Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Hard Disk Drives DAVID McKENDRICK University of California, San Diego The hard disk drive (HDD) industry represents an interesting exception to received wisdom about American industrial competitiveness. Until recently, scholars have been pessimistic about the competitive prospects of much of U.S. industry, observing that "lilt is too late for the United States to regain its position as the exemplar of best practice in the world" (Kogut, 1993~. Florida and Kenney (1990) concluded that America may be good at generating new industries but is bad at sustaining them as they become more mature. American industry in gen- eral was said to have "attitudinal and organizational weaknesses" leading to "shortcomings in the quality and innovativeness of the nation's products" (Der- touzos et al., 1989~. Yet, the experience of the disk drive industry suggests that these characterizations of American industry need not be its paradigmatic form. Like many industries that emerged in the twentieth century, the disk drive indus- try was dominated by American firms during its early years. Unlike other indus- tries, however, the United States never relinquished its leadership. American companies hold more than 85 percent of the global market, an even greater share than they did in the late 1970s. Why has the United States been so consistently successful in this industry? This paper argues that the industry's globalization was an important factor in iThis research was supported by the Alfred P. Sloan Foundation, grant numbers 95-6-13 and 97-1- 10. The author is grateful to Allen Hicken, John Richards, Peter Gourevitch, Roger Bohn, Frank Mayadas, and David Mowery for careful and insightful comments on an earlier draft. For assistance with data collection and compilation, the author thanks Allen Hicken. He also thank James Porter, president of Disk/Trend, Inc., who not only reviewed this paper but also has generously shared his data, time, and knowledge about the disk drive industry. Mark Geenen, president of TrendFOCUS, Inc., kindly provided data on the media and heads segments of the industry. 287
288 U.S. INDUSTRYIN2000 sustaining the American competitive advantage. This is not to say that other, less global factors often invoked to explain the success of certain nations in particular industries do not apply to the disk drive industry. Home market demand, form of industrial organization, innovative capabilities, and the role of institutions, such as universities, government agencies, business associations, and other regional and national entities all play or have played some role in the industry's evolu- tion. The principal point here is that foreign investment not only complements innovation, style of industrial organization, and the favors conferred by historical chance, but it is also critical to sustaining industrial performance. The disk drive industry offers a fascinating context for charting globalization as well as industrial evolution more broadly. American disk drive firms in par- ticular have accumulated the organizational skills necessary for managing the geographic separation of R&D, production, and distribution to achieve econo- mies of location. This kind of dispersion has been increasing in other industries as well, but, because the HDD industry is farther ahead than most in globalizing its activities, its experiences may provide a glimpse of what may come for other parts of the American economy and a touchstone for the maintenance of indus- trial leadership. NATIONAL EMBEDDEDNESS AND PATH DEPENDENCE The home market confers advantage upon national firms, and the success of customer industries confers success upon their suppliers. This national embed- dedness is especially true for nascent industries. An industry emerges through the cumulative interactions of entrepreneurs and organizations. Interdependen- cies are established through the sharing of information and resources, and clus- ters of firms begin to form (Van de Ven and Garud, 1989~. Clusters are largely national or regional phenomena, with firms serving national customers before growing through foreign trade and investment (Chandler, 1990~. Ties among national firms persist as they expand into international markets. In his study of competitive advantage, Porter (1990: 138) finds that "a group of internationally successful domestic firms, selling worldwide, channelLed] global demand to the domestic supplier industry." The Origins of the Disk Drive Industry One possible explanation for the success of the American HDD industry, therefore, is American success in the computer industry. This explanation seems reasonable on the face of it. At the time IBM shipped the first rigid disk drive in 1956, the United States was already the world's dominant computer producer and exporter. Although Europe contributed enormously to the technical development of the early computer industry, American firms led the world in computer instal- lations, and many of these same firms developed their own HDDs. General Elec
HARD DISK DRIVES 289 trio, Control Data, Burroughs, and Digital Equipment followed IBM's entry into HDDs in the 1960s. Some independent companies, such as Bryant Computer Products and Data Products, also emerged in the early 1960s to develop disk drives for sale to computer manufacturers that had not yet made their own, nota- bly Sylvania, RCA, Honeywell, and Univac. In the late 1960s, after IBM secured its position as the clearly dominant mainframe maker, a new wave of independent companies emerged to make disk drives that were "plug compatible" with IBM systems: Memorex, Potter Instrument, Marshall Laboratories, and Information Storage Systems. Without incurring IBM' s R&D expenses, the plug compatible companies were able to offer disk drives identical to or better than IBM's at a much lower price. Plug compatibility was not limited to IBM systems but ex- tended to systems made by other computer manufacturers as well. A parallel trend, but on a smaller scale, was evident in Japan and Europe. In Japan, the principal computer companies made their own disk drives: NEC, Fujitsu, Hitachi, and Toshiba all entered in the mid-to-late 1960s. Only in the 1970s did Japanese companies attempt to market disk drives to non-Japanese customers in the U.S. market; until then the size of the market for Japanese com- puters limited the market for their disk drives. A smaller domestic market also meant fewer independent Japanese disk drive companies entered in the 1970s as alternative sources of supply; the principal ones were Mitsubishi and Hokushin Electric Works. In Europe Siemens and Philips made disk drives for their own computer systems, while Data Recording Instruments and BASE produced for the original equipment market (OEM). Data Recording Instruments was Europe's first firm to ship HDDs in 1968. Honeywell-Bull (later CII-Honeywell Bull and then Bull Peripherals) engaged in both captive and OEM production. In Eastern Europe COMECON organized the computer industry in such a way that DZU of Bulgaria was designated as the principal disk drive supplier for all computers in the region and became the most vertically integrated producer in the world. Only in rare cases did European disk drives find their way into American or Japanese com- puter systems. Thus, throughout the 1960s and 1970s, the relative positions of the U.S., Japanese, and European disk drive industries could be explained by incorporation of their products into the systems manufactured by their respective national or regional computer industries.2 During the 1970s captive production remained the largest channel for disk drives, though the relative importance of the original equipment market grew. Led by Control Data, Diablo Systems, CalComp, and Memorex, the OEM segment reached $631 million in sales revenues in 1979 but was still well below the $2.8 billion associated with captive production (Disk/ 2A major exception to this general tendency was the success of Control Data in selling to European computer manufacturers. It claimed the bulk of the world's shipments of "noncaptive" drives in the 1 960s and still almost half by the late 1970s. IBM's disk drives were solely for IBM computers.
290 U.S. INDUSTRYIN2000 Trend, 1980~. In 1979, American firms had 81.1 percent of the global HDD market, Japan 14.3 percent, and Europe the remainder. Between them, IBM and Control Data controlled just short of 40 percent of the market. The Personal Computer and the Desktop Disk Drive Up to this point the story conforms strongly to an explanation of competitive advantage through path dependence and increasing returns; the large U.S. market for mainframes, and later minicomputers, gave the American disk drive firms an unassailable long-term advantage. But it does not account for the divergence in the fortunes of the American disk drive and computer industries after 1980 when both came under greater global competitive pressures. For the computer industry, a watershed event was the debut of the IBM PC in 1981. The PC defined the dominant design in the industry for many years (Langlois, 1992; Anderson, 1995~. In addition to setting the standard for what a desktop computer should look like, it featured an open architecture that attracted the entry not only of some of IBM's established mainframe and minicomputer rivals but de novo start-ups that set out to manufacture IBM clones. Compaq and Dell became two of the most important American entrants, but more interesting are the many new clone makers that emerged outside the U.S., especially Taiwan, Korea, and Japan. Daewoo, Epson, Hyundai, Acer, and scores of other smaller companies collectively dispersed the production of computers. As a result, the global market share of U.S. computer makers steadily eroded during the 1980s and early l990s. The U.S. share of the worldwide computer market, including mainframes, fell from 88 percent in 1983 to around 56 percent by 1992. During the same period Japanese market share in the computer industry increased from 8 percent to 30 percent.3 The same open architecture that attracted the new clone manufacturers also stimulated entry into peripheral equipment. Where mainframe and minicomputer manufacturers made many of their own peripherals and components, the assem- blers of personal computers almost entirely outsourced their production. Japa- nese, Korean, and Taiwanese producers of keyboards, floppy disk drives, moni- tors, DRAMs, and motherboards displaced U.S. firms in peripherals and components even more dramatically than American companies had been dis- placed in the PC market. Given these trends, and the development of national clusters of computer- related capabilities in these countries, one might have expected other Asian com- panies to erode America's position in HDDs. Much the same competitive dy- namics faced the HDD industry as disk drives were adapted to fit into a PC. Drive sizes decreased from 14-inch and 8-inch diameters in the 1970s to 5.25 inches in 1980 and 3.5 inches in 1983. An explosion of some 100 new entrants, 3Global computer market shares were calculated from the Datamation 100 for various years.
HARD DISK DRIVES 291 intense competition, and shakeouts occurred between 1980 and 1996. By 1996 fewer firms made disk drives than at any time during the previous 20 years (Fig- ure 1~. The HDD landscape became littered with the graves of once prominent · · . Amencan compames. But whereas most of the rest of the Amencan computer peripherals industry has largely vanished, the Amencan HDD industry remained dominant in the face of competition from Asia and Europe.4 Although U.S. firms such as Pnam, Pra~netek, Conner Penpherals, Ministor, and Hewlett Packard exited, so did firms from other countnes. Mitsubishi, Matsushita, Rodime (the first firm to introduce the 3.5-inch disk dnve), Olivetti, BASE, Sony, Philips, and Siemens are among the formidable foreign companies unable to remain in the industry. Asian and European PC makers bought HDDs from U.S. firms. South Korea, for example, depended almost entirely on Amencan companies to meet the HDD requirements of its major PC exporters (MR, l991~. In Europe PC companies such as Amstrad also purchased Amencan disk dnves. Compared with the computer industry, the Amencan HDD industry held a roughly steady 75 percent of the global market throughout the 1980s and then increased its share to more than 80 percent by 1992 (Figure 2~. By 1995 U.S. global market share reached 85 percent, where it had been in the early 1970s. 80 70 60 to - 50 LL ° 40 a) ~ 30 no 20 10 ~ Total Number of Firms / Entries _~ \ - Exits CD ~ 0 0 0 ~ ~ ~ ~ Us (D ~ ~ ~ rot oo oo oo of co oO oO 0 ~ ~ ~ ~ 0 ~ ~ ~ ~ ~ Year FIGURE 1 Number of firms in the HDD industry, 1976-1996. rat oo 0 0 ~ cot ~ ~ us (D oo on oO ~ ~ ~ 0 cr ~ 0 cn ~ 0 ~ 0 4An important exception is the printer industry. Although the United States lost the impact printer market, it has a huge lead in laser printers.
