The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail
- Description
- Product Details
- About the Author
- Read an Excerpt
ISBN-13: 9781633691780
Media Type: Paperback
Publisher: Harvard Business Review Press
Publication Date: 01-05-2016
Pages: 288
Product Dimensions: 5.40(w) x 8.20(h) x 1.00(d)
Clayton M. Christensen is the Kim B. Clark Professor of Business Administration at Harvard Business School. He is the author of eight critically acclaimed books, including the bestsellers The Innovator’s Solution, How Will You Measure Your Life?, and Disrupting Class. Christensen is the cofounder of Innosight, a management consultancy; Rose Park Advisors, an investment firm; and the Innosight Institute, a nonprofit think tank.
Chapter One How Can Great Firms Fail? When I began my search for an answer to the puzzle of why the best firms can fail, a friend offered some sage advice. "Those who study genetics avoid studying humans," he noted. "Because new generations come along only every thirty years or so, it takes a long time to understand the cause and effect of any changes. Instead, they study fruit flies, because they are conceived, born, mature, and die all within a single day. If you want to understand why something happens in business, study the disk drive industry. Those companies are the closest things to fruit flies that the business world will ever see." Indeed, nowhere in the history of business has there been an industry like disk drives, where changes in technology, market structure, global scope, and vertical integration have been so pervasive, rapid, and unrelenting. While this pace and complexity might be a nightmare for managers, my friend was right about its being fertile ground for research. Few industries offer researchers the same opportunities for developing theories about how different types of change cause certain types of firms to succeed or fail or for testing those theories as the industry repeats its cycles of change. This chapter summarizes the history of the disk drive industry in all its complexity. Some readers will be interested in it for the sake of history itself. But the value of understanding this history is that out of its complexity emerge a few stunningly simple and consistent factors that have repeatedly determined the success and failure of the industry's best firms. Simply put, when the best firms succeeded, they did so because they listened responsively to their customers and invested aggressively in the technology, products, and manufacturing capabilities that satisfied their customers' next-generation needs. But, paradoxically, when the best firms subsequently failed, it was for the same reasons--they listened responsively to their customers and invested aggressively in the technology, products, and manufacturing capabilities that satisfied their customers' next-generation needs. This is one of the innovator's dilemmas: Blindly following the maxim that good managers should keep close to their customers can sometimes be a fatal mistake. The history of the disk drive industry provides a framework for understanding when "keeping close to your customers" is good advice--and when it is not. The robustness of this framework could only be explored by researching the industry's history in careful detail. Some of that detail is recounted here, and elsewhere in this book, in the hope that readers who are immersed in the detail of their own industries will be better able to recognize how similar patterns have affected their own fortunes and those of their competitors. HOW DISK DRIVES WORK Disk drives write and read information that computers use. They comprise read-write heads mounted at the end of an arm that swings over the surface of a rotating disk in much the same way that a phonograph needle and arm reach over a record; aluminum or glass disks coated with magnetic material; at least two electric motors, a spin motor that drives the rotation of the disks and an actuator motor that moves the head to the desired position over the disk; and a variety of electronic circuits that control the drive's operation and its interface with the computer. See Figure 1.1 for an illustration of a typical disk drive. The read-write head is a tiny electromagnet whose polarity changes whenever the direction of the electrical current running through it changes. Because opposite magnetic poles attract, when the polarity of the head becomes positive, the polarity of the area on the disk beneath the head switches to negative, and vice versa. By rapidly changing the direction of current flowing through the head's electromagnet as the disk spins beneath the head, a sequence of positively and negatively oriented magnetic domains are created in concentric tracks on the disk's surface. Disk drives can use the positive and negative domains on the disk as a binary numeric system--1 and 0--to "write" information onto disks. Drives read information from disks in essentially the opposite process: Changes in the magnetic flux fields on the disk surface induce changes in the micro current flowing through the head. EMERGENCE OF THE EARLIEST DISK DRIVES A team of researchers at IBM's San Jose research laboratories developed the first disk drive between 1952 and 1956. Named RAMAC (for Random Access Method for Accounting and Control), this drive was the size of a large refrigerator, incorporated fifty twenty-four-inch disks, and could store 5 megabytes (MB) of information (see Figure 1.2). Most of the fundamental architectural concepts and component technologies that defined today's dominant disk drive design were also developed at IBM. These include its removable packs of rigid disks (introduced in 1961); the floppy disk drive (1971); and the Winchester architecture (1973). All had a powerful, defining influence on the way engineers in the rest of the industry defined what disk drives were and what they could do. As IBM produced drives to meet its own needs, an independent disk drive industry emerged serving two distinct markets. A few firms developed the plug-compatible market (PCM) in the 1960s, selling souped-up copies of IBM drives directly to IBM customers at discount prices. Although most of IBM's competitors in computers (for example, Control Data, Burroughs, and Univac) were integrated vertically into the manufacture of their own disk drives, the emergence in the 1970s of smaller, nonintegrated computer makers such as Nixdorf, Wang, and Prime spawned an original equipment market (OEM) for disk drives as well. By 1976 about $1 billion worth of disk drives were produced, of which captive production accounted for 50 percent and PCM and OEM for about 25 percent each. The next dozen years unfolded a remarkable story of rapid growth, market turbulence, and technology-driven performance improvements. The value of drives produced rose to about $18 billion by 1995. By the mid-1980s the PCM market had become insignificant, while OEM output grew to represent about three-fourths of world production. Of the seventeen firms populating the industry in 1976--all of which were relatively large, diversified corporations such as Diablo, Ampex, Memorex, EMM, and Control Data--all except IBM's disk drive operation had failed or had been acquired by 1995. During this period an additional 129 firms entered the industry, and 109 of those also failed. Aside from IBM, Fujitsu, Hitachi, and NEC, all of the producers remaining by 1996 had entered the industry as start-ups after 1976. Some have attributed the high mortality rate among the integrated firms that created the industry to its nearly unfathomable pace of technological change. Indeed, the pace of change has been breathtaking. The number of megabits (Mb) of information that the industry's engineers have been able to pack into a square inch of disk surface has increased by 35 percent per year, on average, from 50 Kb in 1967 to 1.7 Mb in 1973, 12 Mb in 1981, and 1100 Mb by 1995. The physical size of the drives was reduced at a similar pace: The smallest available 20 MB drive shrank from 800 cubic inches ([in..sup.3]) in 1978 to 1.4 [in..sup.3] by 1993--a 35 percent annual rate of reduction. Figure 1.3 shows that the slope of the industry's experience curve (which correlates the cumulative number of terabytes (one thousand gigabytes) of disk storage capacity shipped in the industry's history to the constant-dollar price per megabyte of memory) was 53 percent--meaning that with each doubling of cumulative terabytes shipped, cost per megabyte fell to 53 percent of its former level. This is a much steeper rate of price decline than the 70 percent slope observed in the markets for most other microelectronics products. The price per megabyte has declined at about 5 percent per quarter for more than twenty years. THE IMPACT OF TECHNOLOGICAL CHANGE My investigation into why leading firms found it so difficult to stay atop the disk drive industry led me to develop the "technology mudslide hypothesis": Coping with the relentless onslaught of technology change was akin to trying to climb a mudslide raging down a hill. You have to scramble with everything you've got to stay on top of it, and if you ever once stop to catch your breath, you get buried. To test this hypothesis, I assembled and analyzed a database consisting of the technical and performance specifications of every model of disk drive introduced by every company in the world disk drive industry for each of the years between 1975 and 1994. This database enabled me to identify the firms that led in introducing each new technology; to trace how new technologies were diffused through the industry over time; to see which firms led and which lagged; and to measure the impact each technological innovation had on capacity, speed, and other parameters of disk drive performance. By carefully reconstructing the history of each technological change in the industry, the changes that catapulted entrants to success or that precipitated the failure of established leaders could be identified. This study led me to a very different view of technology change than the work of prior scholars on this question had led me to expect. Essentially, it revealed that neither the pace nor the difficulty of technological change lay at the root of the leading firms' failures. The technology mudslide hypothesis was wrong. The manufacturers of most products have established a trajectory of performance improvement over time. Intel, for example, pushed the speed of its microprocessors ahead by about 20 percent per year, from its 8 megahertz (MHz) 8088 processor in 1979 to its 133 MHz Pentium chip in 1994. Eli Lilly and Company improved the purity of its insulin from 50,000 impure parts per million (ppm) in 1925 to 10 ppm in 1980, a 14 percent annual rate of improvement. When a measurable trajectory of improvement has been established, determining whether a new technology is likely to improve a product's performance relative to earlier products is an unambiguous question. But in other cases, the impact of technological change is quite different. For instance, is a notebook computer better than a mainframe? This is an ambiguous question because the notebook computer established a completely new performance trajectory, with a definition of performance that differs substantially from the way mainframe performance is measured. Notebooks, as a consequence, are generally sold for very different uses. This study of technological change over the history of the disk drive industry