8
Issues for U.S. Policy: Global Technological Stratification and U.S. Technological Capabilities

UPSTREAM TRENDS: THE SEMICONDUCTOR EQUIPMENT INDUSTRY

The development and growth of the semiconductor equipment industry over the past several decades give evidence of the technological stratification of the semiconductor industry. This stratification is due to the specialization of technical expertise and to the U.S. investment community's increasing preference for focus over integration. In the early days of the industry, companies were virtually self-sufficient. In the 1960s, Texas Instruments, the largest semiconductor company at the time, grew its own material, made its own packages, and designed and built its own equipment. By the late 1960s a few equipment manufacturers had started to build systems, and firms would design and specify what they wanted. In the 1970s it became clear that semiconductor firms could not remain technologically self-sufficient, as equipment makers started to suggest equipment that the firms were not intimately involved in designing such as ion implanters at National Semiconductor. The role of equipment makers became more important as the industry shifted from being labor to capital intensive. In 1967 a competitive wafer fabrication line cost about $1 million, whereas a world-class volume fabrication facility costs well over $100 million today.

By the beginning of the 1990s the semiconductor industry had come to depend on sophisticated technology in the areas of wafer manufacturing (lithography, deposition, packaging), chip design—computer-aided design



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



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

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

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy 8 Issues for U.S. Policy: Global Technological Stratification and U.S. Technological Capabilities UPSTREAM TRENDS: THE SEMICONDUCTOR EQUIPMENT INDUSTRY The development and growth of the semiconductor equipment industry over the past several decades give evidence of the technological stratification of the semiconductor industry. This stratification is due to the specialization of technical expertise and to the U.S. investment community's increasing preference for focus over integration. In the early days of the industry, companies were virtually self-sufficient. In the 1960s, Texas Instruments, the largest semiconductor company at the time, grew its own material, made its own packages, and designed and built its own equipment. By the late 1960s a few equipment manufacturers had started to build systems, and firms would design and specify what they wanted. In the 1970s it became clear that semiconductor firms could not remain technologically self-sufficient, as equipment makers started to suggest equipment that the firms were not intimately involved in designing such as ion implanters at National Semiconductor. The role of equipment makers became more important as the industry shifted from being labor to capital intensive. In 1967 a competitive wafer fabrication line cost about $1 million, whereas a world-class volume fabrication facility costs well over $100 million today. By the beginning of the 1990s the semiconductor industry had come to depend on sophisticated technology in the areas of wafer manufacturing (lithography, deposition, packaging), chip design—computer-aided design

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy and engineering (CAD/CAE), and other tools to design the electronic and process properties of the devices—and design of the systems architecture embodied in chips. Although some might argue that Hitachi possesses most of this capability, the Texas Instruments of the 1960s is gone forever, and no company can push technology forward on its own. Semiconductor equipment companies play an important role in the overall technical progress of the industry. Technical stratification has been partially spurred by the preferences of investors. There was a flurry of investment in supplier companies in the mid-1980s, and some good companies such as Novellus were born, but interest has flagged because development cycles are typically long by venture-investing standards and because the condition of the U.S. semiconductor industry has made it a worrisome customer target. Presently, even U.S. companies that offer significant improvements on processing technology have trouble obtaining financing. One example is Hampshire Instruments, an x-ray lithography company that has faced continuing financial challenges even as technical challenges have been conquered. Thus, in general, even new U.S. equipment companies with superior technology are not current investment targets because the capital required is high, the development cycles are long, the U.S. customer base is weak, and leading Japanese customers have their own suppliers. A significant portion of the technology required for future semiconductors is not gaining new company investment support, which leaves technology development to existing U.S. companies and foreign firms. The competitive and technical landscape differs among various segments of the equipment industry. First, in lithography equipment, where technology is tied to resist and U.S. companies invented the concepts for all leading machines, the United States has the technology for the next generation—both x-ray and "phase shift mask" with eximer lasers (today's steppers). Yet there are no commercially "leading" companies resident in the United States, the leaders being Nikon, Canon, and ASM Lithography. American entrants GCA, SVG and Ultratech are followers. In deposition equipment, Applied Materials, the strongest U.S. equipment company, has world-class technology that must be demonstrated before it can sell its products. It has acquired a strong position in Japan through hard, long effort but has growing competition and a customer base that understands its technology to some degree in order to use it. In years past, Applied Materials would get advance payments from customers, as did other equipment suppliers, and then work with them to develop the equipment and processes necessary to build the next generation of devices. Today IBM and, to some degree, SEMATECH will support the development of a machine in this way, but the balance of the load is carried by the equipment maker.

