¹ A description of how disk drives work, as well as definitions of technical terms used in this history, are included in the Appendix. Because few data were available about the industry prior to the publication of Disk/Trend Report, most of the statistical analyses employed in this article begin in 1976.

² In this context, “architecture” refers to the system that defines the way in which computer components interact with each other. See Henderson, Rebecca M. and Clark, Kim B., “Architectural Innovation: The Reconfiguration of Existing Systems and the Failure of Established Firms,” Administrative Science Quarterly, March 1990, 9–30.CrossRefGoogle Scholar

³ Details about the role that the captive disk drive operations of IBM and Control Data played in developing many of the key technologies used in the OE market industry are recounted in a companion paper. See Clayton M. Christensen, “Industry Maturitv and the Vanishing Rationale for Industrial Research and Development,” Harvard Business School Working Paper, 1993.

⁴ This is consistent with the importance of corporate headquarters activities in industries studied in Porter, Michael, The Competitive Advantage of Nations (New York, 1991).Google Scholar

⁵ This section's information about IBM's pioneering work was drawn from James Engh, “The IBM Diskette and Diskette Drive”; J. M. Harker, et al., “A Quarter Century of Disk File Innovation”; and L. D. Stevens, “The Evolution of Magnetic Storage” — all in the twenty-fifth anniversary issue of the IBM journal of Research and Development 25 (Sept. 1981). Other useful information sources were H. V. Bordwell, “Cornerstone of a Division,” Reflections (a periodical publication of IBM's Santa Teresa Laboratory), June 1984, 6–11; “The IBM 350 RAMAC Disk File,” American Society of Mechanical Engineers, Santa Clara Valley Section, Feb. 1984; “How One Company's Zest for Technological Innovation Helped Build the Computer Industry,” IBM Corporation, San Jose Calif., 1984; “Disk Storage Technology,” IBM Corporation, San Jose, Calif., 1980; a long and delightful personal interview with Mr. Reynold Johnson, head of the IBM team that developed the first disk drive, 5 May 1992, in Palo Alto, Calif.; and personal interviews with twelve other earlv members of IBM's disk drive team.

⁶ The term “Winchester” was the name of IBM's project to develop the Model 3340. The name was chosen by the project's manager, who owned a 30–30 Winchester rifle. These numbers matched the objectives originally specified for the 3340 project, to develop a drive with 30 megabytes each of fixed and removable capacity. Other industry participants subsequently borrowed the term for their sealed-system drives, and “Winchester” joined the ranks of cellophane and nylon as a generic name for a category of products. James Porter, editor of Disk/Trend Report, interview with author, October 1991, Mountain View, Calif.

⁷ Each of these LA-area firms had been acquired by the mid-1970s by a larger firm—Century by Calcomp, and then Xerox; ISS by Univac; Pertee by Adler; Wangco bv Perldn Elmer; and Kennedy by Allegheny Ludlum. Under the acquirors' management, Pertee, Wangco, ISS, and Kennedy evaporated rather quickly. Century hung on with roughly flat revenues (and dramatically declining market share) until 1988, when it was finally closed by Xerox.

⁸ The dollar figures for captive shipments shown in Table 1 and all subsequent tables in this article have been adjusted to reflect OE market, rather than retail pricing, levels. This enables a clearer comparison of market and captive activity.

⁹ Histories of the venture capital and IPO market financing activities of these start-ups can be found in Sahlman, William and Stevenson, Howard, “Capital Market Myopia,” Journal of Business Venturing 1 (1983): 7–30.CrossRefGoogle Scholar

¹⁰ Three firms treated as related-market firms in this study entered the industry by acquiring start-up firms. Between 1969 and 1973 Data 100, Electronic Memories, and Calcomp acquired lomee, Caelus, and Century Data, respectively. lomee made disk-pack products; Caelus made primarily the disk packs themselves, along with a few drives; and Century made fixed-disk drives. In 1988 Western Digital, a controller manufacturer (and therefore classed as a forward integrator when it entered in 1988), acquired the floundering disk drive operations of Tandon, which had entered on a related-technology basis.