292 90 80 70 60 50 a, 40 30 20 10 O U.S. INDUSTRYIN2000 i _ _ _ _ U.S Compute Japanese Computer - 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 thy) O ~CO ~lO By ~ 0 A) O ~N ~Jo 10 CC) r- Go of) ~a:) oc) Go Go co oo oo o) CD o) ~CO CO CO 01 ~CD ~ ~CD CD ~can ~ Year FIGURE 2 HDD and computer market shares, 1979-1996. Source: The Data Storage Industry Globalization Project Worldwide; Disk/Trend, Inc. Although the market share for American computer manufacturers fell throughout the 1980s, the American floppy drive industry practically disappeared, and the world increasingly turned to non-American suppliers of other computer components and peripherals, American firms continued to be the overwhelming source for HDDs. It is clear that the disk drive industry owed its birth to the American computer industry. But given the fate of other computer peripherals, a path-dependence argument is incomplete. The United States has been the inven- tor of other promising technologies that it relinquished to the Japanese. Why could it hold onto HDDs but not floppy disk drives, monitors, or optical storage devices? One possible factor is innovation. INNOVATIVE CAPACITY Many argue that Japanese and American firms have distinctly different meth- ods of innovation and innovative capabilities compared with their Western coun- terparts. Japanese firms are said to possess several organizational, incentive, and communication advantages that are conducive to innovation (Aoki and Rosen- berg, 1987; Aoki, 1990~. The Japanese system has its strengths and weaknesses relative to the stylized facts about American innovative capabilities. New prod- uct introductions tend to be faster; strengths in incremental product modifications based on careful engineering make Japanese firms better at innovations along a predictable technological trajectory (Imai et al., 1985; Odagiri and Goto, 1993;
HARD DISK DRIVES 293 Mansfield, 1988~. However, the links between scientific research and invention are weaker in Japan than they are in the United States. Although Japanese firms are adept at "the better known, closer-at-hand technologies," they are less suited to choose "bolder, riskier, and more visionary technologies" that lead to pivotal new products or process technologies (Okimoto and Nishi, 1994~. Given these characteristics, Japanese firms would be expected to perform better in situations where continuous incremental improvements, tight engineer- ing tolerances, and manufacturing strength are the bases of competitive success and less well in segments incorporating radical or unproven technologies that rely on more fundamental technical research. Innovations in Disk Drives Rapid product or process innovation is a necessary condition for competitive success in all high-technology industries. This is especially true for disk drives. Although technological advances in semiconductors have generally been credited for most of the price and performance improvements in computers, fewer people are aware that progress in disk drive speed and capacity kept pace. The amount of data that can be stored on a square inch of a disk grew almost 30 percent a year between 1957 and 1990; since then it has increased about 60 percent a year. Data- transfer rates have increased while average access times have fallen. Between 1980 and 1995, the price per megabyte of storage fell at an annual rate of 40 percent (CRN, 1997~. All of these advances were accomplished on increasingly smaller disk drives. Since the 1970s the disk drive's size, called form factor, has decreased, from 14 inches to 5.25 and 3.5 inches in the 1980s. These "architec- tural innovations" (Henderson and Clark, 1990) challenged the competencies of incumbent disk drive companies, and the inability of firms to make the transition to smaller form factors has been cited as a central reason behind firm failure (Christensen and Bower, 1996~. Can U.S. success in the industry be explained by the greater innovative capabilities of its firms? One American company, IBM, served as the technological fountainhead for the industry and continues to demonstrate remarkable technological leadership.5 As Table 1 shows, IBM established the industry and introduced many key inno- vations the first removable disk pack drive, the Winchester standard, the first drive with ferrite, thin film, and magneto-resistive heads, and the first 8-inch disk drive, which proliferated with the development of minicomputers. More than any other institution, IBM displayed engineering brilliance in overcoming critical technical constraints. The 1301 disk drive in particular pioneered in areas that led to follow-on improvements in storage density and access times. Nonetheless, the co-evolution of technology and competition in the HDD industry confounds the 5For a technical history of IBM's first 25 years of innovation in the industry, see Harker et al. (1981) and Stevens (1981).
294 TABLE 1 IBM "Firsts" in the HDD Industry U.S. INDUSTRYIN2000 Disk Number Diameter Firsts in HDD Model Year Megabytes of Disks (inches) First disk drive First disk drive with a~r-beanng heads First disk drive with removable disk pack First disk cartridge drive First disk pack drive First disk drive with ferrite core heads First track following servo system. First disk drive with low mass heads, IBM RAMAC IBM 1301 IBM 1311 IBM 2310 IBM 2311 IBM 2314 IBM 3330-1 IBM 3340 lubricated disks, sealed Winchester First disk drive with thin film heads IBM 3370 First 8-inch HDD IBM 3310 First disk drive to use MR heads and PRML IBM 681 1956 5 1962 28 1963 2.68 1965 1.024 1965 7.25 1966 29.2 1971 100 1973 70 4 1979 571.4 1979 64.5 1990 857 50 25 6 1 6 11 12 7 24 24 14 14 14 14 14 14 14 6 8 12 5.25 Source: Quantum Corporation web page based on information in Disk/Trend, Inc. conventional wisdom about differences in Japanese and American innovative ca- pability in two ways. First, Japanese firms have been stronger than theory would predict in techno- logically advanced new products. One measure of this strength is the shift to different form factors, each representing architecturally distinct product genera- tions. IBM introduced the 14-inch and 8-inch form factors, but since then young entrepreneurial firms, rather than older incumbents, have pioneered most archi- tectural innovations (Christensen and Rosenbloom, 1995~. Theory suggests that Japanese firms would lag behind their American com- petitors in the shift to new form factors. At first glance this seems to be the case. Eight of the first ten companies to introduce 5.25-inch drives were American, led by Seagate in July 1980. The other two were European Rodime and Olivetti. Three other firms shipped 5.25-inch drives by the end of 1981, but only one of those, Nippon Peripherals, was a Japanese firm. By the end of 1982, 13 more firms had begun shipping 5.25-inch drives, 7 of these were Japanese firms, in- cluding Fujitsu and Hitachi. In 1983, 14 more firms, 5 of which were Japanese, made the shift to 5.25-inch drives. Thus, from 1980 to 1993 only 13 of 41 HDD firms that shipped 5.25-inch drives were Japanese, and these firms were a year or more behind in introducing their drives. Among firms that still made disk drives at the end of 1996, however, the Japanese were quicker than most of their U.S. counterparts in moving to 5.25-inch drives. Fujitsu, Hitachi, and NEC all shipped 5.25-inch drives before or concurrently with Quantum, Maxtor, Micropolis, and IBM. Western Digital, another leader in 1996, did not make disk drives until 1988 when it acquired Tandon's HDD operations.
HARD DISK DRIVES 295 A similar story can be told regarding the shift to the 3.5-inch form factor. The European company Rodime was the first to ship 3.5-inch drives, in Septem- ber 1983, and the next three were American Microcomputer Memories, Micro- science International, and MiniScribe all of which shipped in 1984. The first Japanese firm to ship a 3.5-inch drive was Nippon Peripherals in February 1985. All five of these early innovators have since exited the industry. Of the HDD firms surviving at the end of 1996, the first to ship 3.5-inch drives was Hewlett Packard in March of 1985; but Hitachi, Fuji Electric, NEC, and Fujitsu followed close behind. More important, as occurred with the transition to 5.25-inch drives, these Japanese firms were quicker to make the shift to 3.5-inch drives than every U.S. firm that is a leader today. IBM did not introduce 3.5-inch drives until May 1986. Seagate first shipped 3.5-inch drives during the third quarter of 1987, the same date the Japanese firm Toshiba began shipping and one year after the Japa- nese firm Seiko Epson had begun shipping 3.5-inch drives. Quantum and Maxtor did not make the move into 3.5-inch drives until 1988, and Micropolis waited until 1991. In the shift to the 2.5-inch form factor the Japanese firm JVC was among the first movers. Other Japanese firms were no slower at adopting the new form factor than surviving American firms. Japanese firms have also been among the leaders in incorporating advanced technology in their disk drives, specifically the new, thin film magneto-resistive (MR) recording heads. MR heads are designed to read media with very high recording densities and are the reason that growth in are al density the amount of data that can be squeezed onto a given space of a disk had jumped up to a 60 percent annual rate since 1990. Unlike previous head technologies that function like small electromagnets, MR heads use a thin strip of magneto-resistive mate- rial deposited on the head that senses the strength of the magnetic patterns on the disk and creates corresponding electrical pulses. The MR strip cannot write data, however, and so a traditional thin film component must be placed on the head next to the MR strip (Quantum, 1997; EBN, 1996; BET, 1996~. Because the switch to MR heads requires corresponding changes in media and electronics technologies and because they are very difficult to make, many companies have been slow to commit resources to the new technology, choosing instead to try to increase capacity through conventional technologies. Stylized notions of American and Japanese innovative capabilities suggest that U.S. firms would be more likely to move first into smaller market segments with more sophisticated technology while abandoning to firms from other coun- tries the market segments dominated by older technology. American firms would thus be expected to lead the way into MR technology. Similarly, some would argue that Japanese drive designers would push technological improvements us- ing the inductive thin-film technology with which they are familiar rather than make the complex shift to MR heads. In one sense these suppositions are true; IBM invented MR technology and entered the market with it almost three years
296 U.S. INDUSTRYIN2000 before the nearest competitor. Yet, three of the next six companies to introduce disk drives with MR heads were Japanese companies. Moreover, Japanese technological strength is further revealed by looking at the are al density of a disk drive. Areal density encapsulates in one picture a company's ability to bring together head and media technologies and is a major feature of the technology race in HDDs. As Table 2 shows, the Japanese are also among the leaders in areal density. The table ranks firms according to the disk drive with the highest areal density each offers as of 1997 (Disk/Trend, 1997~. Once again, though IBM is clearly far ahead, three of the top five are Japanese. This ranking changes frequently, as the newest product to the market seems to embody the highest areal density, but the illustration nonetheless demonstrates Japanese innovativeness.6 A second exception to the conventional wisdom is that Japanese firms have also been weaker than theory would predict. Within a given form factor, technol- ogy has evolved in ways that should have given the Japanese an advantage. All companies have technology roadmaps, and technological progress has moved along well-known paths, especially in the technological development of the cur- rent generation of disk drives employing inductive thin film heads and disks. IBM was the first company to ship disk drives with thin film inductive heads in 1979; drives with thin film media appeared four years later. Innovations in areal TABLE 2 Highest Areal Density, as of May 1997 Company Areal density (megabits per square inch) IBM Hitachi Quantum Toshiba Fujitsu Maxtor Seagate ITS Micropolis Samsung 2638.0 2013.0 1646.0 1308.0 1300.0 1193.0 1108.0 1008.0 959.2 884.0 Source: Disk/Trend Report, 1997. 6Firms also competed in the desktop market in terms of "volumetric" density or how much capacity one could cram into the slot allotted to the disk drive. One trick in mechanical design was the introduction of "half-high" disk drives in which more disks were stacked closer together. A company might be a leader in areal density (data on a disk) but a laggard in volumetric density. Some say that IBM did not understand this distinction. Unfortunately, systematic data to test this notion are unavail- able. I thank Frank Mayadas for bringing this to my attention.