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy Similar situations exist for companies in the automation and test equipment fields. They must perform equipment and process development with their own capital and then transfer their process technology to make a sale. The sales of equipment companies necessarily involve the transfer of process technology. The processes by which component materials are developed are really no different. Makers of silicon wafers, exposure resists, masks, packages, and CAD/CAE tools must develop products with their own resources. Venture investment is not attracted to the area, again because of the general weakness of the U.S. semiconductor industry and the concern that any success in Japan will be short-lived since the technology transferred through a sale will very likely leak to local suppliers. Although it is hard to pinpoint aggregate trends in U.S.-Japan alliances in equipment and other upstream sectors supplying the semiconductor industry, there are indications that the Japanese semiconductor equipment industry has been fairly active in U.S. investment recently.32 A key question is whether the U.S.-owned portion of the semiconductor equipment industry can remain viable in the face of the conditions outlined above. If the answer is no, the next question is whether a predominantly Japanese-owned equipment-making industry will be adequate to support the maintenance of semiconductor process design capability in the United States.33 DOWNSTREAM TRENDS: SYSTEMS, COMPONENTS, AND PROPRIETARY ARCHITECTURES On August 12, 1981, IBM Corporation launched its first personal computer, creating an industrial miniboom for semiconductors, disk drives, software, and countless other products. In executing its strategy, IBM chose to ally itself with smaller companies for critical components, notably Intel and Microsoft. Spawned by the microprocessor, the personal computer was a triumph of miniaturization that transformed the way many people work and eroded 32   See Phyllis A. Genther and Donald H. Dalton, "Japanese-Affiliated Electronics Companies: Implications for U.S. Technology Development," NTIS, March 1992, pp. 7–11. According to Japan Economic Institute and U.S. Department of Commerce data the number of Japanese-owned semiconductor manufacturing equipment plants increased from 13 in 1989 to 23 in 1990, while employment increased from 1,745 to 1,960. This "indicates growth in the supplier network of Japanese-affiliated companies . . . and suggests the replication of keiretsu structure linkages in the United States." 33   Concerns have been raised (and denied by Japanese spokespersons) that Japanese semiconductor equipment makers may delay or deny sales of their most advanced products to U.S. companies. See U.S. General Accounting Office, International Trade: U.S. Business Access to Certain Foreign State-of-the-Art Technology (Washington, D.C., September 1991).

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy the sales of much larger machines. In fact, the opportunity in personal computers was so great that a myriad of companies sprang up around the world. In less than 10 years, the total worldwide market grew to $150 billion annually. Times have changed, however. With the personal computer industry maturing, the growth of sales of personal computers has slowed—from 34 percent in 1988 to 10 percent in 1990. This is taking a toll on computer makers as well as on the semiconductor manufacturers who supply them. Today, faced with price cutting, moribund profits, and an increasingly difficult future in a saturated personal computer market, the largest players in the computer industry are attempting to redefine the global market battlefield through new partnerships and strategic alliances. These alliances, along with major technological changes, will have a profound effect in determining who will be the winners and losers over the next decade in the semiconductor, software, and computer businesses. To be sure, the new personal computer alliances have vulnerabilities and face a number of challenges. Focusing the resources and coordinating the strategies of a large number of companies on a common goal is fraught with difficulties, and new technical developments or counter-strategies can raise obstacles to collaboration or rob an alliance of its rationale. Whether or not the alliances described below ultimately meet their objectives, market trends may force companies to contemplate and launch other, similar efforts. The two alliances that have attracted the most attention are the ACE consortium and the Apple-IBM alliance. ACE was announced on April 9, 1991, and was launched by 21 industry-leading hardware and software companies. Led by Compaq Computer Corporation the alliance announced a set of specifications for an advanced computing environment. The ACE consortium's ambitious goal is the creation of a hardware and software system that can be used in everything from laptops to mainframes. The group intends to generate compatible computers that will be equal in popularity to IBM's personal computer. The new machines will use one of two operating systems and two microprocessor chips. The two hardware architectures are based on the MIPS R3000/R4000 RISC family and the Intel 80X86 family of microprocessors. In the months following the ACE announcement, the consortium quadrupled in size to include more than 80 computer companies from the United States, Japan, and Europe. Successful implementation of the ACE initiative would position its members to define industry direction by thwarting rivals in the commercial sector and gaining access to markets for networked computers. If ACE succeeds, the big potential winner in the semiconductor industry would be MIPS, which licensed its RISC processor technology to several U.S. and Japanese semiconductor manufacturers. Two in particular, LSI