¹¹ The dimensions of relatedness among the different activities of diversified firms have been extensively studied by industrial economists. Prominent among these are Wrigley, Leonard, “Divisional Autonomy and Diversification” (Unpub. DBA diss., Harvard University Graduate School of Business Administration, 1970)Google Scholar, which offers a taxonomy of diversification strategies. MacDonald, James M., “R&D and the Directions of Diversification,” Review of Economics and Statistics, 1985, 583–90CrossRefGoogle Scholar, discusses the relationship between a firm's R&D strategy and patterns of diversification. The related-market and related-technology rationale for diversification were first identified by Rumelt, Richard, Strategy, Structure and Economic Performance (Cambridge, Mass., 1974), 17.Google Scholar

¹² The term “independent” here refers to firms that were not vertically integrated into computer manufacturing. In Japan all independent firms were either related-technology or related-market firms; there were no start-ups. There was one European start-up, Rodime, founded in Scotland by engineers defecting from Burroughs Corp. Rodime pioneered the 3.5-inch drive and for a time was quite successful, with revenues exceeding $100 million. It withdrew from the market in 1991. No attempt was made in this history to sort Japanese or European firms into related-market, related-technology, or forward integrator categories.

¹³ Chandler, Alfred D. Jr, Strategy and Structure: Chapters in the History of the Industrial Enterprise (Cambridge, Mass., 1962);Google Scholar Chandler, and, The Visible Hand: The Managerial Revolution in American Business (Cambridge, Mass., 1977).Google Scholar

¹⁴ The peripherals division of Control Data Corporation was by far the largest vertically integrated manufacturer in the OE market, and the 1989 shift in market share from the vertically integrated to the start-up categories reflects Seagate Technology's purchase of the Control Data disk drive operations in 1989.

¹⁵ The original Quantum Corporation was founded in 1979 to make 8-inch drives. It was one of the most successful 8-inch drive makers, but it missed the 5.25-inch generation almost completely. As its revenues were evaporating in 1986–87, Quantum merged a partially owned subsidiary, Plus Development Corporation, back into the parent company; canceled all of its 8- and 5.25-inch production arrangements, and used Plus's 3.5-inch diameter “Hardcard” architecture as the basis for a new business. The executives of Plus became the executives of Quantum. These executives, as well as most industry observers, consider this to have marked the closing of the “old” Quantum, and the birth of an essentially new company.

¹⁶ The unusually steep rate of price decline measured by the experience curve seems, in part, to be due to the substitution of Winchester-architecture for removable disk-pack architecture between 1980 and 1985—when the individual points on Figure 1 decline most rapidly. For drives of equivalent capacities, cost per MB of a Winchester drive was typically 30 percent lower than that of removable-disk architecture. It appears that the experience curves within removable-disk and Winchester architectures followed approximately a 70 percent slope, which is typical for such curves in the electronic components industry.

¹⁷ Henderson, and Clark, , “Architectural Innovation,” 9–30.Google Scholar

¹⁸ In industry parlance, the different sizes of Winchester disk drives are called the 14-inch form factor, the 8-inch form factor, etc. An example of the architectural uniqueness of these form factors is that in the 8-inch drive, a 110-volt AC motor was typically positioned in the corner of the system, and it drove the disks by pulleys and a belt. In reducing the size to 5.25 inches, the motor was changed to a 12-volt, DC, flat “pancake” design and positioned beneath the spindle. Such a rearrangement in the wav components relate to each other, where the fundamental technological concepts of magnetic recording on rotating disks powered by electric motors are preserved, is the essence of Henderson and Clark's definition of “architectural” innovation. The architectures listed here were those that came to be broadly adopted in large segments of the market. Many other architectures were introduced, and the manufacturers of some of these products became commercially successful in niche segments of the market.