HARD DISK DRIVES 297 density during the next decade involved improvements to these two increasingly understood technologies. Japanese firms would thus have been expected to ad- vance more quickly along this technological trajectory while simultaneously ob- taining cost advantages through more efficient manufacturing, but the reverse is in fact true. American firms have dominated this largest segment of the disk drive market and are making interesting adaptations to the basic technology.7 In this way, American firms have been most responsible for extending the life of inductive head technology, which innovation theory would not predict. Overall, there is little evidence that the Japanese are less innovative than successful American companies according to these key measures. They have not been far behind their U.S. competitors on the technological frontier, and they have even introduced advanced new products before leading U.S. companies. Although innovation has been necessary for all companies to stay in the game, it has not been a sufficient condition.8 FORM OF INDUSTRIAL ORGANIZATION Many scholars argue that the Japanese form of industrial organization, with its complex interfirm relations, may have distinct advantages (Aoki, 1988; Gerlach, 1992; Teece,1992~. Although the evidence comes almost entirely from the automobile industry, the general claim is that Japanese firms are less verti- cally integrated than their American counterparts and maintain closer relation- ships with suppliers, often through some equity holdings (Aoki, 1990; Hill, 1995; Dertouzos et al., 1989~. By combining market incentives with relational con- tracting, Japanese companies are reportedly more cost effective, flexible, and faster in coordinating operations than their more vertically integrated competi- tors. Was there, ironically, something about the American form of industrial organization that sustained U.S. advantage in the HDD industry? Backward Integration: Components and HDD Assembly The basic issue is whether Japanese and American disk drive firms practiced different methods of organizing production and delivery. I focus on four of the most important disk drive components the recording heads that read and write the data, the disk to which data are written and stored, the motor used to rotate the 7Improvements to inductive technology include "proximity" or virtual-contact heads. These in- volve significant enhancements to etched air-bearing and transducer technologies. sit is important to note that I have not addressed the ability of firms to introduce successive genera- tions of products. When product cycles are so short, firms face intense pressures to stay competitive in terms of capacity, performance, and interfaces. Keeping design teams together in such a pressur- ized environment is difficult. It is possible that American firms have been better at this than those from Japan and Europe. I hope to explore this possibility in a later paper.
298 U.S. INDUSTRYIN2000 disk, and the semiconductors that control the drive and manage the flow of infor- mation between it and the computer. I also consider the extent of contract assem- bly of disk drives. In contrast to the microcomputer industry (see, for example, Langlois, 1992), vertical integration has been an important, although not universally implemented, strategy for HDD firms. I compared the degree of backward integration into components for a sample of 28 firms, which included both surviving firms and firms that exited the industry. Of the 28 firms, 16 were still producing HDDs in 1995. Ten of the firms are Japanese, one is Canadian, and the remainder are from the United States. The Canadian firm, Northern Telecom, is listed with U.S. firms because its disk drive operations were in the U.S. as a result of acquisitions of two American companies. The degree of backward integration across these firms was compared at four different points 1983, 1987, 1991, and 1995. The combined global market share of the 28 firms was 85 percent in 1983,91 percent in 1987, 98 percent in 1991, and 99 percent in 1995. The data show that back- ward integration has clearly been an important strategy in the industry and one that has become more prevalent over time. In 1983,75 percent of the HDD firms in our sample were vertically integrated in one or more key components. This number increased to 91 percent in 1991 and 94 percent in 1995. In-house assembly of HDDs has also been the dominant model in the indus- try, regardless of nationality. Contract assembly relationships have been com- mon, but they have not accounted for a large share of total production. Of the more than 100 firms that shipped disk drives under their brand names since 1976, only 20 used contract assemblers. The majority of firms that engaged contract assemblers did so because they were small and had limited resources or competed in niche segments. The important exceptions to the general model are Quantum and IBM. Probably 30 percent of all disk drives shipped in 1996 were done on a contract basis for these two firms. All of Quantum's disk drives are assembled by Matsushita-Kotobuki-Electronics. IBM has used its former English disk drive subsidiary, spun off in 1994 and now called Xyratex, to assemble drives. In 1997 Xyratex assembled about 2 million drives for IBM. IBM has also used a Thai subcontractor, Saha Union, to assemble 2.5-inch and 3.5-inch drives designed by its Japanese disk drive operation in Fujisawa. With the exceptions of Quantum and IBM, the vast majority of units shipped have come from HDD companies' own factories. All leading firms except Quan- tum have maintained a strong manufacturing capability. IBM, for example, no longer sources drives from Xyratex and is building a plant in Thailand to make its own. Moreover, most of those performing contract work have themselves been HDD firms rather than specialist assemblers. Several of those who contracted work to others also engaged in contract assembly themselves, including IBM. Interestingly, many American disk drive firms performed contract assembly. Of the companies that have contracted all assembly to others, only Quantum and tiny Nomai (France) survive.
HARD DISK DRIVES 299 Despite this dominant theme of in-house assembly and the general trend to- ward backward integration into components, the level of integration varies among firms. There does not appear to be any systematic difference in backward inte- gration between leading Japanese and American firms, however. Highly inte- grated firms exist on both sides of the Pacific. Seagate, IBM, Fujitsu, and Hitachi make virtually all of the key HDD components in-house, although they are not completely self-sufficient. Some Japanese and American HDD firms for a time only assembled HDDs and did not integrate upstream into components namely Kyocera and ITS. Those that never integrated into one of the four components were typically the smallest firms in the industry. Most firms in both countries lie somewhere between the two extremes, producing one or two key HDD compo- nents. If one focuses only on integration into heads and media, the story is much the same. Moreover, the closest observable interfirm relationships appear to be between American and Japanese firms Quantum and MKE, and Integral Peripherals and Fuji Electric, which owns a small share of Integral and at one time assembled Integral drives intended for the Japanese market. Samsung also provided most of the initial $1 million in start-up financing for the American firm Comport and assembled all of Comport's drives. More recently, IBM has announced a con- tract assembly relationship with NEC, which will manufacture IBM drives later in their life cycles and use IBM components. At the component level, the top three independent media companies are Komag (U.S.), Fuji Electric (Japan), and Mitsubishi Chemical (Japan). Only Fuji Electric has an equity relationship with an HDD firm, holding a small percent of Fujitsu. Yet, Fuji Electric's largest customers are MKE/Quantum and Seagate Technology, not Fujitsu (TrendFocus, 1996a). For recording heads, Read-Rite (U.S.), TDK (Japan) and its Hong Kong subsidiary SAL Magnetics, and Yamaha (Japan) are the three largest independent producers (TrendFocus, 1996b), and none has an equity relationship with any HDD producer. Forward Integration: Computer Systems Is there a systematic difference between Japanese and American firms in forward integration into computer assembly? It is true that virtually all of the surviving Japanese HDD firms make computers, and that none of the American firms, save IBM, do. This fact misses much of the evolution of the industry, however. Throughout the 1970s and 1980s many of the largest American HDD manufactures were computer makers. Conversely, numerous Japanese compa- nies began to make HDDs during the 1980s but did not make computers. Nonetheless, conventional wisdom in the industry is that the Japanese HDD industry has been at a competitive disadvantage because of its heavy reliance on captive sales. This dependence, it has been argued, tends to slow the speed and degree of innovation because captive drive makers may not be subject to the same
300 U.S. INDUSTRYIN2000 competitive pressures as noncaptive firms. George Scalise, former Maxtor presi- dent and chief executive, has expressed an additional disadvantage facing captive makers: "The history of computer systems manufacturers in most instances has been that building peripheral products like disk drives is not a core business that can generate volume and economies of scale to be cost competitive" (EN, 8/21/ 89~. Finally, captive HDD manufacturers are said to find it difficult to sell drives to outside computer firms. According to Micropolis founder, Stuart Mabon, "To enter into a large OEM relationship with a disk drive company means disclosing future computer plans. Most large OEM computer companies would prefer not to disclose those plans with a competitor" (EN, 8121189~. The captive market has indeed been whittled away, but the perception of Japanese companies as over-reliant on internal sales can be challenged on at least two fronts. First, as already discussed, captive sales have not made Japanese firms notably slower to innovate than successful American HDD firms. Second, the Japanese HDD industry has not been more reliant on captive sales than the American industry. If anything, the opposite has been true. Between 1983 and 1993 captive sales accounted for a slightly higher percentage of total HDD rev- enue among American firms than Japanese (Figure 3~. Only for the periods be- fore 1983 and after 1994 is it possible to argue that the Japanese have been more reliant on captive sales than the Americans. The differences between Japanese and American firms in their reliance on captive sales are too small to provide a compelling account of competitive advantage. 100 90 80 70 ~ 60 a) 50 40 30 20 10 o ~ U.S. \ Japan \ 1 1 1 1 1 1 1 ~1 1 1 1 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1 11 1 ~00 00 00 00 GO ~ Year FIGURE 3 Captive revenue as a percentage of total revenue. Source: Disk/Trend, Inc.
HARD DISK DRIVES 301 In summary, it is unlikely that variation in mode of organizing among U.S. and Japanese firms is responsible for the sustained U.S. dominance in the HDD industry. Until very recently, the industry has been characterized by a great amount of diversity in organizing, with specialists coexisting alongside vertically integrated firms. During the last decade, both Japanese and American firms have increasingly relied on backward vertical integration while reducing their reliance on captive sales to an in-house computer business. In fact, their integration strat- egies have been remarkably similar. GLOBALIZATION OF ASSEMBLY If the degrees of vertical integration, innovation, and path dependence in the disk drive industry explain only a part of the competitive advantage of U.S. firms, what else accounts for the American dominance?9 One important ingredient has been the globalization of assembly. Innovation is critical, but companies have to be equally effective at transferring new products quickly into volume production while keeping costs down in the face of rapid price erosion. The president of Seagate, the world's largest disk drive company, says that his company is happy to be a follower rather than an innovator but to outproduce its competitors (SoS, 1996~. The centerpiece of this production strategy has been overseas assembly. In general, American firms have not been known for their manufacturing prowess. Yet U.S. disk drive companies have demonstrated that this generaliza- tion does not hold for all industries. American disk drive companies competed squarely in and came to dominate the low-margin, high-volume segments the price and capacity points most in demand by users of personal computers. Judged by what scholars have had to say about the manufacturing failures of American firms in other industries, this is an extraordinary accomplishment. American industry achieved it primarily by being the first as a group to shift assembly offshore to lower-cost locations, where it quickly constituted an entire value chain of activities. If Silicon Valley is the geographical synonym for innovation, then non-Japan East Asia has come to signify low-cost assembly and logistics man- agement. A look at the movement of the industry overseas and how it furthered U.S. competitiveness is instructive. Home-Based Assembly: 1956-1982 In the 1960s and 1970s, before the introduction of the 5.25-inch disk drive, assembly of disk drives by American firms occurred primarily in Silicon Valley, the Los Angeles area, Minneapolis, Oklahoma City, and the region around Bos- ton. Some HDD firms that were vertically integrated into computers also as 9This section draws heavily on McKendrick and Hicken (1997).
