenty-first century—wireless and wired telecommunications and digital home appliances.6 The Japanese semiconductor industry, with an unprecedented level of government backing, is embarking on an intensive series of leading-edge R&D projects with the objective of recapturing world leadership in microelectronics from the United States, based on improved capability with respect to systems-on-a-chip. A Japanese semiconductor analyst recently commented on this effort as follows:
The 21st century will be the end of the PC era, and the arena of competition will change. Instead of the PC, cell phones and digital consumer equipment, two areas in which Japan is dominant, will pull semiconductor technology forward. A favorable wind is blowing for Japan, which was defeated in the PC era.7
Technological Parity?
Similarly, the European industry is building on European strengths in key, rapidly growing end-use markets—most notably telecommunications—to carve
out a significant market segment based on the design and production of specialized devices for these markets, which are growing rapidly as a proportion of the total end-market for semiconductors. The principal European and Japanese R&D programs are major, sustained, long-term efforts of five or more years’ duration that are often succeeded by follow-on efforts of comparable or greater length.8
Industries in Taiwan, Korea, and Singapore are seeking to achieve and sustain technological near parity with U.S. producers by acquiring technology through foundry relationships and other collaborative arrangements. These arrangements involve the cession by U.S. firms of key segments of the manufacturing process that they once performed entirely by themselves. China, while not a major factor today, is aggressively pursuing acquisition of foreign technology and know-how, and in the view of many will emerge as a major U.S. competitor within 10 years. Increasingly, all of these countries are competing with the United States for the same limited pool of human resources and, to the extent that this rivalry remains a zero-sum exercise, it will inevitably intensify.
These developments abroad are not necessarily negatives for the U.S. industry, and in some cases they will actually enhance competitive opportunities for U.S. firms. The emergence of government-supported semiconductor foundries in Taiwan and Singapore, for example, may well work to the competitive advantage of U.S. producers, and U.S. firms have been beneficiaries of foreign government subsidy programs.9 But, in assessing the prospect that the U.S. semiconductor industry will surmount the challenges it will confront in the new century, the government-driven competitive strategies that are being implemented abroad and their potential effects on the U.S. industry must be taken into account. The intensity and scale of these national efforts are changing rapidly, generally in the direction of a larger government commitment.
The first segment of this report surveys U.S. government programs in microelectronics. The remainder examines programs outside the United States, reviewing national strategies in Asia and Europe.
GOVERNMENT AND INDUSTRY IN THE UNITED STATES
The success of the U.S. semiconductor industry—with its now legendary tradition of dynamic and colorful entrepreneurial initiative, ferocious competition, and dramatic technological breakthroughs—is sometimes held up as the very embodiment of the virtues of the market-driven U.S. economic system. The reality, however, is more nuanced. U.S. government policies have played an important role in the evolution and survival of the semiconductor industry, and the industry’s current success is at least partially attributable to a close cooperative working relationship that has grown up between the industry, universities, and the U.S. government.
As a result foreign countries seeking to create their own indigenous versions of the U.S. semiconductor industry are trying to replicate not only the best features of Silicon Valley and the U.S. venture capital system but also the U.S. industry-university-government research triad. The chairman of the Nippon Electric Company, one of Japan’s largest semiconductor makers, has commented on how the U.S. industry recaptured world leadership from Japan after the mid-1980s—in a manner that he now urges Japan to emulate:
In the U.S. there have been various [government] measures that recognize the importance of the semiconductor industry, which is the basis of defense and all industries; a semiconductor revival resulting from those measures; the activation of private-sector investments, along with that; and ideal cooperation between industry and universities that supports the semiconductor industry...[T]he U.S. activated cooperative semiconductor-related efforts such as SEMATECH that involved industry, government and universities from the viewpoint of the importance of semiconductors as a key technology for all industries. In order to win in the 21st century, we must, at all cost, rework the strategy we have lost.10
The Legacy of Government Support
The U.S. semiconductor industry was given its initial impetus from U.S. government funding of research and development and procurement for military and space exploration programs.11 While the creation of a thriving commercial semiconductor industry was a byproduct rather than an objective of these early government programs, the U.S. defense community grew to recognize the in
creasingly central importance of the commercial industry to national defense, and in subsequent decades took major steps to preserve and enhance the competitiveness of that industry.
The most significant initiative was the Defense Department’s sponsorship of SEMATECH, a research and development consortium established to ensure U.S. leadership in semiconductor-manufacturing technology. This initiative and many others in the microelectronics field benefited from support from the Defense Advanced Research Projects Agency, a small Defense Department agency that supports long-term R&D with commercial as well as military value.12
During the 1980s the U.S. Congress enacted legislation extending copyright protection to semiconductor designs, relaxing antitrust rules for joint research, and providing tax credits for research and development, all of which substantially benefited the U.S. industry. In 1986 the U.S. government negotiated a bilateral agreement with Japan, the U.S.-Japan Semiconductor Arrangement, with the objective of ending Japanese dumping and improving market access for U.S. semiconductor producers in Japan.13
In addition to these individual government actions a system of ongoing collaboration between the industry, the government, and U.S. universities has evolved with respect to long-term research and development. In 1982 the Semiconductor Research Corporation was formed by U.S. semiconductor device firms to undertake silicon-based R&D in U.S. universities.14 While U.S. companies contributed most of the Semiconductor Research Corporation’s funding, four U.S. government organizations participate in and contribute to its funding, and a number of its key leaders have backgrounds in Department of Defense microelectronics R&D programs.15
Beginning in 1992 the Semiconductor Industry Association, representing U.S. device manufacturers, brought together representatives of the private sector, the government, and U.S. universities to develop a National Technology Roadmap for Semiconductors, a description of and timetable for achieving technological targets necessary to ensure continued advances in the performance of integrated circuits.16 SEMATECH, while no longer federally funded, continues as an active participant in joint industry-government activities such as the ongoing development of the roadmap and R&D partnership with the National Laboratories.
Support: Past, Present, and Future
This multifaceted collaborative industry-government-university R&D effort is widely credited abroad with playing a major role in reversing the relative competitive decline of the U.S. industry in the late 1980s and its return to world leadership in the 1990s and into the present.17
Yet the U.S. government’s involvement in microelectronics has remained a subject of controversy. The end of the Soviet military threat raised questions about the continuing need for defense spending in this field, at least on the same scale as during the Cold War.18 The Defense Advanced Research Projects Agency’s annual funding of microelectronics R&D—the principal channel of direct federal financial support—is expected to decline.19 In addition, critics, including some executives in the U.S. semiconductor industry, have blasted federal
16 |
See Semiconductor Industry Association Web site at <http://www.semichips.org>. |
17 |
“A major factor contributing to the U.S. semiconductor industry’s recovery from this perilous situation [in the 1980s] was a U.S. national policy based around cooperation between industry, government, and academia.” See Hajime Susaki, chairman of NEC Corporation, in FBIS, January 12, 2001, translation of “Japanese Semiconductor Industry’s Competitiveness: LSI Industry in Jeopardy,” Nikkei Microdevices, December 2000, pp. 245-48 (JPP20010112000011). The U.S. industry undertook a comprehensive effort to “incorporate scientific methods into semiconductor production management technology, on-site maintenance, and the like.... I would like to point out that the earnest national effort made by the U.S., in particular, was a very important motive force behind the American revitalization.” See FBIS, August 1, 2001, translation of Michio Mizogami, “Prescription for Japan’s Revival,” Nikkei Microdevices, August 2000, pp. 207-09 (JPP20000817000061). National Research Council, Conflict and Cooperation reviews the SEMATECH program (p. 141) and provides an early review of the then emerging Japanese and European programs (pp. 19-24). For a review from the perspective of the U.S. industry, see Andrew A. Procassini, Competitors in Alliance: Industrial Associations, Global Rivalries, and Business-Government Relations. Westport, Conn.: Quorum Books, 1995. |
18 |
Conrad W. Holton, “Federal Funds for Chip Research Dwindle Under Congressional Pressure,” Solid State Technology (July 1996), p. 86. |
19 |
Scott Nance, “Broad Federal Research Required to Keep Semiconductors on Track,” New Technology Week, October 30, 2000. Sonny Maynard, Semiconductor Research Corporation, cited in presentation by Dr. Michael Polcari, “Current Challenges; A U.S. and Global Perspective,” National Research Council, Symposium on National Programs to Support the Semiconductor Industry, October 2000. |
programs in microelectronics as corporate welfare.20 Thus, even though federal funding for SEMATECH ended in May 1997, debate has continued within the Congress and the Executive branch as to whether and to what extent the U.S. government should continue to invest federal funds in supporting R&D in microelectronics.21 Many observers argue that if anything, the role of the government should be further curtailed. This trend, if it continues, will run directly counter to those in Europe and East Asia, where governments are dramatically escalating their levels of direct and indirect funding in this sector.
Challenges Facing the U.S. Semiconductor Industry
Confronting the “Brick Wall”
For three decades advances in semiconductor technology have followed Moore’s Law, which predicts that transistor performance and density will double at a predictable and relatively constant rate (Moore originally postulated a doubling every three years, but the rate of doubling has been adjusted over time). This dynamic, which has resulted in a relentless driving down of the cost of electronic functions and concurrently an exponential increase in performance, underlies the information revolution that is credited by many analysts with the U.S. economic resurgence of the 1990s. As the economist Robert Gordon has noted,
[The] Clinton economic boom is largely a reflection of Moore’s Law. [T]he recent acceleration in productivity is at least half due to the improvements in computer technology.22
For many years it has been recognized that there are limits to the miniaturization of semiconductor components; at some point optical lithography—the predominant mode of semiconductor manufacturing—will no longer be workable. Although scientists have repeatedly succeeded in pushing optical lithography beyond what were once seen as its absolute limits, a more fundamental array of technological barriers may portend the end of Moore’s Law. In September 1999
Paul A. Packan, a highly regarded Intel researcher, warned that miniaturization of electronic components had been taken to such extremes that arcane physical effects at the molecular and atomic level now loom as a major obstacle to further miniaturization.23 Moore’s Law, he said,
seems to be in serious danger. Fundamental thermodynamic limits are being reached in critical areas, and unless new, innovative solutions are found, the current rate of improvement cannot be maintained.24
“No Known Solutions”
Packan pointed out that “solutions for these problems have not been found,” and the 1999 International Technology Roadmap for Semiconductors concurred, indicating that industry will face technical challenges in the next five years “for which there are no known solutions.”25 These multiple technological obstacles have come to be known collectively as “the brick wall.” U.S. companies have developed techniques that may enable them to squeeze one or two more generations of miniaturization from refinements using conventional methods, but this will merely defer the confrontation with the brick wall.26
Some U.S. scientists are experimenting with non-conventional solutions to some of the problems identified by Packan and the roadmap, such as the use of non-silicon and biological materials, but the scale of these efforts is grossly inadequate for the task.27 As Gordon Moore has expressed it, the U.S. industry is currently living off the benefits of investments in basic microelectronics R&D that were made in prior decades.28
Shortfalls in Research
Because the nature of the technological solutions to the brick wall is not known, the financial, human and infrastructural resources needed to surmount it cannot be quantified with precision, although it is generally agreed they are substantial. A 1995 working group, established under Semiconductor Research Corporation auspices, examined this issue based on circumstances prevailing at the time. It identified an annual shortfall of $492 million in basic microelectronics R&D. It concluded that while $153 million of this shortfall consisted of topics suitable for university R&D, U.S. universities had the capacity to perform only $48 million of this total.29
In a 1999 “back of the envelope” analysis the Defense Advanced Research Projects Agency concluded that the basic research funding gap had widened to an estimated annual shortfall of $1.2 billion. The growth in the research gap between 1994 and 1999 reflected not only an increase in research funding needs from the $750 million range to about $1.4 billion but also a drop in the level of funded basic research from $270 million in 1994 to $155 million in 1999.30
Shortfalls in Human Resources
The looming shortfall in human resources is probably even more serious than the research-spending gap. U.S. universities are graduating progressively fewer students with degrees in electrical and chemical engineering and in the physical sciences, notwithstanding the immediate demands of the private sector and the need to find solutions to the technological brick wall that will require a substantial increase in the number of trained people committed to R&D. To compound
the problem the proportion of foreign students in these fields at U.S. universities has increased dramatically, and a growing percentage of these graduates are returning home rather than remaining in the United States. Japan now produces 75 percent more engineers than the United States, and China produces over twice as many.31
Existing U.S. Industry and Federal Efforts
The magnitude of the technological challenge and the looming resource short-falls confronting the United States in microelectronics should not obscure the fact that the private sector and the U.S. government continue to devote very substantial resources to microelectronics R&D, and that a number of successful or promising industry-government R&D efforts are under way. These programs and perhaps others like them could provide a foundation for a national effort to surmount the brick wall.