revealed two types of technology change, each with very different effects on the industry's leaders. Technologies of the first sort sustained the industry's rate of improvement in product performance (total capacity and recording density were the two most common measures) and ranged in difficulty from incremental to radical. The industry's dominant firms always led in developing and adopting these technologies. By contrast, innovations of the second sort disrupted or redefined performance trajectories--and consistently resulted in the failure of the industry's leading firms. The remainder of this chapter illustrates the distinction between sustaining and disruptive technologies by describing prominent examples of each and summarizing the role these played in the industry's development. This discussion focuses on differences in how established firms came to lead or lag in developing and adopting new technologies, compared with entrant firms. To arrive at these examples, each new technology in the industry was examined. In analyzing which firms led and lagged at each of these points of change, I defined established firms to be those that had been established in the industry prior to the advent of the technology in question, practicing the prior technology. I defined entrant firms as those that were new to the industry at that point of technology change. Hence, a given firm would be considered an entrant at one specific point in the industry's history, for example, at the emergence of the 8-inch drive. Yet the same firm would be considered an established firm when technologies that emerged subsequent to the firm's entry were studied. SUSTAINING TECHNOLOGICAL CHANGES In the history of the disk drive industry, most technology changes have sustained or reinforced established trajectories of product performance improvement. Figure 1.4, which compares the average recording density of drives that employed successive generations of head and disk technologies, maps an example of this. The first curve plots the density of drives that used conventional particulate oxide disk technology and ferrite head technology; the second charts the average density of drives that used new-technology thin-film heads and disks; the third marks the improvements in density achievable with the latest head technology, magneto-resistive heads. The way such new technologies as these emerge to surpass the performance of the old resembles a series of intersecting technology S-curves. Movement along a given S-curve is generally the result of incremental improvements within an existing technological approach, whereas jumping onto the next technology curve implies adopting a radically new technology. In the cases measured in Figure 1.4, incremental advances, such as grinding the ferrite heads to finer, more precise dimensions and using smaller and more finely dispersed oxide particles on the disk's surface, led to the improvements in density from 1 to 20 megabits per square inch (Mbpsi) between 1976 and 1989. As S-curve theory would predict, the improvement in recording density obtainable with ferrite/ oxide technology began to level off toward the end of the period, suggesting a maturing technology. The thin-film head and disk technologies' effect on the industry sustained performance improvement at its historical rate. Thin-film heads were barely established in the early 1990s, when even more advanced magneto-resistive head technology emerged. The impact of magneto-resistive technology sustained, or even accelerated, the rate of performance improvement. Figure 1.5 describes a sustaining technological change of a very different character: an innovation in product architecture, in which the 14-inch Winchester drive is substituted for removable disk packs, which had been the dominant design between 1962 and 1978. Just as in the thin-film for ferrite/oxide substitution, the impact of Winchester technology sustained the historically established rate of performance improvement. Similar graphs could be constructed for most other technological innovations in the industry, such as embedded servo systems, RLL and PRML recording codes, higher RPM motors, and embedded interfaces. Some of these were straightforward technology improvements; others were radical departures. But all had a similar impact on the industry: They helped manufacturers to sustain the rate of historical performance improvement that their customers had come to expect. In literally every case of sustaining technology change in the disk drive industry, established firms led in development and commercialization. The emergence of new disk and head technologies illustrates this. In the 1970s, some manufacturers sensed that they were reaching the limit on the number of bits of information they could pack onto oxide disks. In response, disk drive manufacturers began studying ways of applying super-thin films of magnetic metal on aluminum to sustain the historical rate of improvements in recording density. The use of thin-film coatings was then highly developed in the integrated circuit industry, but its application to magnetic disks still presented substantial challenges. Experts estimate that the pioneers of thin-film disk technology--IBM, Control Data, Digital Equipment, Storage Technology, and Ampex--each took more than eight years and spent more than $50 million in that effort. Between 1984 and 1986, about two-thirds of the producers active in 1984 introduced drives with thin-film disks. The overwhelming majority of these were established industry incumbents. Only a few entrant firms attempted to use thin-film disks in their initial products, and most of those folded shortly after entry. The same pattern was apparent in the emergence of thin-film