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy Logic Corporation and Nippon Electric Corporation (NEC), should benefit most from being suppliers to the ACE alliance of computer original equipment manufacturers (OEM). Furthermore, NEC's ability to supply large quantities of high-density DRAMs to the alliance members would help ensure a strong challenge in the industry. The big potential loser in the industry would be Intel, which has enjoyed a virtual microprocessor monopoly in personal computers. If ACE succeeds, software portability across a wide range of hardware devices from a multitude of vendors would force Intel to offer more aggressive price-performance characteristics for its architecture. During the year after ACE was formed, the consortium has run into a number of obstacles, although the members remain publicly committed to its goals. For example, Intel has accelerated the commercialization of its next generation microprocessors in response to the ACE push for a RISC personal computer standard and to the introduction of MPUs compatible with its X86 line by Advanced Micro Devices and others. Compaq decided that the performance of Intel's new products will obviate some of the rationale for RISC personal computers, and dropped out of ACE in April 1992.34In addition, MIPS was acquired by workstation maker Silicon Graphics in early 1992, leading some analysts to question whether all ACE computer makers would have equal access to MIPS designs. In the spring of 1992, it was unclear whether the leading companies in ACE would be able to provide strong enough leadership to push the consortium to a timely achievement of its objectives. A second significant alliance was launched on July 5, 1991, when Apple Computer and IBM Corporation joined forces in a wide-ranging pact to jointly develop powerful, easy-to-use desktop computer systems. Formerly fierce rivals, the two companies will cooperate in the development of next-generation computer operating software and advanced RISC hardware. Furthermore, to pave the way for its joint venture with Apple, which is known as Taligent, IBM forged a series of pacts with software publishers, computer companies, and chip makers in which member companies will develop software and hardware to take advantage of the systems that Apple and IBM develop together. Within weeks of announcement of the Apple-IBM partnership, Motorola and Siemens AG were invited to participate as key suppliers. Motorola will second source the IBM-designed RS6000 microprocessor (also known as the Power PC chip), and Siemens will produce the world's most advanced computer memory chips (at first the 16 megabit DRAM). 34   Peter Lewis, ''Whither the ACE Consortium? Maybe Nowhere," The New York Times, May 10, 1992, Section 3, p. 10.

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy The bold moves by these two computer giants are likely to change the ground rules for competition by forcing a redefinition of the personal computer. Codesigning technology is only the veneer on their strategy; the real purpose of the alliance is to preserve and enhance their enormous marketing clout. By deciding not to compete with each other, they have sent a strong message to Compaq, Sun Microsystems, AT&T/NCR, and numerous Japanese companies: recalibrate your strategic aspirations or face competing head-to-head against Apple-IBM. The big potential winner in the semiconductor industry is Motorola. Long a friend of Apple and IBM, and far behind Intel in microprocessor market share, Motorola may do very well as the designated microprocessor supplier to the alliance. Another potential winner, as a group, would be Europe's struggling chip manufacturers. Bringing German electronics giant Siemens AG into the Apple-IBM camp will have two major effects. First, by sharing the burden of building a $700 million advanced chip factory in France, and sharing proven 16 megabit DRAM production technology, the partners build a base for their efforts to develop a production worthy 64 megabit DRAM. Second, because the agreement with Siemens allows for additional European partners, European computer makers will be less likely to form their own alliances in response to a perceived threat from Apple-IBM. Whether or not Taligent can deliver what its parents intend for it is still unclear. Since the announcement of the alliance, both Apple and IBM have made moves that appear to be aimed at hedging their bets on Taligent. Still, the partners have devoted considerable resources to the venture. The recent trends in alliances among software, semiconductor, and computer companies may well signal a shift in the balance of power in the computing industry. The key unsettled question in the 1990s is whether any alliance can protect its members from increasingly fierce competition. Some recent trends need to be examined to answer this question. Historically in the electronics industry, as semiconductor technology becomes cheaper the prices of the personal computer or workstations using memories and microprocessors decrease. Lower prices have led to higher sales, fueling the development of entirely new applications. This remarkable relationship—cheaper memory leads to cheaper computers, which then expands the market for software—will be severely challenged in the next few years. There is wide agreement that software represents perhaps the key battlefield for U.S. computer companies in the 1990s. The United States dominates the world's software market share at a time when software is becoming an increasing portion of value-added information technology. In 1990, the U.S. world market share was approximately 75 percent in software, 65 percent in computers, and 40 percent in semiconductors.

