The Future of the Microprocessor Industry”
Total Page:16
File Type:pdf, Size:1020Kb
MASSACHUSETTS INSTITUTE OF TECHNOLOGY SLOAN SCHOOL OF MANAGEMENT 15.912 Technology Strategy Professor Rebecca Henderson “The Future of the Microprocessor Industry” Final Paper Juan Chaia Paulo Marchesan Bernardo Neves Cambridge, Massachusetts. May 11th, 2005 15.912 Technology Strategy Massachusetts Institute of Technology Professor Rebecca Henderson Sloan School of Management EXECUTIVE SUMMARY Intel has been one of the most successful companies in modern corporate history. They are the clear leader in the microprocessor industry, in which they have set the pace of technological advance in the past three decades. They were able to do this because of the uniqueness of its technology at the beginning, and the development of strong complementary assets, namely manufacturing expertise and branding, later on. As a consequence, Intel has been able to capture a significant portion of the value created by the microprocessor industry. However, the electronic microprocessor technology is reaching maturity, and may be subject to a disruption within the next two decades. In this paper, we predict that such disruption may come from microphotonics. Microphotonics technology, which very crudely uses photons for the transmission and processing of data, has been on the spotlight for at least a decade. According to experts from MIT, it may be ready to be used on commercial chips in a decade. Some large companies around the world, such as Pirelli, IBM, Lucent and others, are already making big bets that this will be the next chip technology. Our paper microphotonics analyzes different scenarios that the industry leader, Intel, may face if indeed microphotonics turns out to be the disruptive technology in the microprocessor industry. For each potential scenario, we give suggestions of strategic actions that the company may take. Under some scenarios, Intel should vertically integrate; in others, it should license the technology to avoid a supplier hold-up problem. In any case, it is a clear conclusion for us that if this technology becomes disruptive, Intel should be the first to market or it will face serious trouble surviving. This means tracking very closely the evolution of microphotonics, and being prepared to invest heavily to be ahead when it is mature enough, and to potentially change Intel’s current R&D structure. On the bright, there is still a window of opportunity for Intel until the technology is ready for commercial applications. On the not so bright one, Intel is facing a classical dilemma between diverting R&D resources to a promising yet unproven technology and devoting them in full to the exhaustive exploitation of its current technology, which still has some earnings’ generation potential ahead. 1. INTRODUCTION The key questions were are trying to answer in this paper are what the future of the microprocessor industry will look like, what will most likely be the next disruptive technology, and above all “What should Intel do?”, considering the dynamic and competitive environment in which it may operate in the future. While this paper is not intended to be a comprehensive and exhaustive discussion about either the theory of technology strategy or the microprocessor industry, in the ensuing sections we hope to shed some light onto questions such as: “When and how should Intel play the game?”; “How do the industry structure and value chain are likely to change?”; “How should the R&D process be structured?”, etc. The Future of the Microprocessor Industry 1 15.912 Technology Strategy Massachusetts Institute of Technology Professor Rebecca Henderson Sloan School of Management 2. INDUSTRY HISTORY AND EVOLUTION Since the appearance of microprocessors in 1971, the number of transistors (the building blocks of these integrated circuits) per unit area has been doubling every 1.5 years. This has translated into a significant increase in the clock speed of the chips. Clock speed is measured in megahertz (MHz), and it determines how many instructions per second the processor can execute2. After decades of dramatic improvement in performance, the microprocessor industry appears to be entering the mature stage. The evolutionary stage of an industry can be assessed by plotting the performance evolution of its products over time, according several performance dimensions. Perhaps the best way to measure performance in the microprocessor industry is in terms of clock speed (see graph below). The graph shows that microprocessors’ performance evolution turns out to be a classical S-Curve1. (see figure 1, appendix 2). As the technology approaches its natural technological limit, we can expect longer periods of time between generations of chips with lower incremental performance, as well as the appearance of a disruptive technology – which may significantly affect the industry, and Intel in particular. One of the reasons behind the exhaustion of a particular technology is that it starts losing functionality for the customer. In