Industry R&D
The U.S. semiconductor industry devotes a comparatively large portion of its revenues to research, spending about $9 billion annually on research and development. The industry’s annual R&D expenditures account for 10 to 15 percent of sales revenues—a rate of spending that is substantially higher than that of such technology-intensive sectors as aerospace, computers, telecommunications, and precision instruments.32
The vast preponderance of total private-sector R&D expenditure is directed toward the development of commercial products or specific objectives of immediate concern to the companies involved.33 A few of the biggest companies (IBM,
Intel, AT&T) historically conducted a significant amount of long-term basic R&D in microelectronics, but as competition has intensified “and time to market has become a more important determination of corporate success, even these companies have been forced to channel their research more closely to areas of strategic importance.”34
While the U.S. industry’s applications-oriented pattern of investment has been criticized as myopic, it reflects powerful, widely recognized economic and commercial imperatives:
Long-term research is inherently uncertain, and the payback can be decades in the future. Furthermore, basic research can result in revolutionary technologies that undermine the existing product areas upon which the established firms rely on for their success. Economists have also described the so-called appropriability problem: because of the broad applicability of the results of fundamental research, it is difficult—if not impossible—for a company to capture all the economic returns from its research investments—and to prevent their competitors from doing so. Hence, individual firms tend to invest less in fundamental research than would be optimal for society.35
The Focus Center Research Program
Reflecting the growing belief that a greater research effort is required, the U.S. semiconductor industry in 1997 launched the Focus Center Research Program. The Microelectronics Advanced Research Corporation’s (MARCO) Focus Center Research Program is a wholly owned and separately managed subsidiary of the Semiconductor Research Corporation, a not-for-profit research management organization that funds and operates university-based research centers in microelectronics.36
34 |
An IBM executive, Kathleen Kingscott, commented on this type of basic R&D in 2000: “They’re not going to do that anymore. IBM used to do it; AT&T used to do it; but we can’t do that anymore. In this capital-intensive industry companies are very hard pressed to find the resources to move from one generation to the next. It’s particularly true in the smaller semiconductor supplier companies.” Cited in Nance, op. cit. Erich Bloch and Jerry Sheehan, “Federal IT Funding: Creating an Infrastructure, Growing an Infrastructure, Growing an Industry,” IMP Magazine, March 1999, http://www.cisp.org/imp/march_99/03__99bloch)sheehan.htm. |
35 |
Ibid. |
36 |
About MARCO, MARCO Web site at <http://marco.fcrp.org> (accessed January 26, 2001). MARCO has its own management personnel but uses the infrastructure and resources of the Semiconductor Research Corporation. |
MARCO has been charged with the establishment of the Focus Center Research Center, which is designed to support pre-competitive, cooperative, long-range microelectronics R&D at U.S. universities. Each focus center consists of a team of U.S. universities tasked with conducting exploratory long-range (eight or more years) R&D on silicon-based integrated circuits in order to “address gaps and barriers anticipated in the development of certain technologies” outlined in the roadmap (ITRS).37 The Focus Center Research Program is funded jointly by Semiconductor Industry Association member companies (50 percent), by suppliers (25 percent), and by the U.S. government (25 percent), with a 6-year budget of $300 million.38
The National Laboratories
The U.S. government administers a number of facilities with excellent microelectronics R&D capability. Most notable among these facilities are the national laboratories, which are supervised by the Department of Energy.39 U.S. and foreign companies may enter into cooperative research and development agreements (CRADA) with U.S. government research organizations, including the national laboratories—in effect, drawing on the infrastructure and expertise of those institutions to overcome specific technological hurdles with commercial implications.40
A dramatic example of the CRADA programs’ potential for driving technological advances in microelectronics is a current ongoing effort to develop ex
37 |
Focus Center Research Program, MARCO Web site at <http://marco.fcrp.org>. |
38 |
FCRP Strategic Plan, MARCO Web site at <http://marco.fcrp.org>. |
39 |
The U.S. National Laboratories were originally established under the Manhattan Project and the Atomic Energy Commission. They are owned by the U.S. government and are managed by the Department of Energy (DOE). They perform R&D for DOE programs and R&D work for other federal agencies and the private sector on a cost-reimbursable basis. Under the Federal Technology Transfer Act of 1986 laboratory directors can enter into cooperative R&D agreements with private companies and negotiate licensing agreements and contracts. See generally Manufacturing Studies Board and National Materials Advisory Board, Commission on Engineering and Technical Systems, National Research Council, The Semiconductor Industry and the National Laboratories: Part of a National Strategy, Washington, D.C.: National Academy Press, 1987, pp. 10-11. |
40 |
Under CRADA agreements companies and government laboratories pool resources and efforts on a particular technological problem. The government may provide personnel, equipment, and laboratory facilities. The private-sector partners typically contribute funds and personnel. The government usually holds the patents on technologies developed under these agreements, but it gives its private-sector partners the exclusive license to market the technology, retaining the right to buy any product developed pursuant to the CRADA. Measuring CRADA outcomes is not self-evident, as David Mowery documents in Using Cooperative Research and Development Agreements as S&T Indicators: What Do We Have and What Would We Like? Presentation before National Science Foundation conference, Workshop on Strategic Research Partnerships, October 13, 2000, publication of proceedings pending. |
treme-ultraviolet lithography (EUVL) technology for manufacturing semiconductor devices with line widths as small as 20 nanometers (below that size silicon transistors cease to function normally).41 The EUVL project has already won high praise for overcoming “technical hurdles that once looked insurmountable.”42 The apparent success of this effort, however, will make it possible to squeeze a higher level of miniaturization out of existing silicon-based technologies, not to make the radical leap to an alternative system that would surmount the brick wall.
The National Nanotechnology Initiative
Nanotechnology is the science and engineering of assembling materials and components atom by atom, or molecule by molecule, to create large structures with fundamentally new molecular organization.43 It is highly likely that the technological solutions to the brick wall, when they are found, will reflect advances in nanotechnology.
In January 2000 the Clinton administration announced a National Nanotechnology Initiative pursuant to which 10 federal agencies would sponsor research and development into a broad range of nanotechnology themes. In FY 2000 $270 million in federal funds were allocated to this effort, and President Clinton’s proposed 2001 budget raised this figure to $497 million. The largest block of federal money was allocated to the National Science Foundation ($97 million in 2000), with significant funding also going to the Department of Defense ($70 million) and Department of Energy ($58 million). Approximately 70 percent of the total funding will be directed toward university-based research.
While a significant part of this project will involve either the development of nanoelectronic devices or research regarding materials and processing technologies with microelectronics applications, its scope is far broader and its resources must be spread across fields as diverse as medicine, space exploration, agriculture, environmental protection, and transportation networks.44
The Advanced Technology Program
The Advanced Technology Program was created by an act of Congress in 1988 and is administered by the National Institute of Standards and Technology, which has a large, technically skilled staff.45 Through this program the institute supports industry R&D efforts through provision of startup funding; equipment, facilities, and personnel; organizational and technical support; cost sharing for periods up to five years; and in some cases government participation in joint ventures.46 Financial assistance is provided in relatively small amounts (e.g., up to $2 million per project and normally up to $5 million for joint ventures).
Many U.S. industries, including the semiconductor industry, have benefited from this program.47 The program has been a lightning rod for criticism by opponents of government support for industry, and a number of powerful members of Congress have periodically threatened to shut the program down.48 Nevertheless, the program grew substantially under the first Bush administration. Its subsequent embrace and rapid expansion by the Clinton administration, and concomi
44 |
The National Nanotechnology Initiative implementation plan envisions that nanotechnology “will foster a revolution in information technology hardware rivaling the microelectronics revolution begun about 30 years ago” (ibid., p. 51). The initiative foresees, for example, research that will develop new approaches to nanostructure synthesis to permit affordable fabrication of electronic nanodevices for commercial use (the current estimated cost of one fabrication plant for 70-nm microelectronics is estimated at over $10 billion), ibid., p. 52. |
45 |
See generally National Research Council, Charles W. Wessner (ed.), The Advanced Technology Program: Challenges and Opportunities, Washington, D.C.: National Academy Press, 1999. For a comprehensive look at the program also see National Research Council, Charles W. Wessner (ed.) The Advanced Technology Program: Assessing Outcomes, Washington, D.C.: National Academy Press, 2001. |
46 |
Funding for the program peaked at $341 million in 1995 and has remained at $200 million or less in subsequent years. The program has achieved successes in areas such as medical devices. |
47 |
A project involving the Diamond Semiconductor Group developed technology for enabling faster processing of large semiconductor wafers with better process control; another project involving free research developed process improvements for growing large, silicon-carbide crystals, a semiconductor material used in making optoelectronics devices such as blue light-emitting diodes. National Institute of Standards and Technology, Building Bigger and Better Semiconductor Wafers, February 1999, at <http://www.nist.gov/public_affairs/factsheet/diamond.htm>. |
48 |
“Stark Warning Issued on Advanced Technology Program,” Bulletin of Science Policy News, March 21, 1997. |
tant congressional opposition, resulted in the curtailment of its budget rather than its elimination.49
Programs to support technology development are frequently misunderstood and not uniformly accepted, reflecting a strong bias for market-based solutions and perhaps a lack of familiarity with past U.S. practice. Although small in size, the Advanced Technology Program may be uniquely controversial partly because of its broad mission (to support enabling technologies with economy-wide benefits) and partly because it became involved in the political machinations of the mid-1990s. But, SEMATECH was also controversial, at both its inception and its renewal; opponents initially argued that it would not work and later argued that it had worked too well. The longstanding U.S. debate about the principle of government support for industry, rather than its efficacy, has few if any parallels overseas.
FORMS OF GOVERNMENT SUPPORT OUTSIDE THE UNITED STATES
As a general proposition government interventions in microelectronics are far larger in scale and broader in scope outside the United States. They are also more plainly directed at specific commercial objectives, yet they generate considerably less domestic controversy than do the U.S. programs, which are limited virtually entirely to pre-competitive R&D. Outside the United States, governments also fund pre-competitive research, particularly in Japan and Europe, but they are also providing large-scale financial support for R&D with direct and immediate commercial application. The table at the end of this paper provides a partial summary of significant government measures supporting microelectronics R&D outside the United States.50 In addition, above and beyond assistance for basic and applied R&D, foreign governments are providing capital for the construction of wafer-fabrication facilities and in some countries, such as Taiwan and Singapore, they have taken direct equity positions in semiconductor enterprises.
Governments are also utilizing industrial policy measures to replicate, to the extent possible in a non-U.S. environment, what are seen as the strongest features of the U.S. system. Thus, governments are countering the strong attraction that
the United States has traditionally exerted for foreign engineers and scientists by establishing incentives to draw such individuals into their own microelectronics programs. In Taiwan, Japan, Korea, and the European Union, governments and regional authorities are attempting to create new Silicon Valleys, utilizing various incentives to encourage the clustering of high-technology firms in designated zones. Some are also attempting to create parallels to the U.S. venture capital system by pumping government funds into new “venture” enterprises.
Policies Related to Research and Development
In Japan and Europe, by far the most important forum of government support for the microelectronics sector is in the area of research and development. Government funds are being channeled into a number of large industry-government R&D consortia with both pre-competitive and commercial themes. Typically these projects are sustained, long-term efforts involving the principal firms in each country, with very substantial government funding levels.51 In both Europe and Japan, the main government-sponsored programs are designed to enhance national advantage with respect to what are seen as the fastest potential-growth end-markets, such as wireless and wired telecommunications and digital home appliances. The U.S. industry, which dominates PC-based end markets, is seen as vulnerable as non-PC end markets grow as a proportion of the total market for semiconductors.
Some of the large-scale R&D projects being undertaken in advanced countries outside the United States offer a vehicle for reducing some of the technological burdens confronting the U.S. semiconductor industry. U.S. firms participate in and benefit from some foreign government-supported R&D programs. IMEC, a highly regarded, partially government-funded European microelectronics R&D center, performs leading-edge R&D on a contract basis for non-European firms, including Motorola, Texas Instruments, and AMD.52 Most major U.S. semiconductor producers have R&D centers abroad and/or R&D alliances and
technology exchange arrangements with foreign firms. All these are channels through which they may obtain technology originally developed in foreign government-supported programs.
The historical pattern of transnational cooperative R&D in this strategic industry is, however, uneven and is likely to remain so, reflecting abiding economic nationalism and national security concerns.53 An official from MEDEA,54 Europe’s most significant microelectronics R&D program, explained frankly why participation in that project was limited to European-based firms: “[A]fter all, we have no desire to fund technology transfer to other regions.”55 While Japan has permitted some foreign participation in its microelectronics R&D projects, the issue remains controversial, and foreign involvement is likely to remain closely regulated and monitored.56
Taiwan, Korea, and Singapore are not seeking to emulate the large-scale R&D consortia found in the most advanced countries. Instead they are pursuing R&D strategies characterized by
-
Rapid acquisition of foreign leading-edge technology;
-
Efficient diffusion within the national industry; and
-
Efforts to foster the ability to manufacture leading-edge products.57
Governments adapting this strategy pursue policies to attract inward investment and joint ventures involving advanced foreign semiconductor firms. They
-
Buy foreign technology outright and establish listening posts in advanced countries;
-
Establish diffusion-enhancing research institutes, consortia, and associations; and
-
Offer incentives for repatriation of their nationals working for semiconductor firms abroad.58
Taiwan has been so successful in adapting technology that a TSMC official estimated in 1999 that his firm and Intel were equals in terms of manufacturing technology. “[N]ow our company and Intel are both at the 0.18-micron level, so there is no technological gap.”59
Policies to Meet Burgeoning Capital Requirements
Since the beginning of the 1990s the sheer size of the investments required to establish state-of-the-art semiconductor manufacturing facilities has become an obstacle to growth. With the cost of a single fab rising above $1 billion and with markets volatile, the risks have become too great for many private companies to bear. In response, in the European Union national and regional governments have committed substantial investment resources to ensure the retention of at least some state-of-the-art manufacturing capacity within the European Union.60 In Korea the government has traditionally channeled capital to the semiconductor industry through its influence over the banking system. While the implementation of International Monetary Fund-backed reforms has ostensibly ended this practice, there is substantial evidence that the industry in Korea continues to receive capital assistance from the government through a variety of channels.61 Still, the most dramatic and comprehensive government support for capital in
vestment in microelectronics is now found in two significant emerging producing countries, Taiwan and Singapore.
Government Assumption of Capital Costs
Taiwan and Singapore have exploited the capital investment dilemma confronting the global semiconductor industry, in effect, by assuming the costs and risks of the high-investment levels themselves. In mid-2000 Taiwanese planners envisioned that Taiwanese firms would build a total of 21 new 300-mm fabs and 9 new 200-mm fabs by the year 2010. A Japanese survey taken in mid-2000 of known planned 300-mm wafer-fabrication facilities around the world concluded that half of the total facilities either were in Taiwan or were being wholly or partially financed by Taiwanese firms in other countries.62 The government of Singapore has publicly set a goal of 20 fabs by the year 2005.63 A substantial proportion of the fabs that are planned or being built will be dedicated to foundry operations—the manufacturing of foreign firms’ designs on a contract basis.
Taiwanese Policy
The aggressive capital investment programs under way in Taiwan supports its acquisition-and-diffusion technology strategy and partially compensates for the relative absence of larger-scaled R&D consortia. The performance of foundry manufacturing services creates a close, interdependent relationship between the foundry and the designing firm, and has enabled Taiwanese firms in particular to surge to the forefront with respect to semiconductor manufacturing technology.