302 U.S. INDUSTRYIN2000 sembled disk drives in Europe. IBM manufactured drives in Germany, England and Italy; Control Data manufactured in Portugal and had a joint venture in Ro- mania; and Burroughs had operations in Scotland. Burroughs also had assembly operations in Brazil and Canada. Japanese and European companies, with the exception of Germany's BASE, which had an operation in Silicon Valley, all assembled in their home countries. The principal rationale among the firms that located assembly outside their home market was proximity to customers: placing assembly in those markets where governments, banks, and insurance companies- the primary customers for their computer systems were likely to look favorably upon firms committed to local assembly of systems and peripherals. In 1982 and 1983 Seagate, Computer Memories, Ampex, and Tandon, all independent producers, became the first companies to move HDD assembly to locations for reasons other than access to host country markets. These firms began to assemble drives in what they saw as the best location from a cost stand- point, selecting low-wage areas in Asia, particularly Singapore. By the end of 1983 assembly was scattered geographically, but still overwhelmingly in the countries where the firm had its headquarters. Virtually all of the production of HDDs in 1983 was concentrated in two countries, the United States (72.3 percent of shipments) and Japan (12 percent of shipments). With almost 5 percent of global shipments, Europe produced more disk drives than all of Asia outside of Japan. In 1983 U.S. firms produced some 93 percent of their drives in the U.S., while Japanese firms produced all of theirs in Japan. A Shift in the Center of Gravity: 1983-1990 The experiences of Seagate, Tandon, and Computer Memories in Southeast Asia began to influence other American HDD firms. The perceived success of Seagate' s Singapore facility, in particular, spurred several other HDD producers to adopt a similar cost-based siting strategy. Table 3 shows the movement of overseas disk drive assembly among firms headquartered in America, Japan, and elsewhere. Many American firms followed Seagate's lead and chose Singapore as their first overseas manufacturing site. American HDD companies also opened overseas facilities in other low-cost Asian locations such as Taiwan and Hong Kong. In the span of just seven years, a dramatic change in the locus of assembly occurred. By 1990 Singapore was the world's largest producer of HDDs, ac- counting for 55 percent of global output, measured in shipments, with the rest of Southeast Asia accounting for only a percentage point more. As more firms located in Southeast Asia, supporting industries emerged in the region so that by 1990 three-fourths of the parts needed to produce a disk drive could be purchased there (LAT, 1990; BT, 1993~. The revealed global strategies of American and Japanese firms could not have been more different. By 1990, eight years after the first HDD was produced in Singapore, American firms assembled two-thirds of
303 V, ¢ m V ~ · ~ m ~ ~ 00 o ca ¢ O . 0 V, ca 0 . C) to ca cq cq 0 ~ ~0 A 0 ~0 0 0 0 0 0 0 0 . 0 ° }0 O.~ of ~ . to a' ~ v, ~ ~ v, a 0 ~0 ~ ~ ~ 2 ~ ~ ~ ~ ~ R ~ ~ ~ ~ j g A, 2 2 V, o so o Cam V, o .. VimVim .. .~v ~En em ~ ~ oo oo oo oo oo oo .~ a ca ~ ~ ~-= ·= ~ ~ .~ ~ ~ ~ ~ ~ ·~ ~ E~ m v, v, ~ ~ E~ v, o ·.G ~Eo, C: ~ ~ b .= b~o ~ .e ~ o ~ ~ oo oo oo 4= ca ;^ V,
304 o C) o ca o .~ to ca O - ~ S=5~5~ ca so D 3 ~ ca .0 .a ~ E~ ~ ~ o v ca ~ ~ o o .~ C) .O C) ca .C) REV o ~ v .~ Em ca .= · ~ v ¢ Em ca ca . ~ .= · · Cal V ~ z .e so C) ca ca . ~ .= · ca .~ Em ~ o V, o ~_ V ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~= ~ ~I ~ a a IS cd x ~ Mi =' Em Mi .. A ;^ to o Em .~ it ~ ~
HARD DISK DRIVES 305 their disk drives in Southeast Asia. What began as a variation from the norm became a collective phenomenon. In contrast, Japanese companies assembled almost none in Southeast Asia, and only 2 percent in the rest of Asia. Japanese companies instead continued to manufacture predominantly in Japan, where they produced 95 percent of their disk drives. As a group, Japanese firms were clearly hesitant to abandon a strategy that, up to the mid-1980s, appeared to be working namely, exporting from Japan. In 1984, for example, TEAC Corp. was shipping almost 60 percent of its output to the United States. Even as late as 1989, both Matsushita and Hitachi invested in Japanese manufacturing capability for 3.5-inch drives, judging that applying more automation to drive assembly would enable them to overcome the otherwise higher costs of manufacture in Japan. As the yen strengthened against the dollar and they turned their attention abroad, the United States, not Asia, was the site of their first overseas manufacturing investments. Fujitsu opened a U.S. plant in 1986, NEC followed in 1987, and Toshiba entered in 1992. At one point, Fujitsu reportedly intended to manufacture nearly all of its disk drives in the United States (CW,1985~. Toshiba explained that its strategy in HDDs was proximity to the market to respond to market needs more effectively by designing and build- ing products closer to the markets where they were sold (LAT, 1991~. Nor was Southeast Asia the chosen strategy for new Japanese entrants. After they entered in 1985, Fuji Electric, JVC, Seiko Epson, and Alps Electric all confined their manufacturing to Japan. Strategic Convergence: 1990-1996 For high-volume, low-priced, and low-to-medium capacity drives, where cutting costs was paramount, Southeast Asia was clearly the location of choice for American companies, and their strategy increasingly confined the Japanese to niches in the high-capacity segments. This was a surprising switch because high- volume, low-cost manufacturing is an area where the Japanese traditionally ex- cel. Eventually the success of the American firms impelled the Japanese to fol- low with investments in Southeast Asia. Between 1991, when Fujitsu began production in Thailand, and 1996, all the principal Japanese HDD firms gradually shifted manufacturing to Southeast Asia, principally the Philippines. Fujitsu's move to Thailand was motivated when one of its major production facilities in Japan reached maximum capacity (CI, 113192~. In addition to ex- panding the Japanese facilities and investing in the United States, as it had done in the past, Fujitsu decided to manufacture drives in Thailand and retooled an existing recording heads facility for production of low-capacity 3.5-inch drives (IDC, 2128191~. Production stayed at low levels until 1993 when the appreciation of the yen forced Fujitsu to move a large share of its manufacturing to Thailand. By the end of 1995 Fujitsu was doing nearly all of its volume manufacturing at the Thailand facility and a new facility in the Philippines. The president and
306 U.S. INDUSTRYIN2000 CEO of Fujitsu Computer Products of America cited the move to Southeast Asia as one of the prime factors behind the company's rapid growth in 1996. Fujitsu doubled its worldwide hard-drive revenues for 1996 and experienced a 123 per- cent growth in shipments compared with overall 1996 market growth of 17 per- cent (BWI, 6117197~. NEC, Hitachi, and Toshiba soon joined Fujitsu overseas. NEC completed its own HDD facility in the Philippines in 1995 and increased its off-shore produc- tion to 75 percent of total HDD output (COM, 10/9/95~. Hitachi also made its first HDD investment in the Philippines in 1995 and had 90 percent of its 2.5-inch disk drive production there in 1997; it planned to soon make all 3.5-inch drives there as well. By 1995 more than 64 percent of the world's disk drives were produced in Southeast Asia, generating nearly 61 percent of the industry's revenue (Table 4~. HDD production in the United States fell to below 5 percent of world shipments, generating less than 9 percent of world revenues, while production in Japan fell to 15.7 percent of shipments and 13.3 percent of revenue. By 1995 the U.S. indus- try produced two-thirds of its total assembly in Southeast Asia, and Japanese firms had greatly increased their presence in Southeast Asia, producing nearly 55 percent of their HDDs in the region. Virtually all of the remaining drive produc- tion for Japanese firms was still located in Japan 45 percent, compared with 18 percent still located in the United States or Japan for U.S. firms.~° By the mid- 1990s, then, the geographic distribution of Japanese assembly had begun to re- semble that of their American competitors. The Value Chain Follows Through continued investment in the region, nearly every part of the HDD value chain is now produced in Southeast Asia in some quantity, reinforcing its TABLE 4 Distribution of 1995 Production (% Based on Unit Shipments) Locations of production Type of firm U.S. Japan Southeast Asia Other Asia Europe Total All firms 4.5 15.7 64.2 5.7 9.9 100 U.S. firms 5.1 13.0 66.8 3.9 11.2 100 Japanese firms 0.0 45.2 54.8 0.0 0.0 100 Sources: Disk/Trend, Inc., and Globalization Project Database. i°American firms also extended the global assembly strategy to low-cost areas of Europe, Ireland, and Hungary between 1990 and 1995.
HARD DISK DRIVES 307 preeminence as the center of HDD and components production. Seagate offers a good illustration. In almost every year since its initial investment in 1982, Seagate has reinvested in Singapore upgrading existing facilities or building new ones. The largest investments include a $56 million investment in 1988, a $100 million investment in 1992, and a $200 million investment in 1994. Seagate has also invested heavily in Thailand and Malaysia in upstream activities such as motors, heads, and printed circuit board assemblies and has recently opened plants in Indonesia (circuit boards), China (HDDs), and the Philippines (labor-intensive head assembly). Today it is the largest private employer in both Singapore and Thailand. Independent manufacturers of media and heads have also moved into the region, further reinforcing it as the focus of the industry's global strategy. The first head maker to invest in the region was Applied Magnetics, which opened a plant in Singapore in 1983. Read-Rite, another head company, opened or ac quired facilities in Thailand and Malaysia in 1991. The first media maker to locate production outside the United States or Japan was Domain Technology, Inc., which began volume production in Singapore in 1988. Komag and Stor- Media invested in Malaysia and Singapore, respectively, in 1993 and 1995. The first investment in Southeast Asia by a Japanese media company was Hoya Media's Singapore plant in 1996. As of 1995 nearly 70 percent of the firms that make heads or head assemblies had plants in Asia (excluding Japan), while 36 percent of the firms had plants in Southeast Asia. Among media producers, 81 percent had plants in Asia in 1995, while 38 percent were producing in Southeast Asia. American HDD assemblers initiated the move to Southeast Asia and much of the value chain followed. By 1995, more than 60 percent of global employment in the HDD industry, including upstream activities, was in Asia outside of Japan (Gourevitch et al., 1997. After a decade of investment by both multinationals and local supplier firms, low-cost Asia has become the region of choice for the HDD industry. The technical imperatives of the industry ultimately led to a con- vergence of American and Japanese strategic posture. Global Strategy and National Advantage Despite recent Japanese movements into Southeast Asia, American industry was able to sustain its advantage by being the first to implement this global strat- egy. How, exactly, did this strategy confer an advantage on American firms? According to industry participants, American industry's early move into South- east Asia gave it the time to establish regional manufacturing, secure comple ~ iThis percentage is actually understated because it does not capture employment associated with a few of the least expensive components going into a disk drive base-plates, condensers, capacitors, screws, and so forth. These are sourced almost entirely from vendors in Southeast Asia.
308 mentary assets, and move down the learn U.S. INDUSTRYIN2000 . ing curve at a time the competitive logic of the industry dramatically shifted to low-cost, high-volume manufacturing. An industry consultant who set up several overseas production facilities explained the benefits of making disk drives in Southeast Asia in this way: "While loaded labor cost is typically one-quarter of the U.S. equivalent, material cost can be 30 to 45 per cent less than in the U.S. But since the cost of material can account for as much as 80 per cent for some peripheral products, the real savings are achieved through local sourcing for materials rather than in savings of labor costs. Overall, cost savings of 30 to 40 per cent can be achieved in making peripherals in South- east Asia versus the U.S." (EN, October 1984~. Other benefits of assembly in Southeast Asia included lower overhead costs, government incentives, faster in- vestment approvals, and a less costly but developed infrastructure such as preci- sion machinery, die casting, and a pool of skilled personnel in process engineer- ing. Disk drive firms reportedly cut costs by 30 percent by moving production to Singapore (FW, February 24, 1987~. These cost savings coincided with the emergence of the 3.5-inch disk drive to generate enormous advantages in that critical high-volume market. As is shown in Figure 4, the demand for 3.5-inch disk drives exploded in the late 1980s; as Table 5 indicates, the largest part of the 3.5-inch market was for noncaptive sales. It was America' s success in this market that extended its advantage. During this period, America's dominance was led by its independent HDD firms Seagate, Conner, Quantum, Maxtor, and Western Digital and was dependent on their ability to ramp up low-cost, high-volume production in Southeast Asia, where, with the exception of Quantum, they assembled the overwhelming majority of their 3.5-inch drives. It took three years for the U.S. industry to claim more than 50 percent of the noncaptive market. Although each of these firms trailed Hitachi, Fujitsu, and NEC in the introduction of this form factor, this global strategy allowed American industry to claim 90 percent of the noncaptive market by 1991. Operating from Southeast Asia, U.S. companies put tremendous pressure on higher cost Japanese manufacturers even in the Japanese market. The retail price in Japan of a 20- megabyte drive made by Japanese producers cost about 200,000 yen, whereas drives made by Singapore-based U.S. firms were beginning to be sold at a retail price of under 70,000 yen (COMLINE Daily News Computers, April 15, 1988~. This is a surprising twist: Japanese and European firms were early to market with an innovation but were ultimately squeezed by the price competition brought to bear by American firms. American firms developed the ability to evolve quickly to new products, with fast, smoothly executed production ramps in over- seas locations. Effective execution of production ramps also required U.S. manu- facturers to become highly skilled in managing the stream of component parts coming from a diverse supplier infrastructure, both vertically integrated and inde- pendent, located in many countries. In contrast, according to the business press, the cost structure of Japanese manufacturing was simply not competitive with
HARD DISK DRIVES 3~0 27500 25000 22500 20000 17500 ._ ~ ~ 5000 ~3 · - 12500 1 0000 7500 5000 2500 o ' - - - - - - - - -I 1 105,018/ 1 88,357 Worldwide Total l l l 1 3.5" 2/ /~ - ,5" / ~ I J 6.5-14" / 1.8" 0 _ 1 1 1 --7-7--- 1 1 1 1 1 1 ~ ~r I ' '- I ' -' -- oo 0 ~oo 0 ~ ~rat oo co oo oo oo Cut ax ~ax can ~cr ~cr ~on _ _ _ _ _ _ _ _ _ _ _ Year FIGURE 4 The market for disk drives by size (thousands of units). Source: Disk/Trend, Inc. 309 that of American industry in the product segments most in demand. The timing, direction, and scope of globalization thus extended the leadership of America's disk drive industry by enabling it to move down the learning curve in overseas assembly while accumulating effective capabilities in managing internal and ex- ternal international linkages in the value chain.