heads. Manufacturers of ferrite heads saw as early as 1965 the approaching limit to improvements in this technology; by 1981 many believed that the limits of precision would soon be reached. Researchers turned to thin-film technology, produced by sputtering thin films of metal on the recording head and then using photolithography to etch much finer electromagnets than could be attained with ferrite technology. Again, this proved extraordinarily difficult. Burroughs in 1976, IBM in 1979, and other established firms first successfully incorporated thin-film heads in disk drives. In the period between 1982 and 1986, during which some sixty firms entered the rigid disk drive industry, only four (all commercial failures) attempted to do so using thin-film heads in their initial products as a source of performance advantage. All other entrant firms--even aggressively performance-oriented firms such as Maxtor and Conner Peripherals--found it preferable to learn their way using conventional ferrite heads first, before tackling thin-film technology. As was the case with thin-film disks, the introduction of thin-film heads entailed the sort of sustained investment that only established firms could handle. IBM and its rivals each spent more than $100 million developing thin-film heads. The pattern was repeated in the next-generation magneto--resistive head technology: The industry's largest firms--IBM, Seagate, and Quantum--led the race. The established firms were the leading innovators not just in developing risky, complex, and expensive component technologies such as thin-film heads and disks, but in literally every other one of the sustaining innovations in the industry's history. Even in relatively simple innovations, such as RLL recording codes (which took the industry from double- to triple-density disks), established firms were the successful pioneers, and entrant firms were the technology followers. This was also true for those architectural innovations--for example, 14-inch and 2.5-inch Winchester drives--whose impact was to sustain established improvement trajectories. Established firms beat out the entrants. Figure 1.6 summarizes this pattern of technology leadership among established and entrant firms offering products based on new sustaining technologies during the years when those technologies were emerging. The pattern is stunningly consistent. Whether the technology was radical or incremental, expensive or cheap, software or hardware, component or architecture, competence-enhancing or competence-destroying, the pattern was the same. When faced with sustaining technology change that gave existing customers something more and better in what they wanted, the leading practitioners of the prior technology led the industry in the development and adoption of the new. Clearly, the leaders in this industry did not fail because they became passive, arrogant, or risk-averse or because they couldn't keep up with the stunning rate of technological change. My technology mudslide hypothesis wasn't correct. FAILURE IN THE FACE OF DISRUPTIVE Most technological change in the disk drive industry has consisted of sustaining innovations of the sort described above. In contrast, there have been only a few of the other sort of technological change, called disruptive technologies. These were the changes that toppled the industry's leaders. The most important disruptive technologies were the architectural innovations that shrunk the size of the drives--from 14-inch diameter disks to diameters of 8, 5.25, and 3.5-inches and then from 2.5 to 1.8 inches. Table 1.1 illustrates the ways these innovations were disruptive. Based on 1981 data, it compares the attributes of a typical 5.25-inch drive, a new architecture that had been in the market for less than a year, with those of a typical 8-inch drive, which at that time was the standard drive used by minicomputer manufacturers. Along the dimensions of performance important to established minicomputer manufacturers--capacity, cost per megabyte, and access time--the 8-inch product was vastly superior. The 5.25-inch architecture did not address the perceived needs of minicomputer manufacturers at that time. On the other hand, the 5.25-inch drive had features that appealed to the desktop personal computer market segment just emerging in the period between 1980 and 1982. It was small and lightweight, and, priced at around $2,000, it could be incorporated into desktop machines economically. Generally disruptive innovations were technologically straightforward, consisting of off-the-shelf components put together in a product architecture that was often simpler than prior approaches. They offered less of what customers in established markets wanted and so could rarely be initially employed there. They offered a different package of attributes valued only in emerging markets remote from, and unimportant to, the mainstream. The trajectory map in Figure 1.7 shows how this series of simple but disruptive technologies proved to be the undoing of some very aggressive, astutely managed disk drive companies. Until the mid-1970s, 14-inch drives with removable packs of disks accounted for nearly all disk drive sales. The 14-inch Winchester architecture then emerged to sustain the trajectory of recording density improvement. Nearly all of these drives (removable disks and Winchesters) were sold to mainframe computer manufacturers, and the same companies that led the market in disk pack drives led the industry's transition to the Winchester technology. The trajectory map shows that the hard disk capacity provided in the median priced, typically configured mainframe computer system in 1974 was about 130 MB per computer. This increased at a 15 percent annual rate