Taiwan’s TSMC and UMC are considered the best foundry operations in the business. They have learned and perfected semiconductor-manufacturing techniques through the operation of foundries, emphasizing flexibility, speed, high yields, and low cost so successfully that they have eclipsed even the Japanese industry in terms of manufacturing skill. In a new joint venture between Taiwan’s
UMC and Hitachi, UMC is transferring Taiwanese manufacturing know-how to Hitachi—a startling reversal of roles.64
Tax Holidays
Taiwan’s planners estimate that the planned 30 new fabs will cost approximately $60 billion. This huge sum will not be financed directly by the government in whole or in part but by Taiwanese companies, which are expected to raise the money through a combination of retained earnings (50 percent) and funds raised on the private equity markets (50 percent). Even so, they are in a position to make these investments as a direct consequence of government policy. Because semiconductors are considered a strategic high-technology industry, the government provides substantial tax holidays to the semiconductor industry. For example, Taiwanese producers usually pay no income tax. TSMC has accumulated so many tax credits that in every recent year its after-tax earnings have actually been higher than its pre-tax earnings.65 Raising capital in the equity markets has been further facilitated by the lack of a Taiwanese capital gains tax and the government’s identification of the semiconductor industry as the number-one priority industry.66
The Foundry Phenomenon
The advent of the foundry model, which may revolutionize the semiconductor business, is a direct consequence of decisions made by Taiwanese government planners in the late 1980s. UMC and TSMC, the two Taiwanese firms that lead the world in foundry manufacturing, are creations of the Taiwanese govern
ment. TSMC was established as the world’s first pure-play foundry, meaning that it only produces semiconductors on a contract basis for sale by other firms and does not market its own line. This concept was considered so radical and risky at the outset that sufficient private capital could not be raised for the TSMC venture. As a result, 44 percent of TSMC’s capitalization was provided by the Development Fund of the Executive Yuan (Cabinet), a special fund utilized for high-risk industrial projects. TSMC has grown rapidly, in significant part by absorbing other Taiwanese semiconductor firms that have also received government financial assistance.67
Singapore and Malaysia
Singapore, like Taiwan, is making major investments in fabs, which it will produce on a foundry basis. Singapore’s Chartered Semiconductor Manufacturing (CSM) is already the world’s third-largest foundry producer. The government has played a major role in ensuring that the necessary capital is available to achieve its increasingly ambitious expansion plans. The Economic Development Board invests directly in semiconductor enterprises that establish wafer fabs in Singapore, and has helped Chartered Semiconductor Manufacturing to emerge as a major foundry operator. The Economic Development Board manages a number of funds through which it channels grants, investments, and loans to microelectronics enterprises.68 Singapore’s tax rules provide incentives to priority indus
67 |
The Worldwide Semiconductor Manufacturing Corporation and Acer, both of which have been absorbed by TSMC, received substantial financial support from the China Development Industrial Bank, a government bank with the mission of providing long-term credit and investment funds to strategically important industries. “Account Update—Taiwan,” The Asian Banker Journal April 19, 2000; “Management Statement,” China Industrial Development Bank at <http://www.cdcdpbnk.com/english/e-y-2.htm/>. Taiwan’s attraction of high-technology industries is documented in National Research Council, Conflict and Cooperation, p. 33, Box A. |
68 |
The prototype investment was its 26 percent stake in TECH Singapore, established in 1991 with Texas Instruments, Canon, and Hewlett-Packard. The Cluster Development Fund, created in 1993 under the Economic Development Board’s (EDB) direction, made its first investment in 1994 in Chartered Semiconductor Manufacturing’s Fab II ($100 million Singapore). EDB Investments (“EDBI”) funded a fab to produce DRAMs in conjunction with a Japanese-owned firm, Hitachi Nippon Steel Semiconductor Singapore. In 1997 EDBI joined CSM and Agilent Technologies to finance the construction of a fab to manufacture 8-inch wafers for the digital consumer and communications industries. In 1998 Philips and TSMC joined with EDBI to create a foundry operation in Singapore. The EDB reportedly took a 20 percent equity stake. In 1996 EDBI took a 30 percent equity stake in a joint venture between Hitachi and Nippon Steel to establish a wafer-fabrication facility for 64-megabit DRAMs. See “Government Makes First Major Investment from Cluster Development Fund in Wafer Plant,” The Straits Times, February, 26, 1994; Singapore Investment News with Hitachi Nippon Steel Semiconductor Singapore Pte Ltd Company Profile, http://www.has.hitachi.com.sg/hns/profile/c-about.htm.; EDB, “New U.S. $1.2b Wafer Fab to Start Production in 2000,” Singapore Investment News, November 1, 1998; See also FBIS, “Philips Semiconductor Plant in Singapore,” Algemeen Dagblad, September 30, 1998, p. 14 (BR3009120698). |
tries, including the semiconductor industry, on a sliding scale, with tax rates ranging from 13 percent to zero (normal tax rates are 25.5 percent). An enterprise operating a wafer fab in Singapore will be taxed at the zero rate if it utilizes the most advanced technology.
Malaysia appears to be following the Singapore model; its government investment arm, Khazanah Nasional Berhad, has taken or is planning to take large equity positions in several new semiconductor-manufacturing enterprises in Malaysia.69
Implications
The full long-run implications of the foundry phenomenon for the U.S. semiconductor industry are not clear. The ability of U.S. producers to outsource the production of their designs to some of the most efficient and skilled manufacturers in the world reduces the otherwise-enormous investment risks they would face by producing their own devices and helps to ensure a high-quality end product. Many observers in East Asia warn that if the U.S. industry does not make full use of the foundry opportunities available to it in Taiwan and Singapore, others will (e.g., their Japanese and European competitors). Such a development would work to the serious disadvantage of the U.S. industry. On the other hand, the foundry model requires designing firms to share their proprietary designs with the foundry producers, a relationship that has made European policy makers, at least, uneasy.70 If the U.S. industry surrenders too much of the manufacturing
69 |
Khazanah took a 40 percent equity stake in a project involving Atmel of the United States to build a wafer fab in Malaysia’s Kulim Hi Tech Park. It made a 25 percent equity investment in a joint venture with Malaysia’s AIC Semiconductor to establish a semiconductor packaging enterprise. It reportedly will take an equity share of up to 32 percent in Silterra, a new semiconductor foundry; the government reportedly also plans to guarantee $750 million of Silterra’s debt, waive taxes on $1 billion in future profit, and exempt it from import and export duties. See AIC Corporation Berhad, January 18, 2002, http://www.klse.com.my/website/listing/lc/aic.htm; FBIS, October 15, 1997, reprint of Shamsul Akmar and Lee Yuk Peng, “U.S. Firms Investment Arm Tie Up for Wafer Fab Project,” The Star, October 15, 1996, (FTS19971015000152); AIC Corporation Berhad, Press Release, September 10, 1998. |
70 |
In Germany the state government of Bavaria has directed the Fraunhofer Institute of Solid State Technology (IFT), an institute of applied research, to undertake the local commercial manufacture of application specific integrated circuits (ASICs) designed in Europe. The reason for this move was to give local ASIC designers a European alternative to Asian foundries. “The European market for this type of component was worth around DM 4 billion in 1995. However, only half the ASICs sold were produced in Europe. ‘This means that many medium sized firms have to reveal their know-how to foreign manufacturers,’ said [IFT] Deputy Director Konrad Hieber.” See FBIS, August 15, 1996, translation of Andreas Beuthner, “Fraunhofer Tech Transfer to Industry Viewed,” Computer Zeitung, August 15, 1996, p. 4 (FTS19960815000685). |
function to offshore foundries, it is not clear that it can retain its position of world leadership, which is based increasingly on design functions alone.
Taiwan, for example, is promoting an indigenous semiconductor design capability the way it previously encouraged semiconductor manufacturing. Taiwan has over 100 design houses, which benefit from a number of government programs. In January 2000 ERSO (Electronics Research and Service Organization) organized a club to provide local design houses with a platform to exchange intellectual property, the Silicon Intellectual Property Consortium. The objective of the club is to promote the intra- and inter-company circulation of IP and design reuse. Taiwan’s Chip Implementation Center, administered by the National Science Council, is currently designing over 500 devices in its research programs and is turning out excellent design engineers.71
While Taiwanese design houses today do not constitute a significant challenge to the U.S. industry, it is possible to envision future scenarios in which it is common for an OEM to commission a design from a firm in Taiwan or Europe and contract with a foundry like TSMC to manufacture it, effectively bypassing the integrated device manufacturers, which today account for the preponderance of U.S. semiconductor sales. Some transactions of this kind are occurring already.72
Qualified Workforce
The rapid growth of semiconductor industries around the world has created a massive demand for skilled engineers, scientists, and technicians. At the same time the number of individuals graduating from U.S. universities with electrical and electronic engineering degrees is declining.73 The U.S. has been able to offset this trend to a considerable degree because of its abiding appeal to talented immi
grants and foreign scientists and engineers who become U.S. residents.74 A recent article by an official in Japan’s Ministry of Foreign Affairs made the following observation:
Many of the top students at Stanford University are Asian Americans, and 20 percent of the engineers in Silicon Valley are Asian Americans as well…. The source of the “power” of the United States, which was founded by immigrants to begin with, probably lies in the fact that the society is built on its basic openness, social dynamism, and “competition predicated on change.”… For a Japan that does not have a history as a nation of immigrants, it is rather difficult to transform itself into this kind of society.75
Although the United States continues to exert a powerful attraction to foreign talent as a place to live, work, and pursue opportunity, significant counter-vailing forces are inducing an increasing number of foreign graduates of U.S. universities to return home and U.S.-resident foreign scientists and engineers to relocate abroad.
Personnel shortages in the semiconductor industry are reported throughout the developed and newly industrializing world and are forecast to become more acute in the next decade. Planners in Taiwan and Singapore view prospective workforce shortages as the single most important constraint on the industry’s expansion plans. In Europe an estimated 500,000 IT-related jobs are unfilled, and an annual shortfall of 3,500 new graduates in electronic engineering currently exists and is expected to widen.76
All major semiconductor-producing countries in East Asia are implementing substantial programs to train more specialists and engineers for the semiconductor industry, but these are generally acknowledged to be inadequate for the actual need. One result of this global shortage has been a furious competition to attract
skilled specialists from other countries or to establish design centers in other countries to attract local talent.
Taiwanese planners concede that despite numerous training programs, there is no realistic possibility Taiwan can produce enough engineers on its own—and that “we will find them in the U.S., in Southeast Asia, and in Mainland China.”77 Korea has expressed alarm that foreign countries, particularly Taiwan, have been scouting out and recruiting semiconductor researchers and engineers from Korean companies—a concern that is ironic, given Korea’s own aggressive overseas recruiting practices.78 Singapore has implemented an extremely liberal immigration policy, and its Economic Development Board operates an International Manpower Program, which scours foreign universities for talent.79 In China, where local governments once competed with each other to attract inward foreign investment, they are now engaged in a fierce rivalry to attract foreign talent, particularly ethnic Chinese who “bring with them their S&T accomplishments.”80
77 |
Taiwan has established an extensive network of government organizations to attract scientists and engineers to Taiwan, particularly Taiwanese expatriates and ethnic Chinese. The National Youth Commission maintains a database tracking Taiwanese graduates abroad, including over 11,000 engineers, who are periodically contacted and offered incentives to return to Taiwan. Over 100 of the companies in Hsinchu Science Park were founded by returning expatriates, and by the mid-1990s over 6,000 expatriates were returning annually, reflecting the government’s concerted recruiting efforts. National Science Council Yearbook, 1998; Hsinchu Science Park Web site at <http://www.sipa.gov.tw/en/seconde/induse/induse.html>. |
78 |
In 1993 Korea’s Ministry of Science and Technology (MOST) established a program to “attract and make good use of” the approximately 400 ethnic Koreans working overseas in science and technology fields. This effort was intended to encourage these individuals to help Korea “acquire at an early date the newest science, technology and know-how in the R&D stages in advanced countries.” Under the MOST program foreign scientists with five years’ postdoctoral experience abroad are given airfare and high salaries in return for a six-month commitment to work at a South Korean facility. During the first year of this program offers were made to 57 such individuals, the majority of whom had Korean surnames and were living in the United States. One-third of the 31 acceptances were in the electronics field. (Maeil Kjongje Sinmun, January 25, 1994; Hanguk Kjongje Sinmun, December 27, 1994). Korea’s semiconductor makers have an estimated 4,000 professional researchers, of which 400 are said to be targeted by foreign recruiters. An international incident of sorts erupted in 1998 when a number of Samsung semiconductor engineers defected en masse to Taiwan’s Nan Ya Technology Corp., triggering Korean outrage. Korea nonetheless maintains its own aggressive recruiting efforts abroad. Korea’s MOST maintains a database of ethnic Koreans with science and technology skills living abroad who are targeted by recruiters with generous incentives. The Korean Institute for Science and Technology seeks to recruit “outstanding Korean scholars living in Japan.” (Hanguk Kjongje Sinmun, December 27, 1994; Maeil Kjongje Sinmun, March 15, 1993). |
79 |
Singapore’s Chartered Semiconductor, Ltd., has recruited 750 people through this program since 1995. See “Singapore’s Transition into a Knowledge-Base Economy–Investing in the Future Today,” Singapore Investment News (press release from the Economic Development Board, no pagination), April 1, 2000. |
80 |
Between June 1999 and June 2000 the Beijing municipal government enacted at least six sets of incentives to attract ethnic Chinese to its Zhongguancun Science Park. See “China Actively Partici- |
At present the single most powerful policy-related employment inducement worldwide is the compensation packages made possible by Taiwan’s tax laws— a source of much grumbling by competitors throughout East Asia. One of the most commonly cited sources of the startling competitive advantage achieved by Taiwanese firms like UMC and TSMC is that they “can attract and keep the best people” with compensation arrangements that foreign firms cannot match.
