The Future of the Microprocessor Industry

“The Future of the Microprocessor Industry” 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. 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. 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. The “Intel Inside” campaign – launched in 1992 – is a milestone in the development of what is perhaps Intel’s strongest complementary asset – its brand. The campaign4 re-shaped consumer perception, making Intel’s chip “the” key component of their computers, thus allowing Intel to charge a price premium over its competitors and significantly increasing its power along the value chain. Alongside with branding, state-of-the-art manufacturing processes that drastically increase manufacturing yields, have allowed microprocessor producers – Intel in particular – to capture a significant share of the value created by the technology. Suffice it to say that in the last two years, Intel’s gross margins have been around 56% and AMD’s around 36%, while the gross margins of successful computer manufacturers, such as Dell, have been closer to 18%. While Intel and AMD still keep a significant edge in the technology, it has become much hard to compete in the “uniqueness” dimension. The speed of introduction of faster/better microprocessors has significantly increased since 1971, as depicted in the previously shown S-curve for Intel. In order to keep their lead, firms have been increasing their R&D expenditures, which per se reduces their profitability. Current Structural Characteristics and the Nature of Competition The microprocessor industry is a very attractive industry to be in. Barriers to entry are high (huge sunk costs, competition in uniqueness and complementary assets); buyer power is rather low (particularly for Intel that has developed branding); supplier power is very low (essentially commoditized raw materials), currently there are no clear substitutes; and rivalry is rather controlled (not very aggressive in price) – please refer to Appendix 3 for a more detailed analysis of the industry under Porter’s 5 Forces framework However, the industry structure is not static and the actions of the different players deeply influence its evolution. For instance, Intel strategic moves have not only shaped the nature of competition in the industry but also changed the industry structure as a whole. While the development of complementary assets has slightly softened the R&D competition, it did not change the fact that the microprocessor industry is still competing, at least partially, in “uniqueness”. Gaining market share gives the players in the industry a competitive edge by enabling them to invest more in the development of new products, which eventually feedbacks into gaining even more market share (a Product Development reinforcing loop). The ability Intel has demonstrated to be constantly ahead of the game (always being the first to develop the next generation) is, at the same time, derived from and critical to maintain its market share dominance. Additionally, by maintaining the technological lead, Intel has been able to grab a large share of the market, thereby consolidating its market power – especially over its buyers. The technological lead and bargaining 15.912 Technology Strategy Massachusetts Institute of Technology Professor Rebecca Henderson Sloan School of Management power – fruit of the complementary assets intelligently developed – reinforces Intel’s positioning, reduces its cost of capital, and fuels even further the ability of the company to keep innovating (Figure 3, Appendix 2). Economies of scale and learning effects also explain the strong first-mover advantage we observe in the industry. Since the production of the next generation of microprocessors requires huge investments (mostly sunk costs) and learning, being first to market triggers important reinforcing (self-inflicting) loops which give the “pioneer” an even bigger competitive advantage – further economies of scale and learning that make really hard to competitors to catch up. The Name of the Game - Technology The two most important strategic games being played in the industry involve Microprocessor Manufacturers (“Intel vs. AMD”) and PC manufacturers (“Microprocessor vs. PC manufacturers”). In both games, the critical strategic variables are “technology related”. Even tough one could argue the existence of a “capacity game”, the decision about how much production capacity to build is subject to the result of the technological race. The first company to develop the next generation decides how much to produce and the second player has to make his decision constrained by the first mover’s decision. Therefore, the capacity game is a game played within a bigger one: “the technology game”. Intel vs. PC Manufacturers Intel’s choice to develop an open or closed standard for its first microprocessors has been part of an interesting strategic game. In the late 70’s, a pivotal sale to IBM's new personal computer division made the 8088 the brains of IBM's new hit product and propelled Intel into the Fortune 500 list. IBM, however, was aware of the risks of being held hostage (over-dependent on one supplier) and, in order to win the deal, Intel was forced to commit to open the standards of the microprocessors. However, as history has shown, this was not an irreversible commitment and generations later, Intel adopted a private/closed standard, significantly increasing its market power (Figure 4, Appendix 2). Intel did not only play the game; it changed the game. By building complementary assets such as the “Intel Inside” brand, Intel has become less vulnerable to the technological risks, at the same time, increasing its bargaining power and consequently its margins. Intel vs. AMD The technology game played between Intel and its competitors (the most important being AMD) is particularly relevant given the relative homogeneity of the demand for microprocessors. A better processor is likely to be a “better processor” for every user in the market, meaning that this new product could win a huge market share. We find evidence of this homogeneity in the fact that our PCs today have much more powerful processors than what most of our regular tasks require. In this game, winning the technological race is very important because the players are competing for the whole market. Since product innovation is a strategic complement, players are likely to respond to increases in competitors’ R&D expenses by increasing their R&D expenses too – exactly what can be observed in the industry. The relatively high homogeneity of demand influences not only the game being currently played but it also makes the whole industry very susceptible to technological disruptions. 