¹⁹ See, for example, Schumpeter, Joseph A., The Theory of Economic Development (Cambridge, Mass., 1934)Google Scholar; Cooper, A. and Schendel, D., “Strategic Responses to Technological Threats,” Business Horizons 19 (Feb. 1976)Google Scholar; Mansfield, Edwin, et al., The Production and Application of New Industrial Technology (New York, 1977)Google Scholar; and Foster, Richard, Innovation: The Attacker's Advantage (New York, 1986).CrossRefGoogle Scholar

²⁰ Tushman, Michael and Anderson, Philip, “Technological Discontinuities and Organizational Environments,” Administrative Science Quarterly 31 (1986): 439–65.CrossRefGoogle Scholar

²¹ These codes are essentially “markers” recorded by the head on the disk to denote the start of a new piece of data. The markers used in early disk drives consumed a significant portion of the available recording area on the disk. As a consequence, engineers worked to find more efficient marker systems that preserved data integrity but consumed less storage area. The development of RLL codes is one such innovation.

²² Cooper and Schendel, “Strategie Responses to Technological Threats,” and Foster, Innovation: The Attacker's Advantage, are two highly influential studies that take the general viewpoint that radical innovation tends to come from new firms.

²³ See Christensen, Clayton M., “The Innovator's Challenge: Understanding the Influence of Market Environment on Processes of Technology Development in the Rigid Disk Drive Industry” (Unpub. DBA thesis, Harvard University Graduate School of Business Administration, 1992)Google Scholar, chap. 7.

²⁴ See Foster, Innovation: The Attacker's Advantage; and Clayton Christensen, “Exploring the Limits of the Technology S-Curve, Part 1: Component Technologies”; and “Exploring the Limits of the Technology S-Curve, Part 2: Architectural Technologies,” Production and Operations Management 1 (Fall 1992): 334–66.

²⁵ The concept of technological trajectories was first introduced in Dosi, Giovanni, “Technological Paradigms and Technological Trajectories,” Research Policy 11 (1982): 147–62,CrossRefGoogle Scholar

²⁶ The parallelism in the trajectories of capacity improvement across the 8-, 5.25-, and 3.5-inch architectures seems to have occurred because assemblers of each of these architectural generations had reasonable access to the same improvements in basic component technologies. Because component technology improvements were the engine of performance improvement within a given architectural paradigm, one might expect the trajectories to be parallel.

²⁷ These trajectories of capacity demanded were calculated by plotting the hard disk capacity shipped with the median-priced computer system in each market category for each year, and then fitting a best-fit regression line through those points. The trajectories of capacities that the technology was able to supply within each architecture were calculated by determining the average capacity of all models introduced in each year in each architectural form factor, and then using regression analysis to calculate the equation of the best-fit line through them. Details of these procedures can be found in Christensen, “The Innovator's Challenge,” which also describes the trajectories in capacity demanded in the engineering workstation market segment.

²⁸ A similarly detailed account of each architectural transition is contained in Christensen, “The Innovator's Challenge.”

²⁹ This finding is consistent with the observations of Burgelman, who noted that one of the greatest difficulties encountered by corporate entrepreneurs was finding the right “beta test sites,” where products could be interactively developed and refined with customers. Generally, the entré to the customer was provided by the salesperson who sold the firm's established product lines. This helped the firm develop new products for established markets, but did not help it identify new applications for its new technology. See Burgelman, Robert and Sayles, Leonard, Inside Corporate Innovation (New York, 1986), 76–80.Google Scholar

³⁰ Carter O'Brien, former Seagate executive vice-president of marketing, interview with author, 4 May 1992, Scotts Valley, Calif.

³¹ Voice coil motors could position the head much more accurately over tracks on the disk than could stepper motors, since they operated in a continuous mode and could be used with closed-loop servo systems. Voice coil motors were not new to the market when Seagate adopted them in 1984, but they represented a major shift in Seagate's design philosophy.