The market value of UMC and TSMC stock has risen dramatically since the date of original issuance, but employees given stock as compensation are taxed on the face value of the shares, not the market value—which is sometimes over 50 times as high. Moreover, when the shares are sold, whether acquired directly or through the exercise of stock options, no capital gains tax is levied, because Taiwan eliminated its capital gains tax in 1990.81 Thus, engineers signing on with these firms face the prospect of becoming very wealthy in a relatively short time, “and that becomes a big reason why Taiwan can attract the best talent in the high-tech industry from at home and abroad.”82
Government-backed Venture Programs
Citing the enormous success of the U.S. free-market model for developing and commercializing new technologies, a number of foreign governments have begun to emphasize what they characterize as entrepreneurism and venture capitalism in their high-technology promotion policies. In contrast to the 1980s, when governments established a series of semiconductor mega-projects involving the major producers, each aimed at producing a very specific milestone device such as a 1-gigabit DRAM, governments now have begun to encourage smaller players to explore a wider range of new technologies, some of which may grow into internationally competitive products.83
China, for example, is establishing microelectronics incubators, organizations intended to generate small, startup semiconductor enterprises systemati-
|
pates in Fight for Overseas Talent,” SinaNet, August 14, 2000, online article, no page citation, http://www.zgc.gov.cn/news/epnews/0008114-4.htm and “Beijing Issued Implementation Rules on Re-cruiting Talents from Outside City,” Beijing Morning Post, August 14, 2000, online article, no page citation, http://www.zgc.gov.cn/news/epnews/000814-5.htm. |
cally.84 Japan’s Ministry of Education is funding a Venture Business Laboratory Program at Japanese universities to promote R&D, foster “talented people brimming with entrepreneurial spirit,” and generate “student ventures” arising out of original research.85
The terms employed in the implementation of these policies—“venture” is the overwhelming favorite, although Singapore has coined the word “technopreneurship”—convey the notion that these countries are emulating the free market policies that have helped foster U.S. venture capitalism, such as an open and transparent securities market and minimal government intervention in choosing the technologies to be financed. A closer look at the actual venture capital programs in microelectronics outside the United States reveals that while they seek to replicate the success of the U.S. system in promoting innovation, the role of government is much more pervasive, and many of the fundamental aspects of traditional industrial policy remain. For example,
• Governments choose the technologies that are to be supported for development.
Bureaucrats continue to consult with industry and academia to identify promising products: the conventional industrial policy practice of providing financial incentives to firms that they hope will be winners, with recognition that some will be losers. In a change from the past, however, the decision-making tends to be more decentralized and flexible, with broad sectors specified by long-range policies but individual funding decisions made by industry-government committees and specific government venture funds. In Korea, for example, the venture companies supplying most of the capital are creations of various government ministries. In Taiwan a number of startups have arisen from government-initiated projects at ERSO, the government microelectronics research institute.86
• Governments provide the funds for development and commercialization of new technologies through these government-supported ventures.
Japan, which has come to view venture businesses as the driving force behind U.S. economic growth in the 1990s, has deployed a broad array of government financial support measures to fund startups in microelectronics. This includes MITI’s Support Program for System LSI Development.87 There are as well benefits available under the New Business Creation Law88 and the Creative Activity Law89, and funding by the Japan Development Bank.90
In Korea the government has established dozens of venture capital companies to provide direct equity injections, equity, loans, and managerial advice to small- and medium-size enterprises; although some of these venture companies have been nominally privatized and have received some private funds, they continue to obtain funds from ministry “promotion funds” for industry.91
87 |
This program, established in 1998, is designed to provide government financial support to small semiconductor design companies and “contribute to strengthening the design capabilities of our country’s semiconductor industry.” MITI Machinery and Information Industries Bureau. See Denshi Kogyo Nenkan 1998, Tokyo: Dempa Publications, Inc., May 10, 1998, p. 212. |
88 |
This law provides for R&D subsidies to small- and medium-size ventures to develop new technologies. One current program funded under this law supports development of system-on-a-chip advanced design technology. Tsushansho Koho, May 16, 2000, pp. 1-4. |
89 |
Under this law a small- or medium-sized venture whose business plan is approved by a prefectural governor can receive low-interest loans, tax incentives, and assistance in lining up investors in order to develop new technologies. Venture businesses benefiting from this law include the Nomura Electronics Technology Research Institute (researching VLSI composed of ceramic-type condensers); Tekunosemu (development of test contracts for the next generation of semiconductors); and Futo Electronics Industry Co. (development of high-quality glass packaging for semiconductor sensors). Nikkan Kogyo Shimbun, p. 3, April 12, 1997; Nihon Keizai Shimbun, March 29, 1996, p. 1; MITI SMEA Web site at <http://www.chusho.miti.go.jp>. |
90 |
In April 1995 the Japan Development Bank established the New Business Support Program to fund venture businesses. This program allows small firms to pledge their intellectual property as collateral, and most of the initial loans went to ventures in the semiconductor and computer industries. One of these laws went to the multimedia venture, Dome, to develop a “semiconductor for animation compression.” In 1997, however, Dome filed for bankruptcy. Yomiuri Shimbun, July 8, 1995, p. 17; Nikkei Sangyo Shimbun, November 21, 1996, p. 1; July 7, 1997, p. 24. |
91 |
Korea Develoment Bank Web site at <http://www.kdb.co.kr/web>; KTB Venture Capital Web site at <http://ktbuc.com/main.htm/>; Korea Technology and Banking Network Web site at <http://ktb.co.kr/eng/investment.htm>; “ROKG, Business to Raise W150 Billion for IT Venture Fund,” Korea Times, January 12, 2000; Press release, Ministry of Science and Technology, KAIST’s High Technology Complex (HTC) Now Open, March 12, 1999, http://www.most.go.kr/press-e/51.htm. |
• Governments continue to support the large semiconductor producers through the venture programs.
Although the beneficiaries of government venture backing in the first instance are small- and medium-size enterprises, the ultimate beneficiaries are often the large, internationally competitive firms. In Taiwan several of the more successful smaller companies arising from ERSO activities have been absorbed into TSMC and UMC, a pattern that relieves the large private producers of the costs borne by ERSO for semiconductor projects that fail.
In Korea the explicit aim of supporting small- and medium-size enterprise ventures in the semiconductor area is to provide the semiconductor chaebol with less expensive local inputs and equipment, and R&D projects that are designed to support chaebol product lines are given priority in receiving government research support.92 The government has also partnered with the chaebol in investing venture capital in U.S. firms that can provide Korean industry with promising technologies.
New Silicon Valleys?
Every country outside the United States that has made a major effort to promote an indigenous semiconductor industry has implemented programs to replicate what it views as the best features of Silicon Valley. Such features include the clustering of mutually supportive enterprises and research organizations within a limited geographic area.93 Typically the government sets aside land for high-technology industrial parks and offers investors incentives for locating there. The sites are chosen in areas that highly educated scientists and engineers would find attractive places to live and work or, at least, preferable to other alternatives in the region.
The government usually ensures onsite technical and infrastructural support in the form of applied research institutes, nearby universities with good electronics programs, national laboratories, secure supplies of electricity and pure water,
92 |
For example, in 1996 the Korean government, academia, and industry established a program to assist the transfer of technology to small- and medium-size enterprises, the University Industry Technology Force (UNITEF). A small- or medium-size enterprise facing a technological hurdle can apply to UNITEF for assistance and, if approved, a research task can be assigned to one of 500 professors taking part in the program. The chaebol help govern and manage this program, and small- and medium-size enterprise subcontractors of the chaebol are given priority in receiving technical support through UNITEF. See “Korean Engineering Professors from a National Group,” Asian Technology Information Program, May 9, 1997 (Report AT1P97.042), no page citation, http://www.atip.or.jp/public/atip.reports.97/atip97.042html. |
93 |
Annalee Saxenien’s review of the growth of Silicon Valley provides a recent example of the cluster phenomenon. Annalee Saxenian, Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge, Mass.: Harvard University Press, 1994. For a multi-faceted view, see Martin Kenney, (ed.), Understanding Silicon Valley: The Anatomy of an Entrepreneurial Region, Stanford: Stanford University Press, 2000. |
and even pilot integrated-circuit production lines to assist local companies in refining process skills. Some of these initiatives have failed or face prospects that are at best uncertain. However, a number of “new Silicon Valleys” have succeeded in attracting a critical mass of semiconductor device makers and various supporting enterprises in concentrated geographic zones that offer significant operational and cost advantages.
• Hsinchu Park
Taiwan’s Hsinchu Science-Based Industrial Park is universally acknowledged as a dramatic success, reflecting not only the generous incentives available to investors but also the presence near the park of (what is probably the foremost applied research organization in the world) the Industrial Technology and Research Institute (ITRI) and its microelectronics affiliates, ERSO.94
The production of integrated circuits accounts for nearly 60 percent of the total revenues generated by firms located in the park.95 Hsinchu and Taiwan’s other science-based industrial parks foster close physical proximity of OEMs, device makers, design houses, and other enterprises engaged in the production of IT equipment. This greatly enhances efficiency and shortens cycle times. U.S. companies that operate in the parks comment that “everything we need is right here,” and that transactions and interactions that require several days in the United States (because of physical distances) can be achieved in a matter of hours in Hsinchu.
• Kumamoto Prefecture
Japan’s Kumamoto Prefecture on Kyushu (Japan’s “Silicon Island”) was one of 26 sites designated in 1984 for development as a “technopolis”; while many technopolis sites failed to thrive, Kumamoto has become a complex of semiconductor businesses, including companies such as NEC, Fujitsu, NTT, Omron, and Tokyo Electron, and the U.S. testing firm Teradyne.96 In addition to incentives offered at the national level97 the prefectural government offered investors exemptions from fixed property taxes, reduced enterprise taxes, low-interest loans, and direct subsidies. The prefecture has also sponsored a collaborative semiconductor R&D project to develop sub-0.1-micron manufacturing technology.98
94 |
A newly established domestic or foreign company in Taiwan’s science-based industrial parks is exempt from corporate income tax for the first 5 years. The rate of corporate and commodity tax on products or services exported by enterprises in the parks is zero. Taiwan Ministry of Economic Affairs Web site at <http://it.moeaidb.gov.tw/committee/english/b-4.html>. |
95 |
Hsinchu Science-Based Industrial Park Administration. |
96 |
Handotai Sangyo Shimbun, June 14, 2000, p. 2. |
97 |
Semiconductor firms locating in technopolis sites qualify for accelerated depreciation of buildings and equipment located in the sites. MITI, Handbook on Specific Facilities Subject to Special Depreciation Allowances, August 10, 2000, p. 155. |
98 |
Kumamoto Technopolis Web site at < http://www.kmt-technopolis.or.jp >. |
• Taedok Valley
Korea’s Taedok Science Town (“Taedok Valley”) was established in the 1980s with the objective of geographically concentrating the country’s R&D resources. Roughly 37 percent of the 566 venture firms in Taedok are engaged in some aspect of the semiconductor business, leading Korean observers to characterize Taedok as “Venture Valley.” Taedok has 70 national, public, and private research institutions and over 14,000 employees holding at least a master’s degree.99 The government is establishing an advanced-technology commercialization center in Taedok to provide support for local venture firms.100
• New Sites in Singapore
The government of Singapore has earmarked about 160 hectares of land at three sites for the location of up to 25 wafer fabs and various supporting enterprises.101 These sites (Woodlands, Tampines, and Pasis Ris) are being developed by Singapore’s Jurong Town Corporation, a statutory board created to take over the task of developing industrial estates from the Economic Development Board.102 Since 1995 Jurong Town has “pulled out all the stops to develop facilities—ahead of demand—to attract the big players in the high-value-added wafer-fabrication industry.”103
• “Chinese Hsinchus”
China is “putting into practice what was successful in Taiwan,” establishing high-technology development zones with the industrial infrastructure for semiconductor manufacturing and financial and tax incentives for enterprises locating in the zones. The scale of the Chinese effort is ambitious: “It is the kind of setting in which ten science and industrial parks of the order of Hsinchu Science and Industrial Park … can spring up all at once.” One semiconductor equipment firm predicts that Chinese versions of Hsinchu “will extend over the entire breadth of China in five to ten years.”104
99 |
See FBIS, May 4, 2000, translation of Yi Che, “Reportage: Taedok Valley–Over 14,000 Researchers with a Master’s Degree or Better,” Chugan Choson, May 4, 2000 (KPP20000510000022). |
100 |
FBIS, April 10, 1998, translation of Chong Ch’ang-hun, “Taedok to Become Mecca for Venture Firms,” Chonja Sinmun, April 10, 1998, p. 1 (FTS19980705000165). |
101 |
See FBIS, July 29, l997, reprint of Jennifer Lien, “Germany to Put up First Silicon Wafer Manufacturing Plant,” Singapore Business Times, July 29 1997, no page citation (FTS19970729000242). |
102 |
The chairman of Jurong Town Corporation’s board, Major General Lim Neo Chian, is also the deputy chairman of the Economic Development Board. Jurong Town implements development plans for high technology at the direction of the government. EDB Singapore, 1998 EDB Board, EDB website, www.sedb.com/edbcorp/an_1998_02.jsp. |
103 |
Jurong Town Corporation, Annual Report 1997/98. |
104 |
See FBIS, March 7, 2001, translation of M. Kimura, “Industry, Government and Universities United in Enthusiasm and Talent for LSIs and then LCDs,” Nikkei Microdevices, March 2001, p. 62 (JPP20010307000001). |
EVENTS ABROAD: GOVERNMENT POLICIES AND SIGNIFICANT RECENT TRENDS
The question of whether government interventions in microelectronics are effective or counterproductive and whether they are to be condemned, emulated, or simply ignored will no doubt continue as the subject of active controversy in this country for many years. For purposes of this overview it will suffice to note that government policies outside the United States are contributing to a number of significant trends in the global industry which have implications for the future competitive position of the U.S. industry. These include Japan’s attempt to bring about a national revival in microelectronics; Taiwan’s continued rapid expansion and its potential influence on the emergence of China as a major competitor; and the resurgence of the European semiconductor industry, particularly in the rapidly growing telecommunications field.