310 U.S. INDUSTRYIN2000 TABLE 5 The Market for 3.5-Inch Disk Drives (Thousands of Units) Non-captive U.S. market share Captive Non-captive Total 3.5-inch shipments of non-captive Year shipments shipments shipments (% of total) (% of total 1983 0 2 2 100 0 1984 0 67 67 100 11 1985 23 339 361 94 45 1986 250 1,108 1,358 82 56 1987 1,565 2,703 4,268 63 61 1988 2,310 5,899 8,209 72 71 1989 3,620 10,692 14,311 75 74 1990 3,564 16,336 19,900 82 82 1991 3,866 22,170 26,036 85 90 1992 3,972 32,342 36,314 89 92 1993 3,904 39,368 43,272 91 94 1994 4,364 55,980 60,343 93 95 1995 4,623 73,153 77,776 94 93 1996 4,679 83,678 88,357 95 84 Source: Disk/Trend, Inc. FUTURE PROSPECTS: THE LOCUS OF R&D AND GLOBAL KNOWLEDGE MANAGEMENT Now that locational strategies are similar across companies, what will con- tinue to sustain America's advantage in disk drives? The industry faces two critical challenges. The first is generating and exploiting innovations in a context that has changed considerably since the industry's beginnings 40 years ago. Although IBM remains a powerful presence, the sources of knowledge for the industry have broadened considerably, with universities taking on a more impor- tant role than at any time in the past. The second challenge remains implementa- tion. Now that all the major disk drive firms have physically separated assembly from product development, competitive advantage will continue to hinge on the management of that long-distance relationship. Overall, the United States re- mains well positioned with regard to both challenges. But there is some concern that U.S. disk drive and component companies are underinvesting in R&D at a time when Japanese and Korean companies have evolved into much stronger competitors. Trends in the Locus of R&D The context for innovation in the industry has shifted during the last 15 years. Until the mid-1980s, corporations were the source of almost all research i2This section has benefited from conversations with Barry Schechtman, director, National Storage Industry Consortium; Ami Berkowitz, professor, UCSD; Albert Hoagland, professor, Santa Clara University; Dawn Talbot, librarian, UCSD; and researchers at IBM.
HARD DISK DRIVES 311 leading to innovations in disk drives, and technical advance was very much driven by engineering within the firm, principally by IBM. Today, companies can no longer engineer their ways into higher densities. For example, the industry is concerned that it is approaching a fundamental physical limit to the density of magnetic disk recording in its present form the superparamagnetic limit. At the same time, the research underlying industrial applications is becoming more in- stitutionally and geographically dispersed. Since the mid-1980s, more research underlying industrial applications is being conducted in more locations by a vari- ety of research organizations. It is hard to determine cause and effect in changes in the locus of R&D. One method is to look at patents, which have been hugely important in the history of the industry. One experienced manager claims that Conner Peripherals, which became a $1 billion business in the shortest time in history, owed more for royal- ties than its entire historical profits for the company! Unfortunately, according to industry and university sources, a meaningful count for important patents related to magnetic recording is nearly impossible because one cannot distinguish the critical patents from the minor ones. As a convenient measure of the changes in the sources of R&D in the indus- try, this paper counts technical papers in the most relevant journals. Although research on magnetic recording traverses several journals, university and indus- try researchers suggested two in particular. One is a peer-reviewed journal, the IEEE Transactions on Magnetics (IEEE), which I cover in five-year increments since 1970. The second is The Magnetic Recording Conference (TMRC), which invites scholars to submit their papers. TMRC was used as a proxy for research quality because only those perceived to be doing the most interesting or advanced research are invited. Unfortunately, TMRC has published only since 1990, so it provides a picture of only current research. The institutional sources of research were also tracked: firms, universities, or "other," principally government or quasi- government research labs. Table 6 summarizes the geographic trends in the sources of R&D, three of which I note here. First, while the relative share of publications by United States authors has declined since 1970 in IEEE, research by the United States is still the most influential as measured by invited publications in TMRC. Second, and somewhat surprisingly, the share of publications by authors from Japan and Eu- rope has remained fairly steady, with Europe showing considerably more strength in magnetics than one would think from its market presence in HDD and compo- nents. Of course, magnetics research is applicable to several branches of indus- try, so this count is an imperfect measure of capability in data storage. Third, however, when measured by TMRC, which is more specific to magnetic record- ing research, the positions reverse, with Japan having a higher representation than Europe. Research publications were also classified according to the institutional af- filiation of their authors. The percentages in Table 7 refer to the share of a given
312 TABLE 6 Location of Authors Publishing in Academic Journals on Magnetics, 1970-1995 U.S. INDUSTRYIN2000 Total Cross numberof United % of % of % of % of national % of Year articles States total Japan total Europe total Other total Authorship total IEEE Transactions on Magnetics 1970 37 19 51 4 11 9 24 5 14 0 0 1975 47 25 53 6 13 13 28 3 6 0 0 1980 68 30 44 12 18 20 29 6 9 0 0 1985 64 24 38 14 22 16 25 5 8 5 8 1990 81 27 33 10 12 23 28 16 20 5 6 1995 83 23 28 15 18 22 27 15 18 8 10 TMRC 1990 19 12 63 4 21 2 11 0 0 1 5 1991 23 19 83 3 13 1 4 0 0 0 0 1992 25 22 88 1 4 0 0 0 0 2 8 1993 28 24 86 2 7 2 7 0 0 0 0 1994 30 24 80 3 10 2 7 0 0 1 3 1995 28 23 82 4 14 1 4 0 0 0 0 region's publications for example, the share of all American papers that were contributed by American firms or universities. The most obvious fact about this research is the extent to which corporations are still involved in fundamental research in the United States and Japan but not elsewhere. Among firms, IBM remains the strongest company in magnetic recording technologies for hard disk drives and still has an enormous reservoir of technical talent. But its share of articles in IEEE declined during the period covered, from 40 percent of all ar- ticles contributed by U.S. firms in 1970, to 22 percent in 1985, and to zero in 1995. Quantum was the only disk drive firm that contributed articles that year. This last figure for IBM is surely an outlier, however, since IBM was affiliated with between 20 and 40 percent of all American papers to appear in TMRC be- tween 1990 and 1995. There is some similarity between the United States and Japan in the institu- tional affiliation of authors. As expected, university research has featured more prominently in both countries during the last two decades. Of all U.S.-authored papers appearing in IEEE in 1990, almost 60 percent were by researchers affili- ated with universities, and 25 to 45 percent of American papers appearing in TMRC were contributed by universities. Magnetics research in Japan is slightly more concentrated in Japanese firms than in universities or other institutions. Japanese firms accounted for more than two-thirds of Japanese papers appearing in IEEE in 1970 or 1975, still more than half in 1995, and virtually all of the high- quality research on magnetic recording as measured by TMRC. Japanese univer
313 C) a' _' Cq C) ·_4 a' bC o Cq o C) ¢ Cq ·_4 Cq s~ o ¢ o o · _4 ¢ E~ s~ ~ ~o ca s~ ~ :~=,~, o .~ .~ ~ ca ca V ·~L, o ca .= ca s~ ~ ._ s~ ~o ca .= ca .~ o ._ o ca · ~ ca . - o ca ca 4= . ca C) ·~ o ca .= o ~ ~ CM E~ o o o o o o ~ o ~ oo o o o o ~ o o ~ CM o o CM CM o ~ oo o oo o oo ~ oo o ~ o o oo CM ~ CM ~ o ~ oo o ~ oo ~ ~ ~ ~ ~ oo oo ~ ~ ~ ~ ~ o ~ o ~ ~ ~ o ~ ~ ~ ~ o o ~ o ~ o ~ ~ ~ oo oo ~ ~ o o ~ o o o o o ~ o ~ o o o ~ o ~ o o ~ ~ o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o 8 8 ° 8 8 8 o o o o o o o ~ o o ~ o o ~ o o ~ o o ~ o o ~ o o o ~ o o o o ~ o o
314 U.S. INDUSTRYIN2000 sities are stronger in magnetic recording than they were in the 1970s, however, contributing anywhere from 40 to 55 percent of the publications in IEEE since 1980. Overall, the United States remains strong in research related to HDDs, with American universities becoming a more important source. The Japanese also have research depth in magnetic recording, some of which may be concealed by these measures. In particular, it is likely that not all relevant Japanese research appears in these journals.~3 But it is also true that research with commercial usefulness can come from anywhere, and companies need to have the capability to evaluate and incorporate new knowledge into industrial applications. In 1988, for instance, two scientists from France and Germany working independently published papers that directly influenced the development of so-called "giant" magneto-resistive heads, which IBM planned to incorporate into its disk drives in 1998. Following is a description of the efforts underlying these trends. Basic and Applied Research in the United States: University and Collaborative R&D Magnetics research is the basic foundation for the disk drive industry, and Europe was the pioneer in magnetism. BASE (Germany) and Philips (The Neth- erlands) had strong labs, and Philips is still strong in magnetics related to optical storage. In the United States, Bell Telephone Laboratories, IBM, General Elec- tric, and Westinghouse all had magnetic recording research programs during the 1940s and 1950s. The material basis was a continual refinement on earlier devel- opments in Europe. By 1980 only two large groups in the United States were still at work on applied magnetics in any systematic way, one at Bell Telephone Labo- ratories, and the other at IBM. IBM is still the primary American source of industrial research with a direct impact on the HDD industry; Seagate, Quantum, and Western Digital do not have the R&D resources of IBM. But IBM is no longer the only locus of important research relevant to magnetic recording. Government Support. Compared with many other important industries, gov- ernment support for data storage in general, and HDD in particular, has been i3For a time, the Center for Magnetic Recording Research (CMRR) at the University of California, San Diego, hired translators in Japan and the United States to monitor technical developments in magnetics appearing in Japanese technical journals. CMRR then sent out abstracts to its member companies, who then might request the full text. Because of industry interest, this cost quickly esca- lated and CMRR dispensed with the practice. But the experience indicates that interesting Japanese research does not always appear in English language journals, even if they are often the most desir- able outlet for a Japanese researcher. Between 1985 and 1994, the IEEE published the IEEE Trans- lation Journal on Magnetics in Japan, which translated important research from Japanese-language journals. After some lag, these articles were cited 111 times in English language publications. I am grateful to Dawn Talbot for this information.