³² This issue is covered in detail in chapter 7 of Christensen, “The Innovator's Challenge.” The primary issue investigated there deals with this question: If groups of entrant and established firms were dealt exactly the same set of components, would one group of firms consistently design higher-performance drives than the other group for a given level of component technology? In other words, might it be possible that established firms' engineers were somehow locked into an obsolete way of thinking about system design, so that their products were not as efficient in extracting performance out of a given set of components as entrant firms' engineers might have been? The results showed that, though there were consistent differences among firms in the architectural efficiency of their drives, there was no statistically significant difference between established and entrant firms' architectural efficiencies.

³³ By 1988, because the 3.5-inch drive had begun to encroach on Seagate's desktop 5.25-inch product sales, Seagate's revenues began to stagnate. Its executives responded by acquiring the disk drive operations of Control Data in 1989. Seagate then combined its volume manufacturing expertise with Control Data's advanced component technologies (such as thin-film head manufacturing) to forge a strong market position in the rapidly growing engineering workstation market. Most industry observers credit the acquisition as having saved the company, and I would agree. By 1993, Seagate had essentially been driven into a weak number three position in its original market—personal computing—by the two firms that pioneered the 3.5-inch drive, Conner Peripherals and Quantum Corporation. Seagate's corporate strength in the early 1990s derives almost exclusively from its Control Data acquisition.

³⁴ Vertically integrated disk drive manufacturers were not considered in the analysis that follows, on the assumption that the architectural technologies they employed in disk drives were largely determined by their strategies in downstream computer businesses. For example, IBM's decision of when to launch 5.25-ineh and 3.5-inch architectures was determined by when the firm's personal computer business needed such drives, rather than being independent, strategic decisions made by the management of IBM's disk drive operation.

³⁵ The 8-, 5.25-, 3.5-, and 2.5-inch Winchester architectures on which this study focuses were the architectural innovations that penetrated major portions of the market. There were, however, many more innovative architectures introduced by firms that never became commercially successful, or that, though successful, remained confined in a relativelv small market niche. An example of the former was a head-per-track 14-inch drive that Alpha Data attempted to sell for over a decade. The 3.9-inch removable-cartridge drive introduced by Syquest in 1982 was a commercially successful architecture that has remained in a relatively small niche market for back-up desk-top devices. Maxtor's effort to pack more capacity into the standard full-height 5.25-inch form factor by building the motor into the spindle, so that additional disks could be stacked on the spindle, was another architectural innovation of the non-form-factor sort. These innovations, though they did not capture large market shares, were nonetheless considered as new architectures in the analysis reported here, because each addressed new, emerging market segments.

³⁶ The founders of those four start-ups had no prior direct experience in the disk drive industry.

³⁷ Shugart itself was founded in 1973 to manufacture 8-inch floppy drives and, by all accounts, was the primary driver behind the creation of the OE market for the floppy disk drive industry. When its sales had reached $17 million in 1977, it was acquired by Xerox for $40 million. Xerox guided Shugart into rigid drives in 1977, and although the foray generated about $400 million in cumulative revenues, it was never profitable. Xerox closed Shugart down in 1985.

³⁸ The $9.143 billion in cumulative revenues shown in Figure 7 for Shugart's progenitors does not include the approximately SI billion in revenues booked bv Seagate in 1989 from its acquisition of Control Data's disk drive operations.

³⁹ An initial version of this history can be found in Christensen, “Industry Maturity and the Vanishing Rationale for Industrial Research and Development.”

⁴⁰ Although none of the IBM engineers or executives interviewed for this history cited antitrust pressure as a force that kept IBM from selling its components in the OE market, many outside observers believe that IBM was reluctant to sell key components in the open market for fear of fueling the U.S. Department of Justice's antitrust suit against the company, pending at this time.