Japan Seeks an Industry Revival
At the beginning of the 1990s the U.S. and Japanese semiconductor industries stood approximately in a position of parity as the dominant players in the global industry. Since then Japan’s position has eroded rapidly, with its market share dropping from about 50 percent of the worldwide total in 1990 to 26 percent in 1998. The Japanese industry concentrated heavily on DRAMs, which became a low-priced commodity with the entry into the market of Korean and (more recently) Taiwanese DRAM producers. The U.S. industry, with leadership in microprocessors, rapidly gained market share (expressed in terms of share of total revenue) as the explosion in demand for personal computers drove demand for higher-value logic devices. At present Japanese firms are burdened with high-costs and large levels of debt as a result of over-investments in DRAMs.105
Planning a Comeback
The precipitous decline of the Japanese industry’s competitive position has led to a wrenching reappraisal within Japan of what has gone so badly wrong and the emergence in 2000 of a coherent strategy for achieving a national revival of Japanese leadership. The government exercises a more important role in this strategy than at any time since the 1970s.
Leaders of the Japanese semiconductor industry have been unsparing of themselves in analyzing the missteps of the past decade. The have publicly acknowledged strategic blunders (like the over-commitment to DRAMs), complacency about the U.S. resurgence and the rise of Korea and Taiwan, and the erosion of their leadership in manufacturing efficiency.106 The implications of the foundry business model pioneered by Taiwan were not appreciated by the Japanese industry, it is now acknowledged, until the very end of the decade of the 1990s.107
Japanese industry leaders also note that the United States undertook a comprehensive and successful collaborative industry-government effort “to incorporate scientific methods into semiconductor production technology, production management technology, on-site maintenance, and the like,” while Japan by contrast “lost the sharing of roles among government, industries and universities” and thus “lost our national strategy for the LSI industry.” Similarly, the strong backing of governments in Korea and Taiwan is seen by these leaders as a major reason underlying why “Japan was usurped of its position” by industries based in these countries, “just as the U.S. was by Japan 20 years earlier.”108
Japanese Consortia
In April 1994, 10 major Japanese semiconductor manufacturers established the Semiconductor Industry Research Institute Japan (SIRIJ), which was to serve as a think tank for the Japanese semiconductor industry, analyzing the overall industrial environment, formulating strategy, and drafting industrial policies.109
“This move can be described as an attempt by the Japanese manufacturers, whose market share has stagnated, to stage a comeback.”110 SIRIJ has developed a succession of policy recommendations for government promotional measures that have subsequently been adopted by the Japanese government.
In March 1995 SIRIJ released a report that called for the establishment of a new generation of R&D consortia.111 Japanese industry leaders noted with irony that Japan ended its subsidized large-scale joint R&D projects in microelectronics in the 1980s in response to criticism by the United States, while the U.S. launched SEMATECH and seized global leadership from Japan.112 By the mid-1990s the Japanese government had concluded that the eroding national position in microelectronics was a more serious problem than any that might arise from friction with the United States.
A new generation of government-supported joint R&D projects was initiated, bringing to an end the so-called “15 years blank period,” which began at the end of MITI’s first VLSI project in 1980. The most important project was the Association of Super-Advanced Electronics Technologies (ASET), a MITI-sponsored consortium engaging 21 Japanese companies to pursue many microelectronics themes, including new forms of lithography (X-ray, laser, electron-beam, extreme-UV). ASET received about $50 billion yen (about $430 million) in government funds between 1995 and 2000. The government’s Japan Key Technology Center founded a seven-company consortium to develop 400-mm silicon wafers, the Super Silicon Crystal Research Institute (SSi).113 In addition to these government-sponsored programs the Japanese industry established two major privately funded research consortia, STARC, which is patterned on the SRC in the
United States, sponsors university-based R&D,114 and Selete, which concentrates on technology for manufacturing 400-mm wafers.115
Revival of the Japanese Semiconductor Industry
In March 1999, amid a sense of crisis arising out of declining semiconductor sales, SIRIJ established a “Semiconductors in the New Century Committee” (SNCC) to draft proposals for a new generation of government-industry R&D projects that were to revive the Japanese semiconductor industry. In March 2000 SNCC delivered a final report, Proposal: Revival of the Japanese Semiconductor Industry, to MITI. The report recommended that industry, government, and universities join together in a new generation of cooperative R&D projects. These would build on recently established consortia like Selete and STARC and would emphasize system-on-a-chip technologies. SNCC’s chairman declared that “no matter what is said overseas, Japanese [semiconductor producers] who have been defeated in a landslide in the 1990s have no option but to form an all-Japan alliance in order to launch a counterattack.”116
The Ministry of the Economy, Trade, and Industry (METI)—previously called MITI—appears to be implementing the basic elements of the SIRIJ program. This program calls for a developmental effort aimed at “system on chip, which is said to be the brains of the industry.”117 Japanese industry leaders calculate that the biggest growth markets for semiconductors in the twenty-first century will not be in PC equipment, where the U.S. industry leads, but in cell phones and digital consumer equipment, where Japan leads. The new-generation R&D projects are designed to establish further Japanese dominance in these areas.118
114 |
The Semiconductor Technology Academic Research Center (STARC) is co-funded by 11 Japanese firms. It funds university R&D based on three criteria: (1) R&D for new technologies that will become industry standards or mainstream products; (2) pre-competitive R&D that can be transferred to industry for application in 5-10 years; and (3) R&D that fosters young researchers likely to make future contributions to the industry. (Denshi, p. 34, November 1995; Nikkei Sangyo Shimbun, November 12, 1998, p. 19; STARC Web site at <http://www.starc.or.jp>. |
115 |
Semiconductor Leading Edge Technologies (Selete) is a privately funded joint venture company established by 10 Japanese semiconductor producers in 1995 to evaluate next-generation manufacturing equipment and materials. This organization parallels some of the functions of the U.S. SEMATECH consortium and was in fact established as a Japanese alternative to a proposal by SEMATECH for an international joint venture for evaluating next-generation semiconductor manufacturing equipment. Denshi, November 1995, p. 34; Nikkei Sangyo Shimbun, February 14, 1996, p. 11; Tokyo Semiconductor, June 2000, pp. 36-40; Selete Annual Report Fiscal 1998 and Fiscal 1999. |
116 |
Nikkei Sangyo Shimbun, December 6, 1999, p. 8. |
117 |
Nikkei Microdevices, December 1999, pp. 98-103. |
118 |
See FBIS, January 2, 2001, translation of “From Stagnation to Growth, the Push to Strengthen Design,” Nikkei Microdevices, pp. 106-124 (JPP20010125000012). |
The head of its Machinery and Information Industries Bureau, Akira Kubota, summarized the plan:
[T]his time the government will take the lead in semiconductor projects that involve joint efforts by industry, government, and universities. Leading-edge technology development will be the government’s role. First, we considered building clean rooms at national laboratory sites. That is because nothing will come of silicon technology research themes unless there are clean rooms. For the clean-room construction we appropriated 16.5 billion yen in the FY 1999 supplementary budget. Next, we will launch two semiconductor projects. One is the development of basic next-generation LSI technology using the government’s clean rooms. In that effort we will work with industry in the development of new materials, measurement technology, and so forth. The other semiconductor project has to do with the development of equipment, including a small-scale production line.119
The specifics of the new government-industry effort were clarified in late 2000 and early 2001, indicating a dramatic resurgence in government support for this sector:
• Super Clean Room
METI will build a 4,500-square-meter “super clean room” in Tsukuba Science City to foster industry/government/academia cooperation in the development of system-on-a-chip technology.120 The project aims primarily at improving the performance of transistors belonging to the 70-nm technology generation through research on high-k gate dielectric process technologies and other relevant technologies. The research targets also include low-k interlayer dielectric process technologies and lithography and mask technologies for the 70- to 50-nm technology generation. METI reportedly sought 8.1 billion yen (about $70 million) in the FY 2001 budget for this project.121 In all, the government is expected to invest 30 billion yen (about $260 million) in the 7-year project.122
• Future Information Society Creation Laboratory
The Japanese government plans to invest 30 billion yen (about $260 million) in this 5-year project, which will create a “new, small-scale and very short-term
microelectronics production line” and develop super-high-speed LSI devices (SOILSI) and high-performance system LSI devices, “rather than in the area of memory, as in the past.”123 The small-scale line is seen as an advantage in producing devices for digital home electronics, an area where consumer preferences change quickly and short delivery time is required. The laboratory is headed by Tohoku University Professor Tadahiro Ohmi, who has been serving on various government advisory councils, including the Industrial Technology Council, an advisory body to METI for industrial technology policy. He is regarded as one of the most influential scholars with respect to government promotion of the semiconductor industry in Japan.124
• “Asuka” Sub 0.10-Micron Project
In September 2000 it was reported that 11 Japanese semiconductor producers had agreed to invest 76 billion yen (about $750 million) in a 5-year project to develop manufacturing technology for circuit widths of 0.10 to 0.07 micron and less by 2005. STARC will oversee Asuka’s R&D with respect to design technology, and Selete will manage development of device and process technology.125 A staff of 250 researchers from the 11 core companies will be assigned to Selete and 90 to STARC.126 The Electronic Industries Association of Japan has asked the government for assistance for the Asuka project in four areas.
-
Joint research in METI’s Super Clean Room;
-
Research on materials and measurement-related component technologies;
-
Support for semiconductor manufacturing equipment makers’ development of leading-edge technologies; and
-
Support for the development of system LSI design technology.127
METI reportedly will assist this effort with its own parallel R&D effort and is requesting $60 million for FY 2001 for the first year of an anticipated 7-year effort.128 (See Table 1.)
TABLE 1 Government-Supported Microelectronics R&D Initiatives Outside the United States
Country |
Project |
Research Period |
Government Contribution |
Themes |
Japan |
Next-Generation Semiconductor R&D Center (super clean room) |
2001-08 |
$300 million ($60 million in 2001)a |
Process and device technology for 70-mm generation |
Japan |
Future Information Society Creation Laboratory |
2001-06 |
$300 million |
Create small-scale, very short-term semiconductor production line |
Japan |
Asuka |
2001-06 |
Will use METI super clean room |
Develop design technologies for 0.10- to 0.07-micron system-on-a-chip and device process technologies |
Japan |
NEDO projects |
2001- |
Budget for one NEDO project, development of a gas-cluster ion-beam system is reportedly 2 billion yen (about $20 million) |
Cluster ion-beam process technology, system-on-a-chip design technology, advanced parallel-compiler technology |
Japan |
ASET |
1995- |
$500 million |
Lithography, semiconductor manufacturing technology |
Japan |
Selete |
1996- |
Manufacturing technology for 300-mm wafers |
|
Japan |
STARC |
1995- |
Basic research |
|
European Union |
MEDEA |
1997-2000 |
$720 million (est.) |
Process technology, design, applications |
European Union |
MEDEA Plus |
2001-09 |
$1,350 million (est.) |
Systems-on-a-chip, UV lithography |
European Union |
PIDEA |
1998-2002 |
$135 million (est.) |
Packaging and interconnection |
Germany |
Semiconductor 300 |
1996-2000 |
$680 million |
300-mm wafer technology |
France |
Crolles I and II |
1998- |
$136 million (est.)d |
Pilot 300-mm fab |
Belgium |
IMEC |
Permanent research institute |
$40 million/year (est.) |
System-on-a-chip designs; next-generation (sub 0.10-micron) production technology; packaging |
France |
LETI (GRESSI, PLATO and PREVUE) |
Permanent government laboratory |
Performs R&D for transfer to industry |
CMOS technologies, alternatives to CMOS, UV silicon semiconductor technology |
Germany |
BMFT programs |
Permanent |
Direct funding of R&D and contributions to FHG institutes of applied research |
Flexible manufacturing, microsystems, non-silicon semiconductor technology |
Taiwan |
ASTRO |
2000- |
Government will fund half |
Technology induction, upgrading of local industry |
aMETI requested $60 million in FY2001 budget for first year of a 7-year project. bPrivately funded but received NEDO contract to develop technology to cut PFC use. cMostly private funding; Key Technology Center provided subsidy for CAD software development. dCrolles I reportedly received subsidies of FF 900 million to FF 1 billion. Additional subsidies have been requested for Crolles II. |
• NEDO Initiatives
METI’s satellite R&D organization, the New Energy and Industrial Technology Development Organization (NEDO), will contract with domestic organizations to foster “cluster ion-beam process technology,” “system-on-a-chip advanced design technology,” and “advanced parallel-compiler technology.”129 One of these projects got under way in October 2000: a 5-year effort to develop a full-scale gas cluster ion-beam system for ultra-low-energy ion implantation, lateral surface sputtering, and formation of high-quality thin films.
The project, based in Kyoto University, involves a variety of private companies and government and university research institutes. The budget is 2 billion yen (about $20 million). The director of the project, Isao Yamada, explained that given the relative scarcity of venture capital in Japan to support R&D projects of this scale, “the development of key technologies such as semiconductor processes can only be carried out in Japan with financial assistance from the national government.”130
• Nanotechnology Initiatives
The Japanese government has since 1985 been promoting nanotechnology R&D; Japanese scientists believe their country is dominant in this field.131 However, the Clinton administration’s announcement of the U.S. National Nanotechnology Initiative caused a “nanotech shock” in Japan, which saw a U.S. threat to its perceived leadership. METI announced plans to form a Japanese Nanotechnology Consortium, and plans were announced to set up a Nanomaterial Research Center linked to industry and academia. METI’s Electrotechnical Laboratory is “pushing forward with the front line” of nanotechnology research, examining themes such as three-dimensional optical device structures and devices in which memory can be read or written “by means of a single photon.”132 The government’s fiscal 2001 budget for nanotechnology research was increased by 25 percent, to 38.2 billion yen.133
• Promoting Design Capability
METI secured $68 million in its FY 2000 budget for “strengthening semiconductor design capabilities” and “semiconductor microprocessing basic technology.”134 It requested about $100 million in the FY 2001 budget for R&D on microelectronics-related subjects, “system-on-a-chip advanced design technologies,” and semiconductor device process technologies.”135
Telecommunications Policy
The Japanese effort to challenge U.S. leadership in microelectronics emphasizing non-PC-based applications has been assisted both directly and indirectly by Japanese telecommunications policy. Nippon Telephone and Telegraph (NTT) holds a dominant position in Japan’s telecommunications market. Although NTT has styled itself a private entity since 1985, the government retains a 53.15 percent equity stake; NTT remains in many respects a public organization. NTT— the most profitable enterprise in Japan—utilizes revenues garnered from its nearmonopoly of telephone service to fund R&D in telecommunications-related fields, including microelectronics.136 In recent years NTT has pursued research themes emphasizing applications with respect to cellular phones and other forms of wireless communications.137
The deregulation of telecommunications in the United States is widely cited as a key factor in U.S. dominance of the Internet in the 1990s, and NTT’s regulatory policies—which have restricted competition and prevented low-cost Internet access in Japan—are widely criticized in Japan as the main reason Japan has lagged in incorporating Internet technology throughout its economy.138
However, with respect to wireless communications, U.S. regulatory policy has fostered the emergence of five incompatible wireless standards, making it difficult to establish a uniform subscriber base. By contrast, Japan utilizes a single standard and partly as a result is far ahead of the United States in wireless communications.139 Most Japanese semiconductor makers are placing a high priority on developing devices that build this growing leadership, which they see as a high-volume technology driver enhancing their overall competitiveness.140
IT Strategy
In the latter half of 2000 the Japanese government launched a “Basic IT Strategy,” a sweeping array of promotional measures and legal reforms with the stated objective of overtaking the United States as a “high-speed Internet superpower” within five years.141 One element of this plan is an ambitious public works spending program designed to link most of Japan’s households with high-speed fiber-optic broadband connections by 2005. This would give Japan far more pervasive broadband coverage than the United States.142
Japanese electronics hardware manufacturers are developing products along-side which will take advantage of Japan’s near universal broadband coverage. These include interactive televisions and home appliances. These may, at least
partially, displace the PC and by so doing “blast a hole in U.S. dominance of the Internet.”143 MITI’s investment in a small-scale production line is expressly intended to enable Japanese semiconductor manufacturers to produce devices for digital home appliances running on broadband connections.