the case of the microprocessor, the advance in performance (clock speed) becomes less important for customers as successive generations of chips do not improve significantly the perceived performance of their computers and their applications. Technological Leaps In Micro Processing As history has already shown, the electronic/computer industry is very prone to disruptive innovations. The industry has already gone through three major technological revolutions. Up until this point, processors have evolved from vacuum tubes to transistors to integrated circuits and, finally, to microprocessors. From the first microprocessor - invented in 1971 – to today’s CPUs (such as the Pentium 4 microprocessor) the clock speed has increased more than 10,000 times. Natural Technological Limits in the Microprocessor Industry For several years now, scientists and computer experts have been arguing about the limits in processing power. There are 2 main factors that limit computer chip speed: transmission delays and heat build-up on the chip (please refer to the Appendix 1 for definitions). A few years ago, some experts thought that the speed limit using the existing technology would be around 1 Ghz. However, refrigerating technologies, as well as the use of asynchronous logic and parallel processing has enabled incremental speed increases. 1 It is important to note that in the x-axis we are using time instead of effort. On one hand, data about the amount of investment in R&D was not readily available. On the other hand, time can be used as a proxy for effort in the context of this industry, since competition and demand for more computing power force companies to launch chips as soon as possible, and thus the time between a generation of microprocessors and the next is a reflection of the amount of resources devoted to the development, testing, improvement, and marketing of each type of chip The Future of the Microprocessor Industry 2 15.912 Technology Strategy Massachusetts Institute of Technology Professor Rebecca Henderson Sloan School of Management However, as clockspeed increases at a lower rate questions continued to be raised about the technological exhaustion of microprocessors as we know them. From Intel’s perspective that leads to a critical question: How can Intel anticipate and react to disruption?. 3. DYNAMICS THAT SHAPED THE INDUSTRY STRUCTURE Value creation vs. value capture Intel invented the microprocessor in 1971, and has become since the undisputable industry leader. The company’s products established the industry standard, in large part due to the most significant design win in the company’s history: IBM chose an Intel 16-bit microprocessor – the 8086, for its IBM PC in 1980. From then on, by investing significantly in applied R&D, Intel has been able to shape the game and stay ahead2. With technology advancement, Intel has been able to reap significant profits for some time3. While, historically, Intel has extracted rents from having a technological lead in microprocessors, its most important strategic move consisted in developing a branding strategy that de-commoditized its products Intel’s main competitor, AMD, started in the microprocessor business as Intel’s partner. In 1982, IBM demanded a second source in order to use the Intel 8088 in its IBM PC, and therefore Intel signed AMD as a licensed second-source manufacturer. The agreement was cancelled in 1986; AMD started to produce clones of Intel’s processors and in 1993 Intel challenged AMD’s right to produce those, but ultimately lost its case. AMD plays a very different game than Intel’s. The company invests much less in R&D, but enough to follow Intel’s moves – delayed by a few months. It prices lower and keeps a small market share, knowing that Intel will try to avoid a price war if they remain small. Recently AMD has managed to grow its market share to 16%– which, despite the growth, is still much smaller than Intel’s 82% market share. Uniqueness and Complementary Assets Following the invention of the microprocessor in 1971, Intel was able to capture value because of its uniqueness and technological lead. In the early years, not only the patents, but also the secrecy about its manufacturing processes protected Intel’s uniqueness, and allowed the company to avoid competition. Intel – and AMD in some sense – was also able to develop important complementary assets throughout the years, enabling the firm to translate its lead into substantial long-term value capturing. As illustrated in Figure 2 in Appendix 2 players in this industry still compete in uniqueness, but they have migrated to a form of competition equally based in complementary assets, such as branding and excellence in manufacturing. 2 Intel has consistently invested between 9% and 15% of sales on R&D between 1992 and 2003 (please refer to Table 1 in Appendix 2 for a comparison of R&D spending among selected firms). 3 For example, until the introduction of a clone 386 chip by the competition, Intel had a 100% market share for the 386 series microprocessors.