over the next fifteen years--a trajectory representing the disk capacity demanded by the typical users of new mainframe computers. At the same time, the capacity of the average 14-inch drive introduced for sale each year increased at a faster, 22 percent rate, reaching beyond the mainframe market to the large scientific and supercomputer markets. Between 1978 and 1980, several entrant firms--Shugart Associates, Micropolis, Priam, and Quantum--developed smaller 8-inch drives with 10, 20, 30, and 40 MB capacity. These drives were of no interest to mainframe computer manufacturers, which at that time were demanding drives with 300 to 400 MB capacity. These 8-inch entrants therefore sold their disruptive drives into a new application--minicomputers. The customers--Wang, DEC, Data General, Prime, and Hewlett-Packard--did not manufacture mainframes, and their customers often used software substantially different from that used in mainframes. These firms hitherto had been unable to offer disk drives in their small, desk-side minicomputers because 14-inch models were too big and expensive. Although initially the cost per megabyte of capacity of 8-inch drives was higher than that of 14-inch drives, these new customers were willing to pay a premium for other attributes that were important to them--especially smaller size. Smallness had little value to mainframe users. Once the use of 8-inch drives became established in minicomputers, the hard disk capacity shipped with the median-priced minicomputer grew about 25 percent per year: a trajectory determined by the ways in which minicomputer owners learned to use their machines. At the same time, however, the 8-inch drive makers found that, by aggressively adopting sustaining innovations, they could increase the capacity of their products at a rate of more than 40 percent per year--nearly double the rate of increase demanded by their original "home" minicomputer market. In consequence, by the mid-1980s, 8-inch drive makers were able to provide the capacities required for lower-end mainframe computers. Unit volumes had grown significantly so that the cost per megabyte of 8-inch drives had declined below that of 14-inch drives, and other advantages became apparent: For example, the same percentage mechanical vibration in an 8-inch drive, as opposed to a 14-inch drive, caused much less variance in the absolute position of the head over the disk. Within a three-to-four-year period, therefore, 8-inch drives began to invade the market above them, substituting for 14-inch drives in the lower-end mainframe computer market. As the 8-inch products penetrated the mainframe market, the established manufacturers of 14-inch drives began to fail. Two-thirds of them never introduced an 8-inch model. The one-third that introduced 8-inch models did so about two years behind the 8-inch entrant manufacturers. Ultimately, every 14-inch drive maker was driven from the industry. The 14-inch drive makers were not toppled by the 8-inch entrants because of technology. The 8-inch products generally incorporated standard off-the-shelf components, and when those 14-inch drive makers that did introduce 8-inch models got around to doing so, their products were very performance-competitive in capacity, areal density, access time, and price per megabyte. The 8-inch models introduced by the established firms in 1981 were nearly identical in performance to the average of those introduced that year by the entrant firms. In addition, the rates of improvement in key attributes (measured between 1979 and 1983) were stunningly similar between established and entrant firms. Held Captive by Their Customers Why were the leading drive makers unable to launch 8-inch drives until it was too late? Clearly, they were technologically capable of producing these drives. Their failure resulted from delay in making the strategic commitment to enter the emerging market in which the 8-inch drives initially could be sold. Interviews with marketing and engineering executives close to these companies suggest that the established 14-inch drive manufacturers were held captive by customers. Mainframe computer manufacturers did not need an 8-inch drive. In fact, they explicitly did not want it: they wanted drives with increased capacity at a lower cost per megabyte. The 14-inch drive manufacturers were listening and responding to their established customers. And their customers--in a way that was not apparent to either the disk drive manufacturers or their computer-making customers--were pulling them along a trajectory of 22 percent capacity growth in a 14-inch platform that would ultimately prove fatal. Figure 1.7 maps the disparate trajectories of performance improvement demanded in the computer product segments that emerged later, compared to the capacity that changes in component technology and refinements in system design made available within each successive architecture. The solid lines emanating from points A, B, C, D, and E measure the disk drive capacity provided with the median-priced computer in each category, while the dotted lines from the same points measure the average capacity of all disk drives introduced for sale in each architecture, for each year. These transitions are briefly described below. Read an Excerpt
Insights from the Hard Disk
Drive Industry
TECHNOLOGICAL CHANGES
Table 1.1 A Disruptive Technology Change: The 5.25-inch Winchester Disk Drive (1981) 8-Inch Drives 5.25-Inch Drives Attribute (Minicomputer Market) (Desktop Computer Market) Capacity (megabytes) 60 10 Physical volume (cubic inches) 566 150 Weight (pounds) 21 6 Access time (milliseconds) 30 160 Cost per megabyte $50 $200 Unit cost $3000 $2000 Source: Data are from various issues of Disk/Trend Report.