The Japanese national microelectronics effort has taken several years to launch and has been characterized by no small amount of confusion and disagreement.144 But the outline of a coherent plan has now been established and very substantial public and private resources have been committed to implementing that plan. Professor Tadahiro Omi, who runs the project tasked with developing a small-scale production line, says of this effort that “Japan is pursuing the way to victory.”145
European Resurgence
The European semiconductor industry—long the butt of criticism from both inside and outside of Europe—was thought at the beginning of the 1990 to be in a state of irretrievable decline.146 The billions of dollars of government subsidies poured into attempts to reverse the industry’s fortunes appeared to have no effect. Various strategies based on consortia, national champions, and bureaucratic community-level programs seemed wasteful and futile.
This European decline has reversed through the 1990s.147 All three of Europe’s semiconductor firms are in the top 10 worldwide and are growing faster than some of the other top 10 firms.148 The European technology base is world class, and Europe is said to be on the leading edge in developing next-generation
manufacturing technologies.149 A recent study by the European Electrical Engineering, Electronics, Information Technology Association (VDE) predicted that, based on current observable trends, Germany along with the rest of Europe would lead the world in microelectronics and nanotechnology by the year 2010, “followed by the United States and Japan/Asia.” The VDE chairman, while endorsing these findings, remarked, “I caution against arrogance.”150
One of the major factors in Europe’s remaining success within the microelectronics field has been the region’s strength in non-computer-related OEM markets. These relate primarily to telecommunications but also involve automotive electronics, smart cards, and multimedia consumer products. European producers are far less heavily invested in the production of standard memory chips or major PC microprocessors than are producers in the rest of the world. As a result they have been less affected by the volatility experienced in those product areas.151 Most European fabs combine analog with digital circuitry on the same chip to make semi-custom parts for telecom and embedded system applications.
European fabs are have also learned to efficiently place increasing amounts of interconnectivity onto a complex integrated circuit instead of a circuit board, as well as to mix intellectual property cores on the same chip.152 In the mid-1990s the telecommunications sector began to grow rapidly, “and it became clear that the communications market would [eventually] stand side-by-side with the PC and public service sectors in providing traction for the LSI market.”153 A number
149 |
“An Executive Report on the MEDEA Programme: Mid-Term Assessment,” Future Horizons, online report, http://www.minez.nl/kamerbrieven1999/execut.pdf, accessed on July 19, 1999. ASML’s proposed acquisition of SVGL, the principal remaining U.S. lithography firm, supports this perspective, although at this writing the acquisition has not yet been approved. |
150 |
See FBIS, October 11, 2000, translation of D. Wendeln, “Europe on the Way to World Class in Microtechnology and Nanotechnology,” VDI Nachrichten, October 6, 2000, (UEP20001011000369). |
151 |
See FBIS, October 6, 1995, translation of “Largest Eureka Project a Resounding Success,” Economische Zaken, October 6, 1995, pp. 8-9 (FTS19951006000613); FBIS, May 24, 1996, translation of “Remarkable Results of JESSI ‘Phase Shift’ Project,” Computable, May 24 1996, p. 3 (FTS19960524001698); “European Market Roundup: Continued Growth Predicted Through 2001”. Solid State Technology, March 1, 2000, p. S10. |
152 |
At the STMicroelectronics fab in Crolles, France, chip designers now work at workstations right in the fab, and SOITEC has built a fab manufacturing silicon-on-insulator wafers directly adjacent to the STMicroelectronics fab. “Fab Trends: Are Europe and Asia Leading the Way?” Solid-State Technology, March 1, 2000, p. 8. |
153 |
“European LSI manufacturers are riding on a growth track for 2001 using the sudden growth of the LSI market for communications as the tailwind. Until now U.S. forces have primarily controlled the personal computer (PC) field, which has borne the role of traction for the LSI market, and Japanese manufacturers have controlled the public sector. European LSI manufacturers, on the other hand, have focused heavily on the communications field. The communications sector began growing rapidly in the latter half of the 1990s, thus indicating that the strategy of European LSI manufacturers was right on target. Rapid growth is expected for the communications sector after 2001, and European manufacturers intend to ride this wave and are making necessary preparations to achieve high growth,” |
of industry observers believe that the European semiconductor producers’ strategy of concentrating their efforts in this area is now paying off.154
Europeans freely concede that their often derided, sometimes cumbersome government promotional programs have made an important difference in the rebound of their semiconductor industry. These include:
• “Top down” pre-competitive research and development
“Top down” pre-competitive research and development, sponsored, planned, and implemented pursuant to the European Commission’s Framework Programs, is funded in part from the European Community budget. This category embraced the ESPRIT project (concluded in 1998) and now involves a substantial work program in microelectronics pursuant to the European Union’s Fifth Framework R&D program. The Essential Technologies and Infrastructures division of the Fifth Framework, of which microelectronics and optoelectronics comprise a significant subcategory, is being funded at the level of 1.1 billion euros between 1999 and 2002.155 The microelectronics effort is concentrated on system-on-a-chip technology and telecommunications devices, where the European industry sees its greatest future advantage.156
• “Bottom up” joint R&D by European companies
“Bottom up” joint R&D initiatives by European companies to develop marketable products under the rubric of the EUREKA program are often funded by national governments. The first major microelectronics project of this kind was JESSI (1988-96), a $3.6 billion effort widely credited with a dramatic improvement in the competitiveness of European semiconductor makers.157 Among JESSI
|
in FBIS, January 1, 2001, translation of “3 major European LSI Makers Show Stable Growth Through Large Investments,” Nikkei Microdevices, January 1, 2001, pp. 88-93 (JPP20010131000003). |
154 |
Ulrich Schumacher, CEO of Germany’s Infineon Technologies AG (formerly Siemens), made the following comments in January 2001: “[I]n the communications sector, we are already the world’s leading company in wireless communications. We will strive to become the world’s leader in the wired communication sector in 2001. Concretely, we are the world’s leader in the fields of RF, ISDN, VDSL (very high-bit-rate digital subscriber line), and fiber technology, and we are maintaining the number two position in the base band processing field. Next, we intend to reach the number one position in the remaining fields, and then realize a ‘full system solution’ for which all of these fields are combined with all functions realized in one chip.” FBIS, January 2001, translation of Ulrich Schumacher, “2001 to be a Good Year, Expand DRAM in Areas other than PC,” Nikkei Microdevices, January 2001 (JPP20010131000003). |
155 |
IST Workprogramme 2000 at <http://www.cordis.lu/ist/present.htm>. |
156 |
IST Workprogramme 2000 URL is <http://222.cordis.lu/ist/bwp_en10.htm>. |
157 |
“European Market Roundup: Continued Growth Predicted Through 2001,” Solid-State Technology, March 1, 2000, S10. In a 1996 French analysis of the rebound of Europe’s semiconductor industry three factors were cited: (1) a strong market in Europe, (2) restructuring by the European Union producers, and (3) JESSI. “And the third element playing a role in the European industry’s recovery was JESSI. Launched in 1989, this research program enabled the closing of the European industry’s |
accomplishments were “[contributing] substantially to IMEC’s ascent as the leading center for microelectronics R&D in Europe,”158 the development of two devices that “buttressed...Europe’s leading role in the sector of cellular telephones,” and the fostering of much closer cooperation between European universities and the semiconductor industry.159
JESSI was succeeded by the MEDEA microelectronics project, which began in 1997. It will conclude at the end of 2000 and be succeeded by MEDEAPlus, a 9-year follow-up program with an annual budget of 500 million euros, of which about 40 percent will be government funded.160 MEDEA has reportedly yielded good results in developing technologies to support chips in the telecommunications, multimedia, and automobile industries.161 A related EUREKA project, PIDEA, has been established to address packaging and interconnection technologies, subjects that were not covered by MEDEA.162
• National and regional government financial support for semiconductor R&D and state-of-the-art production facilities
The German federal government and the state government of Saxony heavily subsidized a project to build a pilot 300-mm wafer fabrication facility in Dresden. This became the first such facility in the world to become operational.163 A second 300-mm fab is also being established at the Dresden site by a new entity— Infineon Technologies SC300 GmbH & Co. The government of Saxony will hold an equity position of 115 million euros in this and the German Federal Ministry of Education and Technology is expected to contribute another 75 million euros to this project.164
A similar effort is being sponsored with government assistance at a site in Crolles, France. In 1998 ST Microelectronics established a research facility and a pilot 300-mm fab with government subsidies estimated at Fr 900 million to Fr 1 billion.165 In April 2000 ST Microelectronics and Philips announced plans for a second 300-mm fab at Crolles that would serve as a research facility and a commercial fab whose output the two firms would share equally. They reportedly planned to ask the French government for contributions to the capital and research costs of this project.166
• National and regional government support for major microelectronics research laboratories and organizations
IMEC, founded by the regional government of Flanders in Belgium, is a large, independent, extremely highly regarded center that contracts with European governments and European and foreign companies to conduct microelectronics R&D. IMEC has spun off numerous semiconductor enterprises and has won global acclaim for a succession of technological achievements.167 The gov
163 |
See John Baligu, “Top Fabs of 2000,” Cahners Semiconductor International, September 7, 2000, pp. 1-4, http://www.semiconductor.net/semiconductor/issues/2000/200005/six0005fab.asp. This plant was achieving yields of 90 percent for 64-Mbit DRAMs by March 2000. “Infineon 300-mm Processes Achieve 90 percent yields on 64-Mbit DRAMs,” Semiconductor Business News, April 3, 2000. |
164 |
“Infineon Plans $1 Billion Move to 300-mm Production in Dresden,” Semiconductor Business News, March 31, 2000, <http://www.siliconstrategies.com/story/OEG2000033150023>. |
165 |
Of this total, 200 million was contributed by Department of Isere and 200 million by local administrations. See FBIS, May 13, 1998, translation of Gilles Musi, “Strategy – SGS-Thomson Investing Almost 3 Billion Francs in Grenoble,” La Tribune (Internet Version), May 12, 1998, (FTS19980513000697) and FBIS, May 12, 1998, translation of Anne Chatel-Demerge, “The SGS Thomson Group Invests 3 Billion Francs in Grenoble,” Les Echos, May 12, 1998, p. 13 (FTS19980512000875). |
166 |
See “Chip Giants Prep. Fabs as Wafer Suppliers Gear Up – Europe, Japan Tool 3mm,” Electronic Engineering Times, May 1, 2000. (Nexis reprint). |
167 |
IMEC researchers were the first in the world to make a CMOS image-recording semiconductor that performs as well as a charge-coupled device (CCD), a key component in cameras and camcorders. The CMOS device is far less expensive than a CCD and, as a result, “within a few years video and |
ernment of Flanders has guaranteed to fund nearly half of IMEC’s $90 million annual budget through the end of 2001.168
In France the government Laboratory for Electronics, Technology, and Instrumentation (LETI) is conducting microelectronics R&D that is transferred to private companies, is supporting the creation of startups, and is sponsoring consortia addressing specific microelectronics themes, such as GRESSI (CMOS technology and non-volatile memories), PLATO (sub-1.10-micron CMOS and alternatives to CMOS), and PREVUE (ultraviolet lithography).169
Finally, the German Research Ministry (BMFT) funds microelectronics R&D both directly170 and through Germany’s excellent system of applied research institutes, the Fraunhofer Gesellschaft (FHG).171
Euraccess
The Euraccess initiative is a program organized under European Union auspices and run jointly by IMEC and LETI. It has the objective of “identifying platforms where academic and industrial institutions can jointly study new ideas [with respect to deep sub-0.01-micron microelectronics] and their industrial feasibility.” It establishes a network of hubs and labs—institutions active in leading-edge microelectronics R&D. Hubs examine themes such as the limits of MOSFET functionality, alternative gate dielectrics, optical interconnections, and alternatives to silicon device structures. Research results within hubs are shared within common teams, and industrial partners assume responsibility for transferring the technology developed within the hubs to pilot fabs. “Labs focus on specific technologies, with limited industrial support. Researchers are expected to move between the hubs and labs.”172
Telecommunications Policy
The European resurgence in microelectronics has been based in substantial part on European strength in wired and wireless telecommunications markets.173 As in Japan, European regulatory policy has resulted in a single wireless standard, rather than the five incompatible standards that characterize the U.S. market.174 The European telecommunications industry has itself benefited from a series of long-term, large-scale European Union-sponsored R&D projects that began in the 1980s and continue today.