HARD DISK DRIVES 315 barely measurable. During the 1950s the Naval Research Lab did its own mag- netic recording research; Oak Ridge Laboratories worked on neutron fractions, a subject fundamental to magnetism; while the Office of Naval Research sponsored some research. In the 1950s the National Bureau of Standards also did some research developing the world's first rotating disk storage device, which came to IBM' s attention (Bashe et al., 1986: 280; Rabinow,1952~. But government influ- ence on industry is hard to trace and appears diffuse or very indirect at best. The government's most explicit commitment to the storage industry gained expres- sion only in the 1990s. The Defense Department's Advanced Research Projects Agency (DARPA) and the National Institute of Standards and Technology (NIST) have begun to consider data storage a critical industry and have helped to under- write some recent research through the National Storage Industry Consortium (See below). In addition to providing limited funding, NIST itself engages in magnetics research at its Electronics and Electrical Engineering Laboratory. The lab- oratory's Electromagnetic Technology Division collaborates with the magnetic recording industry in developing metrology to support future recording heads and media with their ever-increasing data density. The division also provides new measurement methods, instrumentation, imaging and characterization tools, and standards in support of the magnetics industry. Compared with private and uni- versity research, its role has been quite small, but NIST has announced its interest in enlarging its magnetic and storage programs (Rhyne, 1996~. University Research. In the 1950s and 1960s the theoretical basis of re- search in magnetism was fairly primitive, and no American university had an academic center doing work in magnetic materials. In fact, the most eminent people in magnetics were in or came from industry. Although important theoreti- cal advances may have come from academe, progress in magnetism was very much a story of individuals. That is, basic research affecting the HDD industry from the 1950s through the early 1980s was excited and generated by individuals and industrial interests rather than organized university or government programs. In the late 1970s, the HDD industry recognized that the technological trajec- tory had gone as far as it could without more fundamental research. The mag- netic recording industry (tape, optical, and floppies as well as disks) in the United States brought its concerns to the National Science Foundation (NSF), arguing that unless academia got involved, the United States risked losing its interna- tional lead in the industry. But the research foundations for magnetic storage are so interdisciplinary that universities typically were not organized to address the industry' s problems. Given that physics, chemistry, materials science, mechani- cal engineering, and electrical engineering are all relevant to storage, it was diffi- cult to find a home department. In recognition of this need, the NSF and the industry funded a Magnetics Research Center at Carnegie Mellon University in 1982, which was subsumed
316 U.S. INDUSTRYIN2000 into the Data Storage Systems Center, an NSF Engineering Research Center since 1990. The Magnetics Research Center opened with mostly federal and some industry funding. Initially, its researchers attacked a broad spectrum of problems in materials development. The center's first major effort was to understand the basic chemical and physical processes that hindered the development of better recording materials and equipment, including the development of a theoretical model of the recording process. Firms did not yet understand many of these fundamental issues. The center's main focus today includes the integration of storage systems into high performance computing and research on high density magnetic and magneto-optic disk and tape recording including the electronic and mechanical subsystems. In 1983, industry took the lead in establishing the Center for Magnetic Re- cording Research (CMRR) at the University of California, San Diego, a location more convenient to Silicon Valley than was Carnegie Mellon.l4 CMRR was established by seven sponsoring firms, with UCSD contributing land and faculty; there was no federal assistance. The number of industrial sponsors has since more than doubled, and they provide part of the research expenses, which are leveraged by grants from federal agencies. To ensure academic independence, no restrictions are imposed on the public release of research results, and faculty and staff agreed not to patent any of their work. CMRR has more focus on long-term fundamental research than CMU tribology, signal processing, mechanics, in- formation theory, micromagnetic modeling, and research on new kinds of mag- netic particles. These are the two most prominent centers for magnetic recording research, but other universities have received industry support and are strong in fields re- lated to disk drives, including U.C. Berkeley, Stanford, Minnesota, Alabama, Washington University, and George Washington University. Santa Clara Uni- versity became an important source of engineering graduates for the industry following the establishment of its Institute for Information Storage Technology in 1984. In fact, university graduates with training in magnetic storage have had a profound impact on the industry. Historically HDD firms had to train graduates for a couple of years because many new engineers knew only one piece of the industry or had little knowledge about magnetic recording. In the near future, the industry will be entirely headed by a generation of graduates of these centers, people who deeply understand all aspects of the industry. National Storage Industry Consortium (NSIC). Formal collaboration among U.S. firms in R&D is a new phenomenon. The big donors to the university cen- ters gradually observed that everybody was working on similar issues, and they wanted more diverse and complementary research. At the same time, the HDD intone might wonder why U.C. Berkeley or Stanford did not become the industry's choice. In fact, both were, but neither institution was ready to commit to building a center.
HARD DISK DRIVES 317 industry thought that it should get more federal monies and questioned why semi- conductors should get such favorable treatment. These concerns spawned the industry consortium idea. It was useful to use NSIC as a forum to govern univer- sity research underwritten by industry donors instead of meeting with each aca- demic group independently to try to guide research. Incorporated in 1991, NSIC is a nonprofit mutual benefit corporation whose membership consists of about 60 corporations, universities, and national labora- tories with common interests in the field of digital information storage, not only HDDs but also optical and tape storage. During its first five years, NSIC started only one project a year, and DARPA or NIST supported 50 percent of the costs. The primary emphasis of these projects was to advance less established storage technologies. In 1996 that focus changed to support projects of more immediate interest to NSIC's members but that were still in the precompetitive stages of research. About 40 percent of the $100 million that has flowed through NSIC has come from federal tax dollars, mostly from NIST and DARPA. To date, the HDD industry has received support for three projects, with much of the work being done by universities. Each project has focused on increasing the perceived limits on are al density. University and consortia! research are helping U.S. HDD firms focus on long- term and fundamental issues that present technology barriers. Although the in- dustry has come to value the role of NSIC and university research generally, there is a perception among many managers and academics that among U.S. firms only IBM is doing sufficient internal applied technology aimed at proprietary solu- tions. These observers are concerned that U.S. disk drive firms, as well as inde- pendent component companies, are too focused on near-term needs and thus not well positioned to absorb new developments that will arise from external sources such as universities. Although fear of foreign competition has always been over- blown in this industry, the changing competitive context may make the U.S. in- dustry more vulnerable than it has ever been. Foreign Sources of R&D Japan is the source for the most significant R&D outside the United States, but R&D efforts in South Korea, Singapore, and Europe are worth noting. Japanese Firms. Among Japanese firms, Fujitsu and Hitachi have the great- est depth in technology. In some areas their technical base rivals IBM's and completely dwarfs that of other U.S. manufacturers. Both companies have strong Japanese research labs. To complement its domestic research related to disk drives, in 1987 Fujitsu acquired Intellistor, a data storage and subsystems design company in Longmont, Colorado. In 1991 it was made part of a new subsidiary, Fujitsu Computer Products of America, and became Fujitsu's North American research arm. Longmont is Fujitsu's only magnetic storage R&D operation out
318 U.S. INDUSTRYIN2000 side of Japan. In 1995, although it held only a small percentage of the market, Fujitsu announced its plan to capture 20 percent of the worldwide market for hard disk drives by the end of 1996. One part of the plan was a commitment to in- crease manufacturing capacity in the Philippines and Thailand, consistent with the convergence in global strategy described in the previous section. The other component was the development of second generation MR head technology and its incorporation into a larger range of drives. Fujitsu has strong internal mag- netic recording research and development capabilities, skills surpassed only by IBM's. In fact, the company experienced more than 100 percent unit growth in 1996 and offers drives with among the highest areal density of any manufacturer as well as a broadened product line that includes high-end drives for servers. Although it has a smaller presence in disk drives than Fujitsu, Hitachi has greater research depth according to some observers. It is strong in science and has contributed more articles to TMRC than any other Japanese disk drive com- pany. Hitachi is also the most likely firm to be first after IBM in introducing giant magnetoresistive (GMR) heads. According to one manager of a U.S. corpo- rate R&D lab, Hitachi lags behind Fujitsu and American firms primarily because it has less effective technology transfer. In 1995, the four largest Japanese HDD firms Fujitsu, Hitachi, Toshiba, and NEC agreed to form a consortium to research and develop HDDs and data storage devices. Like NSIC, the Storage Research Consortium intends to spon- sor work in which industry and universities collaborate. With an expenditure of $2.2 million, its resources are currently smaller than are NSIC's. But the com- panies hope to attract as many as 30 others, expand the consortium's resource base, and involve other parts of the value chain including makers of disks, heads, circuit boards, and test equipment. Membership in the consortium is open to foreign companies that have Japanese development or production facilities. Thus, Seagate is effectively excluded, while IBM, which develops drives in Fujisawa, and Komag, the world's largest independent maker of disks and with a Japanese joint venture, presumably are eligible. It is too early to judge the consortium's impact. Korean Firms. South Korean firms have been at the edges of the disk drive industry for more than a decade, but only in the last few years have they made the financial investments necessary to compete in this high-tech commodity busi- ness not only for volume manufacturing but also for R&D. Hyundai entered the industry in 1995 through its acquisition of Maxtor, one of the most successful American start-ups of the early 1980s. The parent firm has announced its goal is to become the world's second largest HDD supplier by 2000 and the world's leader by 2005. It intends to invest $1 billion by 2000 and set up a global network linking production bases in China and Thailand, compo- nent suppliers in Singapore and Hong Kong, an R&D base in the United States, and headquarters in Korea (CDSN, 1996~.