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy In the area of software, techniques such as object-oriented programming hold the promise for U.S. companies to save tremendous amounts of time and effort in creating programs from ready-made sections of computer code. On the other hand, U.S. companies worry about a new trend of several large Japanese computer companies bringing mass production techniques to writing programs.35 Furthermore, what the Japanese companies cannot make, they can usually buy because of their capital resources. This possibility poses a severe challenge to current U.S. dominance in software. Another area of concern for the U.S. software industry is that as applications become increasingly more powerful, they require vastly more semiconductor memory. This is especially true in computer workstations, which make sophisticated use of high-resolution graphics and digital video. The problem facing the U.S. companies is that memory production today is dominated by Japanese, and increasingly, Korean companies. Furthermore, new memory technology is, for the first time in history, becoming more costly. The real problem is that whenever memory becomes more expensive, computers become more expensive, thereby restricting the market for software. So the problem is two-pronged for U.S.-based companies in the information industries. First, when demand is growing Japanese companies may be able to increase the price of memories and thereby slow down U.S. growth in strategic segments. Second, a slow down in growth gives Japanese companies time to catch-up in software technology and thereby challenge U.S. dominance. This is one reason that both the ACE consortium and the Apple-IBM alliance included allies in semiconductor memory production—to attempt to thwart a possible restriction of memory supplies. Another concern of U.S. companies is the trend toward miniaturization, a trend that plays to foreign strengths. In the high-growth areas of laptop and notebook computers, Japan supplies virtually all the display screens and commands a large share of the market for portable computers. Even in the new area of pen-based computers, most machines not only are manufactured abroad but require high-density memories made largely by Japan. But U.S. software strength also presents opportunities in this new environment. For example, in early 1992 a number of alliances were announced that aim to develop new products that would lie at the intersection of the computer, consumer electronics, and telecommunications equipment markets. Some analysts predict that U.S. software strength will allow computer companies such as Apple to reestablish a strong U.S. presence in consumer electronics. Still, it appears that in all of these nascent alliances—such as Apple's partnership with Sharp and AT&T's linkage with Matsushita—it is necessary 35   Michael A. Cusumano, Japan's Software Factories: A Challenge to U.S. Management (New York: Oxford University Press, 1991).

OCR for page 59
U.S.-Japan Strategic Alliances in the Semiconductor Industry: Technology Transfer, Competition, and Public Policy for U.S. companies to turn to Japanese partners with consumer electronics marketing and component manufacturing capabilities.36 If the battlefield in the 1990s is density memories and sophisticated software, where will microprocessors fit? To be sure, the performance of microprocessors is advancing, largely as competitors pursue the RISC-design technology. Furthermore, added power is needed to run increasingly powerful applications. Yet at the same time, more and more software is being written to run independently of any particular microprocessor. This trend, if it continues, will allow any computer maker to choose any cheap microprocessor, as long as it has the horsepower to run the intended applications. Furthermore, as microprocessors enter an era of commodity pricing, a severe challenge could be posed to the profitability and future of microprocessor vendors that have relied on proprietary architectures to maintain market share. Thus, the personal computer and semiconductor industries have entered a crucial period of alignments and realignments based on global alliances. What emerges from the crucible of alliance-bloc competition will be fascinating to observe. Certainly both the processes and the outcome will shape the structure and dynamics of the personal computer and semiconductor industries (just as RISC alliances will shape the structure and dynamics of the workstation industry). 36   Bob Johnstone. "Future at Hand." Far Eastern Economic Review, April 30, 1992, p. 74.