The first project, RACE (1985-95), successfully achieved the introduction of integrated broadband communication (IBC) systems throughout Europe, either in specialized scientific networks or limited public services.175 RACE was followed by ACTS (1994-98), which pursued a variety of themes related to the implementation of advanced communications systems.176 The Wireless Strategic Initiative
172 |
“Technology Networking Will Speed European Research,” Solid State Technology, March 1, 1999, p. 12; IMEC, “What is Euraccess?” at <http://www.imec.be/EURACCESS/‡‡summary>; “Deep-Submicron R&D Program Hopes to Open Doors to Advances for Europe’s Chipmakers,” MicroMagazine.com, March 1999, at <www.micromayazine.com/archive/99/03/break_1.htm/>. |
173 |
European Union Commissioner Erkki Liikanesi commented in April 2000 that in a competitive comparison between the European Union and the United States, “The positive side is of course mobile communications, because here Europe is the leader,” in FBIS, April 17, 2000, translation of interview by Stefan Krempl with Erkki Likanne, “EU Commissioner: We Need a Sense of Urgency,” Munich Telepolis, April 17, 2000, no page citation given (EUP20000417000266). |
174 |
The European telecommunications standards authority required the European Union member states to develop a single digital network standard, GSM (Global System for Mobile Telecommunications). |
175 |
ACTS Web site at <http://www.ukinfowin.org/ACTS/ANALYSYS/INTRO/chap1.htm>. |
176 |
Official Journal of the European Communities, L.126/1, May 18, 1994. |
(WSI), part of the European Union’s Fifty Framework Program, will evolve into the so-called Wireless World-Net (next generation of wireless systems).177
In addition to these programs a number of stand-alone projects administered by national governments within the European Union are developing microelectronics technologies with wired and wireless applications.178
The Potential Emergence of China
The government of China has been promoting an indigenous semiconductor industry since the 1980s. It has been hampered in this effort by resource shortages and by national security restrictions imposed by more developed nations, including the United States and Japan, on the export of advanced semiconductor manufacturing equipment to China. Existing Chinese wafer fabs can only produce semiconductors on 5- and 6-inch wafers with line widths of 0.5 to 1.6 microns.
A poor infrastructure, low productivity, and the small scale of all indigenous manufacturers disadvantage the industry. China must import almost all integrated circuits needed for the production of color televisions, air conditioners, refrigerators, computers, and communications devices.179 At present China does not account for even 1 percent of the world semiconductor market.180
Despite the difficulties confronting the Chinese industry at present, many industry observers predict that China will emerge as a major competitor in 10
177 |
WSI Web siteat <http://www.ist-wsi.org/project.htm>. |
178 |
For example, the RF Front End Project (1998-2001), 50 percent funded by the German government, is intended to improve and speed up radio frequency (RF) circuit design and help integrate that design into the wireless system design flow. “Joint European Electronics Consortium—Cadence Targets RF Design,” Electronic Engineering Times, March 16, 1998. Since the early 1990s the German Ministry of Research (BMFT) has provided major research funding for optical communications systems. See FBIS, July 31, 1997, translation of Photonik-Foerderschwerpunkt-Photonik in Rahmen des Foerderkonzepts Informationstechnik, Cologne, August 5, 1997, no author given, no pagination (FTS1997073100045497G16107). |
179 |
See FBIS, November 1, 2000, translation of Chu Dechao, “Overview of the Semiconductor Industry in China,” Tokyo Semiconductor FDP World, November 2000, pp. 176-79 (JPP20001127000111). “[O]ur country’s integrated circuit industry has remained rather weak and small in terms of its overall scale and has lagged relatively far behind in terms of its production technology development capability, product design and development capability and standard, and so on, with its product sales volume taking up only 8 per thousand of the world integrated circuit sales volume and less than 20 percent of the domestic market demand. As the majority of our products are of intermediate or low standard, we have had to import core key products from abroad (such as CPU, DSP for mobile telecommunications, and so on).” FBIS, May 15, 2000, translation of Qu Weizhi, vice-minister of Information Industry, “How to Develop Integrated Circuits Industry,” Renmin Ribao, May 15, 2000. |
180 |
See Grant Johnson, “Is China’s Semiconductor Market Worth the Risk for Multinationals? Definitely!” Electronic News, March 1999, p. 10, <http://www.instat.com/insights/semi/1999/china42999.htm>. |
years. (Some Japanese observers, noting that Chinese engineers “absorb [semiconductor] technologies at an unimaginable speed,” believe that it will happen sooner.181) The government has frequently indicated the importance that it places on fostering this industry.182 China already manufactures or assembles a broad array of electronic equipment incorporating semiconductors, accounting for nearly 6 percent of world semiconductor consumption.
By 2010 China is forecast to be the world’s second-largest semiconductor market, after the United States.183 The Chinese government is using the prospect of access to this growing market as well as financial and operational incentives to attract world-class foreign semiconductor device, equipment, and materials producers to invest in China.184 The U.S. producer Motorola is building a facility in Tianjin that, when completed, “will become one of the largest semiconductor, manufacturing facilities in the world.”185 NEC is investing heavily in semiconductor production facilities through joint ventures in Shanghai and Beijing,186 and other Japanese producers have begun to map out ambitious investment initiatives in China.187
Much of the Chinese government assistance is being provided pursuant to Project 909, a $1.2 billion program initiated in 1995 to establish five device manufacturing companies and at least 20 design and development centers in a Pudong
181 |
See FBIS, March 7, 2001, translation of M. Kimura, “Industry, Government and Universities United in Enthusiasm and Talent for LSIs,” Nikkei Microdevices, March 2001, p. 62 (JPP20010307 000001). |
182 |
“As integrated circuits concern economic development and national security, we should promote integrated circuits development by integrating the state will with a market mechanism.... [T]he state should map out preferential policies to support its development, including preferential revenue, investment, capital coordination, and qualified personnel policies, as well as other incentive policies, in order to speed up its development.” See FBIS, May 15, 2000, translation of Qu Weizhi, “How to Develop Integrated Circuits Industry,” Renmin Ribao, May 15, 2000, p. 11 (CPP20000515000066). |
183 |
Michael Pecht, Weifeng Liu, and David Hodges, China’s Semiconductor Industry (Office of Naval Research, 2000) at <http://intri2.org/ttec/aemo/report/index.htm>. |
184 |
See FBIS, July 28, 1997, translation of “State Encourages Foreign-Founded Microelectronics, Basic Electronic Product Items,” Zhongquo Dianzi Bao, January 14, 1997, p. 9 (FTS19970728 001781); FBIS, April 7, 1999, translation of Hu Angang, “How Does China Attract Foreign Direct Investment,” Guangzhou Gangao Jingji, February 15, 1999, pp. 33-37 (FTS19990407001093); FBIS, July 12, 2000, translation of Suo Yan, “PRC’s State Council Drafts New Policies to Develop IT Industry,” Renmin Ribao, July 12, 2000,(CPP20000712000106). |
185 |
The Motorola plant will make devices for incorporation into Motorola handsets, also produced in Tianjin. See FBIS, August 22, 2000, reprint of Matthew Miller, “Beijing Approves Motorola Base, To Work With Taiwan Tycoon, Jiang Zemin’s Sun,” South China Morning Post (Internet Version), August 22, 2000, no page citation (CPP20008220000287). |
186 |
See FBIS, February 2, 2001, translation of “Japan’s NEC to Expand Semiconductor Production in China,” Kagnaku Kogyo Nippo (Nikkei Telecom Database Version), January 12, 2001, p. 9 (JPP20010202000013). |
187 |
See FBIS, March 1, 2001, translation of “LSI Makers Eye China for Developing New Business Strategies,” Nikkei Microdevices, March 2001, p. 72 (JPP20010307000003). |
New Area of Shanghai. Most of these new enterprises are being created in partnership with foreign firms.
U.S. Export Control Policy
U.S. policy toward China at present is to deny approval for export licenses for semiconductor-manufacturing equipment capable of producing devices using design rules less than 0.35 microns. This policy has impeded the development of China’s semiconductor industry.188 However, governments in Japan and Europe are becoming less restrictive, with the result that Chinese fabs are procuring from vendors in those countries equipment that cannot be obtained from U.S. sources.189 NEC, for example, is supplying DRAM manufacturing technology to Shanghai’s Hua Hong NEC Electronics Co., which will use 0.25-micron design rules, more advanced than permitted under the U.S. export control regime.190
Taiwan in China
Just as Taiwan has transformed the global competitive picture during the past five years, Taiwanese initiatives may accelerate China’s emergence as a first-rank global competitor. At present, encouraged by both the Chinese and Taiwanese governments and driven by a growing workforce shortage in Taiwan, the Taiwanese information industry is relocating much of its manufacturing operations to the mainland.191 Taiwanese Minister of Economic Affairs Lin Hsin-yi said in November 2000 that Taiwanese firms controlled 50 to 60 percent of China’s total production of information technology hardware.192 The Taiwanese semiconductor industry is widely expected to relocate most of its low-end (200-mm wafer, 0.25 micron and above) operations to China by 2005. Taiwan’s desktop computers are now largely produced in Taiwanese-owned facilities on the mainland, as well as 56 percent of Taiwan’s motherboards, 88 percent of its scanners, 74 percent of its CD drives, and 58 percent of its monitors.193
The Chinese government, “which has carefully observed how Taiwan succeeds in the LSI industry, hopes to as far as possible put into practice measures that were successful in Taiwan.”194 It is putting in place special incentives to lure Taiwanese semiconductor investment. Meanwhile, the government of Taiwan— although officially banning direct investment in China—is encouraging, through its tax policy, domestic semiconductor producers to relocate their less sophisticated operations to the mainland.195 Some significant moves are already in progress:
-
Chang Ju-ching, former president of Taiwan’s Winbond Semiconductor Manufacturing Company (SMC), has announced plans to use Toshiba’s technology to build two 200-mm wafer fabs in Shanghai, one of which will use 0.25-micron design rules.196
-
Winston Wang, son of Formosa Group Chairman Wang Yung-Chang, has formed a joint venture—the Hongli Semiconductor Company—with Jiang Mianheng, son of China’s President Jiang Zemin. The venture is to build six fabs in Shanghai, three of which will eventually utilize 300-mm wafer technology. Japan’s Oki Electric will reportedly provide technical assistance, and the Chinese government has reportedly provided preferential financing and a tax holiday.197 Wang reportedly will invest $1.63 billion in this enterprise. Ground was broken on the first 8-inch fab on November 18, 2000, with the startup of mass production (50,000 wafers/month) set for the second quarter of 2002. Wang reportedly stated at the ceremony that “as long as the market needs, Hongli can start to produce chips of 0.18 microns instead of chips of 0.2 microns as it is currently planned.”198
194 |
See FBIS, March 7, 2001, translation of M. Kimura, “Industry, Government and Universities United in Enthusiasm and Talent for LSIs,” Nikkei Microdevices, March 2001, p. 62 (JPP20010307 000001). |
195 |
“Investment in Mainland China by Taiwan Enterprises: Present Status, Problems and ROC Government Assistance,” Taiwan Industrial Development and Investment Center, internal memorandum; “Beijing Welcomes Taiwan Semiconductor Firm,” Taiwan Economic News,” June 28, 2000, at <http://www.taiwanheadlines.com>. |
196 |
“Since it is a joint venture between the most powerful young master and the richest one across the Taiwan Strait, government officials on the Mainland naturally give all the green lights. The Shanghai Municipal Government guarantees full support. And the state-owned bank on the Mainland has agreed a generous loan of 2.5 billion US dollars. Moreover, the government will provide the most favored treatment to this special project, including tax exemption for five years, etc.” See FBIS, May 11, 2000, reprint of Wen-Hung Fung, “Taiwan’s IT Production Continues Moving Into China,” Taipei Central News Agency, May 11, 2000, (CPP20000511000146) and FBIS, December 5, 2000, translation of Xia Wensi, “The Eldest Young Master Jiang Sailing Smoothly Through Business World – China’s Telecom King Jiang Mianheng,” Kai Fang, December 5, 2000, No. 168, pp. 11-13 (CPP200012113000040). |
197 |
See FBIS, May 11, 2000, reprint of “Jiang Zemin’s Son, Taiwan Company Team Up on Chip Venture,” Taipei Times (Internet Version) May 11, 2000, no page citation (CPP20000511000120). |
198 |
FBIS, November 20, 2000, translation of Pai Te-hua, Wang Chung-ning and Wang Shih-Chi, “Construction of Taiwan-Funded Microchip Plant Begins in Shanghai,” Chung-Kuo Shih-Puo |
-
Taiwan’s Advance Device Technology has already built a 6-inch wafer fab in southeastern China.199
-
In October 2000 it was reported that Taiwan’s Wafer Works Corporation would set up a foundry in connection with China’s Beijing Oriental Electronics Group, using second-hand equipment.200
In December 2000 TSMC Chairman Morris Chang announced plans to visit Beijing, where he was to meet with several delegations organized by the Chinese government—fueling speculation that TSMC was planning to build a foundry on the mainland. A company spokeswoman said the real purpose of his trip was to participate in a bridge tournament. “Our chairman is an avid bridge player,” she said.201
Chinese engineers and specialists will staff the Taiwanese plants in China and may also help to operate semiconductor facilities in Taiwan, where multiple institutional structures exist to diffuse advanced technology.202 Although the Taiwanese are determined never to surrender “the high end” to China, it is unclear how they can prevent diffusion of advanced technology to China—their own as well as that absorbed from their foreign partners and customers. This diffusion occurs through the mobility of personnel, particularly as the boundaries between the two industries become increasingly blurred.
One U.S. Taipei-based semiconductor executive notes that a decade ago, many Taiwanese managers were working for U.S. companies. Many have since migrated to Taiwan to pursue their own business opportunities. The results of this are by now well known. In a similar manner Chinese engineers working for Taiwanese and other foreign-invested firms on the mainland or for Taiwanese firms in Taiwan will migrate just like the Taiwanese managers. The PRC government is trying to accelerate this process.