HARD DISK DRIVES 319 Maxtor's U.S. R&D team is the center of Hyundai's disk drive operations. Maxtor will concentrate on R&D, marketing, and production of high-end HDDs. Singapore will be the headquarters for procurement and a production plant for high-end drives. Thailand and China will focus on assembly of low-end units. Hyundai will also maintain production in Korea, where the firm's storage divi- sion will be headquartered. Although Hyundai will certainly attempt to strengthen the technical capabilities of its Korean-based operation, it is unlikely that its U.S. R&D work can be overtaken in the near term. In fact, Maxtor recently announced plans to establish a new engineering center in California headquarters, joining its California-based Advanced Technology engineering group, which focuses on heads/media integration. The new California engineering center will comple- ment Maxtor's existing engineering operations in Colorado, where it develops high-end drives for the desktop PC market (BWI, 9112197~. For its part, Samsung has announced that it intends to become the fourth largest HDD company in the world by 2001 (KEW, 1996~. It entered the disk drive industry initially as an investor in and contract manufacturer for a small U.S. start-up. After the American company failed in early 1990, Samsung con- tinued to make its drives and then began to develop its own. But the company limped along for several years with products just behind the market and scouted around for acquisitions in recording heads and another disk drive company while reportedly losing money in the business. It renewed its commitment in 1996 when it completed a $370 million investment in a new HDD plant. Besides adding to its manufacturing muscle, Samsung continued to build up its develop- ment center in San Jose, California, where it had been developing disk drives since the early 1990s. The center is responsible for advanced engineering, prod- uct development and qualification, marketing, product planning, and technical support. With its new manufacturing and R&D resources, some observers think Samsung has the potential to be a force. But given its earlier problems in product development, R&D is likely to stay centered in California for some time. Singapore's Data Storage Institute. Singapore has organized an intensive effort to move up the technology ladder in the HDD industry and has established a Data Storage Institute. Started in 1992 as the Magnetics Technology Centre, its role expanded in 1996 when it became one of three new research institutes at the National University of Singapore. It will receive government support to the tune of an initial S$30 million for the building and S$55 million over three years. The new institute will do research and participate in joint programs with multinational corporations involved in HDD, opto-electronic, and disk media technologies. It has more than 160 researchers and is expected to train about 40 engineers every year for employment in the HDD industry. One interesting aspect of the institute is the help it is receiving from leading American HDD firms and researchers. IBM and Carnegie Mellon University are on its advisory panel, more than half its corporate members come from the Ameri
320 U.S. INDUSTRYIN2000 can industry, and its first project in 1992 came from Seagate. The institute will not focus on fundamental research as much as U.S. universities do. Instead, it will leverage Singapore's strength in HDD manufacturing to improve process technologies, including testing, as well as offer more direct and immediate sup- port to industry than do American research universities. It is also exploring inno- vations at the component level, such as a collaborative effort with IBM, Motorola, Fujitsu, Hitachi, and a local Singaporean company to "push the benchmarks" on channel chips that do the read/write function on hard disks. Early in 1997 the government hived off the Data Storage Institute and other research facilities into separate and independent companies. Their primary roles will be to support industry in technology development through "greater respon- siveness" (BT, 1997~. Given Singapore's place in the global HDD network, and the government's unique commitment to the industry, the Data Storage Institute will likely evolve into a center of excellence. At this stage, the institute comple- ments rather than competes with American university research. The European "Scotsman" Project. Despite its considerable research base in magnetics, Europe has only two indigenous disk drive development companies remaining, Calluna Technologies (Scotland) and Nomai (France), and very few suppliers of primary components. The "Scotsman" (Strategic Components, Tech- nologies and Systems in Magnetic Storage) project is a collaboration initiated in February 1996 under the Esprit research program of the European Union to work on head technology. In addition to Calluna and Nomai, the other members are Myrica (U.K.), which is Nomai's development subsidiary; Silmag S.A. (France); and Xyratex, Ltd. (U.K.), the IBM spin-off. Half the $5 million in funding is provided by the European Commission and half is provided by the partners. The primary technology is expected to come from Silmag, which has developed what some say is a leading recording head technology. Although no one expects Eu- rope to obtain a leading position as a result of the project, it is intended to main- tain European expertise in magnetic storage and in the removable disk drive niches in which Calluna and Nomai operate. The Relationship between Product Development and Volume Manufacture In the course of the industry's evolution, pockets of technical sophistication developed in the United States (the Los Angeles area, Silicon Valley, Minnesota, Colorado, and, to a lesser extent, the Boston region), Japan, and Europe (the United Kingdom, the Netherlands, France, and Germany). Through industry con- solidations, surviving firms have found themselves in possession of R&D assets in more than one location. At the same time, the shift of assembly away from a firm's home base means that the management of technical knowledge between geographically dispersed facilities has become a critical organizational task. Table 8 lists all HDD firms in operation as of mid-1997 and the location of their
HARD DISK DRIVES TABLE 8 Location of HDD Product Development and Assembly, 1997a 321 Product Product Company development Assembly Company development Assembly Seagate California, Singapore Hitachi Japan Japan U.S. Philippines Colorado, Thailand U.S. Samsung S. Korea S. Korea Singapore Ireland California, Oklahoma, Malaysia U.S. U.S. Minnesota, Oklahoma, Micropolis California, Singapore U.S. U.S. U.S. China Iomega Utah, U.S. Malaysia IBM California, Singapore U.S. ITS California, India New York, Thailandb U.S. U.S. Japan Hungary SyQuest California, Malaysia EnglandC U.S. Quantum California, Japan Avatar California, Thailand U.S. Singapore Systems U.S. Ireland Calluna Scotland, UK EnglandC Western California, Singapore Technologies Scotland, UK Digital U.S. Minnesota, Malaysia Gigastorage California, Chinae U.S. U.S. Toshiba Japan Japan Integral Colorado, Singapore California, Philippines Peripherals U.S. U.S. Sequel California, California, Fujitsu Japan Japan U.S. U.S. Colorado, Philippines U.S. Thailand Nomai France EnglandC Scotland, UK Maxtor Colorado, Singapore U.S. Raymond Connecticut, Connecticut, Engineeringf U.S. U.S. NEC Japan Japan Philippines Sagemf France France aFirms listed in rough order of HDD revenue. bContract manufacture by Saha Union. CContract manufacture by Xyratex. Contract manufacture by Matsushita-Kotobuki Electronics. eNot yet shipping: in negotiations with the Chinese government. fMakes small numbers of "ruggedized" drives. Sources: Author's data.
322 U.S. INDUSTRYIN2000 product development and volume manufacturing facilities. Locations for compo- nent development and manufacturing in vertically integrated firms are omitted. One of the most remarkable characteristics of the HDD industry is that, with very few exceptions, product development is geographically separated from volume manufacturing and, among the leaders, by great distances.~5 An important question is whether the trend witnessed over the last 15 years in the internationalization of assembly will extend to other core organizational tasks. What are the implications of the growing importance of university re- search for how globally dispersed a firm's R&D can be? Specifically, will R&D follow manufacturing offshore, or do other forces act as countermagnets? Is the industry defying those who argue that remote manufacturing can cause quality and service problems that outweigh any apparent savings? Three outcomes are possible: R&D could follow assembly abroad; assembly could return to be closer to product development; or the industry could reach some manageable equilib- num. The likely scenario is that the current organization of the industry will per- sist. That is, most disk drive design and development, along with pilot produc tion lines, will remain concentrated in the United States and Japan, and volume manufacturing will continue to be physically separated and situated in countries where assembly is cheaper. Firms offer a number of interrelated reasons for why this kind of organizational arrangement is effective and durable. The short expla- nation is that product transfer has become straightforward, and any costs associ- ated with transfer and coordination are paid for with just one day of high volume manufacturing in a lower cost location. Typically, companies conduct pilot production proximate to product devel- opment because of the greater risk with design during initial assembly. Compa- nies also form product transfer teams consisting of product developers and pro- cess engineers from both the home product development facility and the volume manufacturing facility. Today, a product transfer team might be as big as 40 people, and the team stays with the product from pilot through ramp-up overseas. Then the manufacturing team takes over responsibility for volume assembly. Seagate's Malaysian facility, for example, has had a good experience with product transfer and can ramp up quickly. Six to eight engineers in the United States write the code, do mechanical design and testing, and work with the pro- cess people in the domestic facility to stabilize yields during pilot production, which might involve as many as 10,000 drives if the product is especially ad- vanced. A few weeks before the transfer to Malaysia, quality, operations, and lead operators in Malaysia go the United States to prepare for transfer. Then six to ten people from the U.S. team go to Malaysia for 3 to 5 weeks to ensure a good i5The exceptions are for the very smallest disk drive firms. The last ten companies listed in the right-hand side of Table 8 had less than one half of one percent of the global market in 1996. i6This information was provided by the plant manager during the author's visit.
HARD DISK DRIVES 323 start. The product release date is typically met because, according to the plant manager, "transfer is almost routine now, very smooth." Transfer is also effec- tive to Seagate's Thai facility, which makes less sophisticated products and hosts fewer U.S. engineers during transfer but sends more staff to the United States before transfer because its workforce is less skilled. Western Digital uses a formal new product introduction process that allows it to achieve 92 percent yields in overseas assembly.~7 The firm has a special group to manage the process which it claims gives the company an advantage in time to market, time to volume, and time to high yields. In product design, the company uses a typical "gating" process common to well-managed high technology firms. The product concept and then the product itself need to pass certain gates on its way to manufacturing. These are milestones that have to be met at each step. Design, engineering-level build, test, and tooling buildup take place in the United States because there are still bugs that need fixing. Concurrent engineering is occurring in Asia where production level equipment is introduced and compo- nents are chosen. Transfer teams to and from Asia then ensure a smooth product transition. The transfer team from Asia visits San Jose to work on the pilot line to "wring out" the process, while the U.S. team stays with the product all the way to ensure manufacturability. The transfer process involves 40 people for 30 days. An important reason for the success of this model is that both the technology and the assembly process are better understood than they were 15 years ago, new products are increasingly designed so that they do not disrupt existing manufac- turing processes, and computer information systems lower the costs of long dis- tance management. Although firms vary somewhat in their ability to minimize changes to products that might otherwise require substantial changes in tooling for the assembly process, companies try to maintain substantial commonality in components across products. Western Digital, generally thought of as the leader in this regard, has 70 to 80 percent commonality between products. At the same time that companies have accumulated skills in design, transfer, and ramp-up, the quality of the infrastructure in Singapore and Malaysia has fa- cilitated technology transfer and rapid ramp-up to volume manufacture. The disk drive industry has developed a large base of skilled professionals in the region with specialized industry knowledge, and the Malaysian and Singaporean gov- ernments have been aggressive in offering complementary services, such as rapid investment approvals, access to land, and labor training programs. As a conse- quence, there appear to be considerable cost savings with little lost in product yields or volume output. As product cycles shorten, ramping up has become even faster in Asia. In 1995 Western Digital ramped up production from zero to 750,000 units within three months (CRN, 1995~. In 1996 Quantum/MKE went from zero to 7 million disk drives in nine months (NST, 1996~. Moreover, over i7This information was provided by a Western Digital vice president during the author's visit to the firm's Malaysian facility.