Taiwan has always prohibited direct trade and investment with the mainland, and its stated policy toward the relaxation of restrictions on direct investment in
199 |
“Taiwan’s IC Makers Eye China—May Contract Low Cost Plants There, but the Island’s Government is Balking,” CMP Media Inc. at <http://www.ebonline.com>. |
200 |
See FBIS, October 3, 2000, “Taiwan Company to Set Up 6-Inch Chip Foundry in China,” Taipei Times (Internet Version), October 3, 2000, no page citation (CPP20001003000135) |
201 |
See FBIS, December 11, 2000, reprint of “Taiwan Semiconductor Chair Going to China for ‘Bridge Tournament,’” Taipei Times, December 9, 2000, no page citation (CPP20001211000132). |
202 |
“Taiwan Legislator Warns of High Tech Exodus,” South China Morning Post, July 6, 2000, online report, <Nikkei BP AsiaBizTech http://www.nikkeibp.asiabiztech.com/archive/onnet/200007onnet.html>. |
the mainland remains one summarized by the slogan “no haste, be patient.”203 However, despite official prohibitions the government has supported domestic firms’ investments in China on a controlled basis, providing legal advice and tax benefits to firms making desired investments, but imposing fines and other penalties on firms making “undesirable” investments. Divisions exist within the government and the Taiwanese information technology sector over the appropriate scale and pace of future Taiwanese investments on the mainland.204 Despite these constraints the migration of Taiwanese information technology manufacturing, including the semiconductor industry, to the mainland may already be unstoppable regardless of the policy Taiwan’s government chooses to adopt.
The fact is the new government is powerless to stop a new wave of hi-tech firms investing on the mainland...about 30,000 Taiwan firms have invested more than US$40 billion on the mainland in the past decade, the vast majority of them routed through Hong Kong [and third countries] to skirt a ban on direct investment.205
DIRECTIONS FOR U.S. POLICY
Government intervention in the global semiconductor industry poses complex challenges for the United States. U.S. economic doctrine opposes government intervention intended to produce specific commercial outcomes, but with a few exceptions during the past several decades the United States has not been able to persuade other countries of the wisdom of leaving the evolution of strategic industries like semiconductors to the vagaries of the market.
Instead, beginning in the mid-1980s the U.S. government was drawn into a series of limited market interventions to counter the adverse effects of foreign government measures. In semiconductors these include the Semiconductor Trade Agreement (STA),206 the formation of SEMATECH, and the development of the
long-range industry-government-academia partnership embodied in the Semiconductor Roadmap.
Ironically, as these and other manifestations of government support for the U.S. industry are being phased out, foreign countries are emulating them. These countries see the enhanced role of government as an important element in a strategy to challenge U.S. leadership in information technology in general and microelectronics in particular.
Should We or Shouldn’t We?
It is not within the scope of this paper to address the question of whether the U.S. government should provide support to the U.S. commercial semiconductor industry—such as through financing for capital investment or support for applied research aimed at producing a specific products. Any major-scale proposal of such an initiative would be so controversial that it is unclear whether it could be implemented. Rather—in light of the major technical challenges facing the industry and the scale and increases in interventions in this industry by governments abroad—a more practical question relates to what positive role, if any, the U.S. government can play. Such a role might be a part of a broader U.S. response— one that does not contravene conventional U.S. views about the appropriate place of government in the economy.
The role of the state in the economy is perhaps a less controversial subject today than it was in the era of Hamilton and Jefferson, but it is not a settled question. A rough consensus has emerged over the past century—articulated first by Theodore Roosevelt and reiterated by successive generations of U.S. leaders—that in economic affairs the government should serve as a sort of neutral referee. As such, it would act where necessary to ensure that competition is “fair” and that the public is not victimized by unscrupulous commercial practices. At the same time, this consensus holds that the government should not intervene to promote or protect the interests of individual competitors or sectors.207
In addition, it is widely accepted that the government should take certain affirmative steps to promote general economic welfare, such as: the sponsorship of roads, bridges, and other elements of the transportation and communications infrastructure; promotion of scientific advances; and measures to improve the quality and availability of education and training.208 In the latter half of the twentieth century it also became generally accepted that the government must take certain steps necessary to ensure that industries essential to the national defense exist and remain strong enough to meet the needs of U.S. military forces.
But beyond these limited areas in which a role for the state is generally acknowledged, the consensus unravels. While beleaguered U.S. companies and industries sometimes succeed in securing government assistance in the form of bailouts, import protection, special tax relief, and the like, such measures are almost always controversial and for that reason frequently short lived. An imperative of the global economy, however, is that U.S. preferences and practices be measured against the policies and practices of major and emerging competitors, not necessarily for emulation but for a careful assessment of impact and value.
Meeting Challenges
Arguably, the most serious challenges confronting the U.S. semiconductor industry today are in areas where the government can play an important and positive role without contravening generally accepted U.S. notions of the proper limits on the intrusion of the state in the economy. The brick wall confronting the
208 |
In the nineteenth century the federal and state governments substantially underwrote construction of railroads and in the twentieth century the construction of airports, highways, canals, and ports. Federal funding of research and development has led to such advances as atomic energy, the Internet, the Global Positioning System, lasers, solar-electric cells, storm windows, Teflon, communications satellites, jet aircraft, microwave ovens, genetic medicine, and a wide array of advanced materials and composites. Office of Science and Technology Policy, Fact Sheet on How Federal R&D Investments Drive the U.S. Economy, June 15, 2000, at <http://www-es.ucsd.edu/stpp/whouse(rp).htm. |
industry is the type of large technological hurdle that the U.S. government has previously helped industry surmount through support for basic science and pre-competitive R&D.
-
An increase in the volume of federal funding of programs such as MARCO and the Nanotechnology Initiative—basic research initiatives that have enjoyed strong bipartisan support—would be an important first step in attacking emerging challenges in microelectronics. Ensuring the availability of a trained, educated workforce is a core government responsibility. The workforce shortage confronting the industry can be partially addressed by providing additional resources to U.S. universities and incentives for students and faculty to enter and remain active in fields that are critical to the challenges confronting this industry—electrical engineering, physics, and chemistry.
-
The post-Cold War changes in defense and other research budgets must be clarified and redressed. Without the necessary additional funds the students will not be trained, and in any event, cannot be trained quickly.
-
Consequently, even in the face of reduced retention rates, U.S. immigration policy should be administered in a manner that facilitates the attraction of foreign talent to the United States.
The potential problems posed by foreign industrial policies are more complex. The United States possesses an array of trade remedies that can be invoked against certain defined “unfair” trade practices, but these tools have frequently proven crude and/or ineffective in the complex realm of global competition in technology-intensive industries. However, a very substantial proportion—perhaps most—of the foreign programs summarized in this survey do not constitute “unfair” trade practices as defined by U.S. law and are not proscribed by any existing multilateral rules. They simply exceed abiding U.S. notions of the appropriate role of the state in the economy.
Learning From Past Success
The success of the U.S. semiconductor industry during the past 15 years reflects, in substantial part, a series of improvisations by the government and the industry working together to hammer out solutions to challenges arising out of foreign industrial policies without a fundamental departure from U.S. economic values.
-
The Semiconductor Trade Agreement, while controversial and in some respects unique, was nevertheless consistent with a long line of comparable limited actions by the U.S. government designed to open markets and promote competition—and it was phased out when market-based competition in the industry was restored.
-
SEMATECH was created to address specific national defense-related concerns arising out of the growing dependency of U.S. weapons systems on critical components for which a secure domestic production base was regarded as essential.
The flowering of the joint industry-government-university effort, reflected in the Semiconductor Roadmap, was in part a response to the strategic promotional efforts under way in Japan and Europe. More broadly they reflected a recognition that a cooperative approach is required to sustain the tremendous benefits offered by this rapidly evolving industry.
More improvisation and greater cooperation will be required in the coming decade.
POSTSCRIPT
Since this paper was written in early 2001, a number of developments have occurred which deserve to be noted here. Most dramatically, China is beginning to emerge rapidly as a major production base for semiconductors. A massive influx of foreign investment and skilled manpower, predominantly Taiwanese, is resulting in the establishment of new semiconductor foundries in the Yangtze Delta region and in Beijing. At this writing (at the end of 2002) two of these new foundries are operational, six more are under construction or will enter the construction phase by early 2003, and at least 11 more are planned.209 This expansion reflects a new Chinese government promotional effort designed to replicate Taiwan’s success in microelectronics on a much larger scale in China, drawing heavily on Taiwanese and other foreign capital, management and technology.210 China’s new policy measures closely resemble those utilized by Taiwan,
209 |
In September 2002, Shanghai-based Semiconductor Manufacturing Corp. (SMIC) had two fabs operational and planned at least two more, and Grace Semiconductor Manufacturing International, also based in Shanghai, had two fabs under construction and two more planned (interviews with senior executives at SMIC and Grace, Shanghai Zhangjiung Science & Technology Park, September 2002). In Suzhou, He Jian Technology Corporation, widely reported to be affiliated with Taiwan’s UMC, had one fab under construction and five more planned (interview with officials of the Suzhou Industrial Park, Suzhou, September 2002). Taiwan’s TSMC had announced plans to build at least one fab in Songjiang and has acquired sufficient land for additional facilities (interview with official of Shanghai Songjiang Industrial Zone, September 2002). In Beijing, the Beijing Semiconductor Manufacturing Corp. had two fabs in the early stages of construction and at least three more planned (interview with officials of the Beijing Economic Development Area, September 2002). |
210 |
The principal Chinese policies are spelled out in the Tenth Five Year Plan (2001-2005) — Information Industry, http://www.trp.hku.hk/infofile/china/2002/10-5-yr-plan.pdf, and Circular 18 of June 24, 2000, Several Policies for Encouraging the Development of Software Industry and Integrated Circuit Industry, published in Beijing Xinhua Domestic Service, 04:49 GMT, July 1, 2000. The |
including the establishment of science-based industrial parks, tax-free treatment of semiconductor enterprises, passive government equity investments in majority privately held semiconductor companies, and preferential financing by government banks. In addition, China is providing a protected market for semiconductors manufactured domestically, levying an effective value-added tax of 3 percent for locally made semiconductors versus 17 percent for imported devices.211
The abrupt rise of China as a significant site for semiconductor manufacturing reflects the erosion of several longstanding impediments to China’s development in this sector. With the end of the Cold War, the export control regime administered by Western countries restricting the flow of semiconductor manufacturing equipment and technology to China has loosened substantially, and the new mainland foundries have experienced little difficulty in acquiring equipment and process technology to produce 8-inch wafers using 0.18- to 0.25-micron design rules.212 Taiwan’s legal constraints on investment and technology transfer to the mainland have been relaxed, and many Taiwanese are circumventing such restrictions in any case.213 Finally, in the wake of its entry into the WTO, the Chinese government has abandoned or is phasing out a number of longstanding policies which have deterred inward foreign investment in microelectronics, such
as the general prohibition on 100 percent foreign-owned enterprises and stringent restrictions on “trading rights.”214
Taiwan’s government planners are seeking to adjust to the growing migration of the island’s semiconductor manufacturing operations to China by implementing a new “roots in Taiwan” strategy in microelectronics. This approach accepts the loss of much of Taiwan’s commodity manufacturing functions and production-line jobs to China but attempts to retain in Taiwan the most advanced manufacturing, design, and distribution functions.215 Specifically, the government hopes to sustain a high concentration of 12-inch wafer fabs on the island, to enhance Taiwan’s capabilities with respect to systems-on-a-chip, and to improve Taiwan’s position in upstream (materials and semiconductor manufacturing equipment) and downstream (assembly, test, packaging) functions. Reflecting this new emphasis, government promotional initiatives are curtailing (although not eliminating) financing for wafer fabs and increasing aid for IC design and upstream and downstream microelectronics functions. Revised tax incentives place greater emphasis on R&D, training, and maintaining “operational headquarters” in Taiwan.
Japan has launched a number of government-supported industry-government R&D projects in 2001-2002. The Millennium Research for Advanced Information Technology (MIRAI) was initiated by METI in 2001 to develop next-generation semiconductor materials and process technologies, such as measuring and mask technology for 50 nm-generation devices.216 In 2002, METI launched a 5-year industry-government R&D project to develop extreme-ultraviolet (EUV) lithography for 50-nm device manufacturing in conjunction with an association of 10 Japanese device and lithography equipment purchasers.217 In July 2000, 11 Japanese semiconductor manufacturers established a new R&D company, Ad
214 |
World Trade Organization, Accession of the People’s Republic of China (Decision of November 2001), WT/L/432 (November 23, 2001), Parts I.5, I.7. |
215 |
Taiwan Ministry of Economic Affairs, Promotional Strategies and Measures (2002); Government Information Office Release, Liberalization of Mainland-Bound Investment of Silicon Wafer Plants (2002). |
216 |
Government funding for this seven-year project was set at 3.8 billion yen for the first year. The project is being operated jointly by ASET and METI’s new semiconductor R&D organization, Advanced Semiconductor Research Center (“ASRC”) in the Tsukuba Super Clean Room. MIRAI website, <http://unit.asit.go.jp/asrc/mirai/index.htm>; Handotai Kojo Handobukku (December 5, 2001), pp. 4-5. |
217 |
The producers have formed the Extreme Ultraviolet Lithography System Development Association (“EUVA”) to undertake the project. First-year government funding was set at 1.09 billion yen. Japan Patent Office General Affairs Department Technology Research Division, Handotai Rokogijutsu Ni Kansaru Shutsugan Gijutsu Douko Chosa (May 10, 2001), p. 17; METI, Heisei Yonnendo Jisshi Hoshin (March 8k, 2002), p. 1; Handotai Sangyo Shimbun (January 16, 2002), p. 3. |
vanced SoC Platform Corporation (“ASPLA”) to standardize design and process technologies for systems-on-a-chip utilizing 90-nm design rules. METI reportedly will provide 31.5 billion yen for this effort, which will feature partnership with STARC and Selete. METI’s motivation for supporting this project is to create an “All Japan Foundry”—a standardized production line that can be used by all of Japan’s device makers.218 METI is prodding Japanese device makers — with some success—to consolidate their manufacturing divisions and specialize in design.219
218 |
ASPLA website, <http://www.aspla.com/jp>; Ekonomisuto (July 2, 2002), p. 20. |
219 |
“Next Generation Semiconductor Project — METI Tells Firms to Discard Own Plants,” Nikkei Sangyo Shimbun (June 12, 2002); FBIS, November 1, 2002, “Final Phase of LSI Industry Restructuring, Some Non-Winning Scenarios,” Nikkei Microdevices (November 2002) (JPP2002112000009). |