324 U.S. INDUSTRYIN2000 seas assembly does not appear to cost companies in terms of yield. When IBM shifted production from California to Singapore, it not only ramped up quickly (from October to December), but it did so with no loss of yield. Like other drive companies, IBM could find all the engineering and managerial skills it needed in Singapore, and the Singaporean government facilitated the move by approving the investment quickly and even leasing IBM a plant that the government had specifically prepared for disk drive assembly. According to one IBM manager, the company could ramp to volume manufacturing in Singapore faster than any- where else in the world, including the United States. Coordinating technical activities with volume manufacturing across national boundaries has become standard practice for the industry. The model for Ameri- can firms is design and pilot production in the United States, fast ramp-up in Singapore, and matured products and process transferred out to Malaysia, Thai- land, or China. The system has become routinized, and American firms excel at it. Other than niche players, firms that did not adopt this organizational model, or executed it poorly, have all exited the industry. Summary Some industry managers and academics involved in magnetic recording see the current era as a watershed in industrial applications in magnetics, with univer- sities contributing to industry in fundamental ways. Although differences remain in the priorities and interests of academe and industry, they have become more and more aligned. America's university centers have also had profound influ- ences on the industry in another way; their graduates populate the data storage industry and in the near future will be its leaders. Some in industry, however, express mild concern that some U.S. companies, plus the independent U.S. media and recording head manufacturers, are not investing enough in in-house applied technology to enable them to absorb and commercialize university research quickly enough. By contrast, Fujitsu and Hitachi are much more heavily in- volved in applied technology work, and some American observers fear that the growing importance of more fundamental technological research will play into their strengths, especially now that they have also adopted low-cost manufactur ing strategies. In the near term the United States and Japan are unlikely to be displaced as the centers of research and product development. University research in the United States has become a more significant factor in the industry's technical evolution; the research labs of the big Japanese firms continue to make important advances in data storage; and Korean producers depend primarily on U.S. R&D. The only major impetus to shift the locus of R&D out of these countries comes from Singapore, which assembles more disk drives than any other country in the world. Yet even though Seagate and others may elect to do more product devel- opment in their Singapore subsidiaries, and the Data Storage Institute is making
HARD DISK DRIVES 325 progress, the technical resources and depth in the United States and Japan con- tinue to attract investment in R&D. It also seems clear that American HDD firms know how to manage interna- tional operations and coordination between home-based product development and foreign assembly, including its international supply chain. Barring a discovery that product yields would in fact be greater if assembly were brought home to be closer to product development, there is little indication that the physical separa- tion of development and manufacturing cannot be sustained. CONCLUSION Path dependence, industrial organization, and innovation all contributed to American success in the disk drive industry. The American HDD industry was built by successful computer firms, which enabled the industry to achieve an early lead over European and Japanese drive manufacturers. In addition, product development capabilities and some degree of vertical integration have been nec- essary conditions for industrial performance. Each of these contributed to America' s initial industrial advantage, yet taken together they are insufficient in explaining the ability of American firms to sustain their dominance. Differences between American and Japanese firms along these dimensions do not appear strong enough to explain the persistence of American leadership in the industry. This chapter suggests that a potentially important yet overlooked variable in stud- ies of national industrial advantage may be the scope, timing, and direction of an industry's overseas manufacturing operations. By being the first to shift assembly offshore, American firms were able to learn the organizational technology of international coordination and production. Although their activities were dispersed, they were at the same time concentrated in key regions research and development in the United States, labor-intensive assembly in low-cost Asia, and somewhat more skilled assembly activities in Singapore (Gourevitch et al., 1997~. American firms combined the benefits of low-cost, high-volume assembly with sophisticated management of these value networks. Innovative firms that failed to shift assembly abroad exited the indus- try or else claimed imperceptible shares of the market. The history of the disk drive industry differs from other high-technology industries in additional ways. First, the American HDD industry excels at manu- facturing. This is contrary to what researchers have observed in other industries, where Japanese firms are leaders in manufacturing. The business press initially expected that pattern to hold for disk drives as well: "Once in production, a disk drive is basically a commodity product that must be assembled as quickly and as cheaply as possible something that the Japanese are expert at doing" (BOO, 1984~. If anything, they lagged behind American firms in their ability to ramp to volume manufacturing. Especially interesting is that the vast majority of assem- bly was conducted in-house. Although companies have frequently resorted to
326 U.S. INDUSTRYIN2000 contract manufactunng, such arrangements have played a small role quantita- tively. Second, the American industry was largely ignored by the federal govern- ment and university departments during its first two and a half decades. Cer- tainly, disk drive programs in private firms benefited at least indirectly from fed- eral monies earmarked for computers and semiconductors. Yet technical progress in disk drives went largely unnoticed by those outside the industry and was achieved through heroic mechanical and materials engineering efforts in firms, especially in IBM, rather than through publicly funded research. Moreover, un- like software (Mowery, 1996), where the federal government played a prominent role in developing computer science as an academic field, in data storage the private sector initiated the establishment of academic programs specifically for magnetic recording, although the federal government then stepped forward with critical funding. These programs also emerged much later than those targeted at the computer, semiconductor, and software industnes. REFERENCES Anderson, P. (1995). "Microcomputer manufacturers" in Organizations in Industry: Strategy, Struc- ture and Selection, Glenn R. Carroll and Michael T. Hannan, eds. New York: Oxford University Press. Aoki, M. (1988). Information, Incentives and Bargaining in the Japanese Economy. New York: Cam- bridge University Press. Aoki, M. (1990). "Toward an economic model of the Japanese firm." Journal of Economic Literature 28(March): 1-27. Aoki, M., and N. Rosenberg. (1987). "The Japanese firm as an innovating institution." CEPR Discus- sion Paper No. 106. Stanford University. Bashe, C., L. Johnson, J. Palmer, and E. Pugh. (1986). "Disk storage," in IBM's Early Computers. Cambridge, MA: MIT Press. BT (Business Times). (May 3, 1993). "Plugging into and Asian gold mine." BT (Business Times). (April 1, 1997). "NSTB spins off seven research institutes." 2. BW (Business Week). (February 6, 1984). "The disk-drive boom has suppliers spinning." 68. BWI (Business Wire). (September 12, 1997). "Maxtor to enter server disk drive market." BWI (Business Wire). (June 17, 1997). "Fujitsu ranked the fastest growing hard drive manufacturer." CDSN (Computer Data Storage Newsletter). (1996). Vol. 9, no 10. Mike Cannon, President & CEO of Maxtor: "We'll be profitable in 1997." 1. Chandler, A. (1990). Scale and Scope: The Dynamics of Industrial Capitalism, Boston: Belknap Press. Christensen, C.M., and J.L. Bower. (1996). "Customer power, strategic investment, and the failure of leading firms," Strategic Management Journal 17:197-218. Christensen, C.M., and R.S. Rosenbloom. (1995). "Explaining the attacker's advantage: Technological paradigms, organizational dynamics, and the value network," Research Policy 24:147-162. CI (Computergram International). (January 13, 1992). "Thailand: Fujitsu slashed local production of small hard drives as prices collapse." COM (COMLINE Daily News Computers). (October 9, 1995). NEC hard disk drive plant in Luzon now operational. COM (COMLINE Daily News Computers). (April 15, 1988). "Competition Increasing Between Japa- nese and U.S. Magnetic Disk Drive Makers."
HARD DISK DRIVES 327 CRN (Computer Reseller News). (December 4, 1995). "Western Digital's Burger speaks out on state of storage industry." 133. CRN (Computer Reseller News). (February 24, 1997). "Will emerging new digital technologies ever stop?" CW (Computerworld). (December 9, 1985). "'Made in Japan' tag penetrating components market: U.S.-Japan pacts create added equipment sales." 64. Dertouzos, M. L., R. K. Lester, and R. M. Solow. (1989). Made in America: Regaining the Productive Edge, Cambridge, MA: MIT Press. Disk/Trend. (Various years). Disk/Trend Report: Rigid Disk Drives. Mountain View, CA: Disk/Trend, Inc. EBN (Electronic Buyer's News). (April 29, 1996). "MR technology may finally have its year." BET (Electronic Engineering Times). (August 26, 1996). "MR heads inch toward mainstream." EN (Electronic News). (October 22, 1984). "Peripheral manufacturers making move to SE Asia; cheaper labor, materials draw U.S. firms seeking to be more price-competitive." 53. EN (Electronic News). (August 21, 1989). "Seagate/Imprimis deal forces industry shift." 1. FT (Financial Times). (June 1, 1988). "Seagate sets sail for Singapore." 16. FW (Financial World). (February 24, 1987). "Disk drive technology is changing so quickly that only a few U.S. companies can keep up with it. But those survivors should prosper and soon." 124. Florida, R., and M. Kenney. (1990). The Breakthrough Illusion: Corporate America's Failure to Move from Innovation to Mass Production, New York: Basic Books. Gerlach, M.L. (1992). "The Japanese corporate network: A blockmodel analysis," Administrative Science Quarterly 37: 105 - 139. Gourevitch, P., R. Bohn, and D. McKendrick. (1997). "Who Is Us? The Nationality of Production in the Hard Disk Drive Industry," The Data Storage Industry Globalization Project Report 97-01. San Diego, CA: Graduate School of International Relations and Pacific Studies, U.C. San Di ego. Harker, J.M., D.W. Brede, R.E. Pattison, G.R. Santana, and L.G. Taft. (1981). "A Quarter Century of Disk File Innovation," IBM Journal of Research and Development 255:677-689. Henderson, R., and K. Clark. (1990). "Architectural innovation: The reconfiguration of existing prod- uct technologies and the failure of established firms," Administrative Science Quarterly 35:9- 30. Hill, C.W.L. (1995). "National institutional structures, transaction cost economizing and competitive advantage: The case of Japan," Organization Science 6:119-131. IDC (ID C Japan Report). (February 28, 1991). "Fujitsu will produce small hard disk drives in Thai- land starting this summer." Imai, K., I. Nonaka, and H. Takeuchi. (1985). "Managing the product development process: How Japanese companies learn and unlearn," in The Uneasy Alliance: Managing the Productivity- Technology Dilemma, Kim Clark et al., eds. Boston: Harvard Business School Press. KEW (The Korea Economic Weekly). (October 31, 1996). Samsung Electronics opens a HDD plant. Kogut, B., ed. (1993). Country Competitiveness: Technology and the Organizing of Work, New York: Oxford University Press. Langlois, R. N. (1992). "External economies and economic progress: The case of the microcomputer industry," Business History Review 66: 1-50. LAT (Los Angeles Times). (June 25, 1990). "Why American high-tech firm recruits in Asian ricefields." D3. LAT (Los Angeles Times). (August 6, 1991). "Toshiba develops disk drive of high capacity." D2. Mansfield, E. (1988). "The speed and cost of industrial innovation in Japan and the United States: External vs. internal technology," Management Science 34:1157-1168. McKendrick, D., and A. Hicken. (1997). "Global strategy and population level learning in the hard disk drive industry," The Data Storage Industry Globalization Project Report 97-05. San Diego, CA: Graduate School of International Relations and Pacific Studies, U.C. San Diego.
328 U.S. INDUSTRYIN2000 Mowery, D. (1996). "Spinning off and spinning on?: The federal government role in the development of the US computer software industry," Research Policy 25:947-966. MR (Market Reports). (July 19, 1991). "Korea - disk storage devices." NST (New Straits Times). (April 25, 1996). "Quantum makes new move in drive business." Odagiri, H., and A. Goto. (1993). "The Japanese system of innovation: Past, present and future," in National Innovation Systems, R. R. Nelson, ed. New York: Oxford University Press. Okimoto, D., and Y. Nishi. (1994). "R&D organization in Japanese and American semiconductor firms," in The Japanese Firm: Sources of Competitive Strength, M. Aoki and R. Dore, eds. Oxford: Oxford University Press. Porter, M.E. (1990). The Competitive Advantage of Nations. New York: Free Press. Quantum. (1997). "Storage resource center on the web." Available: www.quantum.com/src/storage_ basics. Rabinow, J. (1952). "The notched-disk memory," Electrical Engineering (August):745-749. Rhyne, V.T. (1996). Congressional testimony on behalf of the National Research Council's Panel for Electronics and Electrical Engineering, the Board on Assessment of National Institute of Stan- dards and Technology NIST Programs. June 25, 1996. SoS (Scotland on Sunday). (September 29, 1996). "Patents pay-off pending." 1. Stevens, L.D. (1981). "The evolution of magnetic storage," IBM Journal of Research and Develop- ment 255:663-675. Teece, D.J. (1992). "Competition, cooperation, and innovation: Organizational arrangements for re gimes of rapid technological progress," Journal of Economic Behavior and Organization 18:1 25. TrendFocus. (1996a). Rigid Media Information Service. Annual Study. Palo Alto, CA: TrendFocus, Inc. TrendFocus. (1996b). HDD Recording Head Information Service. Annual Study. Palo Alto, CA: TrendFocus, Inc. Van de yen, A., and R. Garud. (1989). "A framework for understanding the emergence of new indus tries," in Research on Technological Innovation, Management and Policy, R. Rosenbloom and R. Burgelman, eds. Greenwich, CT: JAI Press.