4. THE LEADER As has been discussed throughout the paper, Intel got a clear advantage in the microprocessor industry, first by moving from open standards to close standards, and then by building a brand strategy. This allowed Intel to avoid commoditization and has given the company an edge in sustaining profitability and basically deciding the direction in which the industry will move. In addition to these strategic moves, another key component of Intel’s success has been a distributed model of R&D, which has historically allowed them to benefit from outside research without losing their focus on excellence in manufacturing. This section of the paper will focus on discussing Intel’s performance so far, as well as to a brief look at the key components of their R&D model, so that we can later discuss whether it is sustainable in the face of a potential technology disruption (like microphotonics). Intel’s Performance If we take a look at the last decade (1995-2004), Intel’s financial performance has been solid, with an average compounded growth rate in revenues of 8.7%, and an average profit margin of 21.2%. Investment in their R&D model has also been heavy, typically in the range of 8-14% of sales, as shown in Summary table below. SUMMARY TABLE. Intel’s Key Financial Figures From 1995 to 2004 Million dollars Net revenue COGS (% sales) R&D (% sales) Net income Profit margin 1995 $ 16,202 48.2% 8.0% $ 3,566 22.0% 1996 $ 20,847 44.0% 8.7% $ 5,157 24.7% 1997 $ 25,070 39.7% 9.4% $ 6,945 27.7% 1998 $ 26,273 46.0% 9.5% $ 6,068 23.1% 1999 $ 29,389 40.3% 10.6% $ 7,314 24.9% 2000 $ 33,726 37.5% 11.6% $ 10,535 31.2% 2001 $ 26,539 50.8% 14.3% $ 1,291 4.9% 2002 $ 26,764 50.2% 15.1% $ 3,117 11.6% 2003 $ 30,141 43.3% 14.5% $ 5,641 18.7% CAGR1995-2004 2004 $ 34,209 8.7% -1.4% 42.3% 6.4% 14.0% $ 7,516 8.6% 0.0% 22.0% As can be seen, Intel’s dominance of the market has paid off financially, and its revenues have continuously grown, and the company historically presents solid profit margins in the range of 22% to 31%, with the clear exception of the difficult period that the technology industry experienced in 2001. Then, the dot-com collapse, together with the clear drift of the world’s economy into a recession (aided to great extent by the terrorist attacks), produced a significant decrease in revenues (21%) but particularly in profit (88%). Despite this fact, we can see that the company has nicely recovered and by 2004 Intel was back again in revenue levels similar to those of 2000, with a profit margin of 22%. This last point is significant in the sense that we see an erosion of Intel’s profitability, given that with similar sales in 2000 their profit margin was 31%. This is due to a great extent to the company’s higher investment in R&D, which is currently at 14% of sales, as a way to respond to the increasing pressures presented by the competition, namely AMD. Intel’s R&D process5 Since its origins as a company in the early 1970’s, Intel decided that its core competencies would be around manufacturing, despite the natural dependency of the industry on research breakthroughs. The company’s “Copy Exactly” philosophy, that allows easy transfer of internal developments from laboratory to fabs, standardizes processes, reduces costs from leverage with suppliers, and sustains manufacturing quality, is almost a religion for insiders. In addition, their “minimum information” principle, which limits idea generation to what’s specifically required to solve a particular problem, also ensures a high capture ratio of the ideas generated inside the company and limits the number of spin-offs from employees. These philosophies together are focused on making the advancements in research directly and easily transferable to the manufacturing environment. Intel’s approach to R&D is built around a global, “distributed” model, in which linkages are established and promoted between the internal labs and the external research community, and that emphasizes a collaborative approach in developing new technologies, leading standards initiatives, and influencing regulatory policy to accelerate the adoption of new technologies. The internal R&D activities are performed in specialized laboratories focused on particular activities, such as chip architecture, microcomputers and components research, etc. These labs report directly to specific product or process organization, so as to ensure the aforementioned direct link between R&D and manufacturing. In the external arena, Intel organizes research conferences and forums each year to bring together internal and external scientists working on cutting-edge research, and has been publishing for several years now the Intel Technical Journal, to publish some of the results of these works. In addition, Intel funds several external research projects on a continuous basis, and works closely with industry researchers, scientists, engineers and students on fields related to its activity, including some of the best universities around the globe. Finally, the company has an active equity investment program that allows it to acquire participation on early stage startups that are developing technologies that may be useful for Intel in the future. Intel’s R&D activities are focused on world-class technology development, particularly in the areas of design and manufacturing of integrated circuits. The financial commitment of the company to R&D is directed towards developing technology innovations, primarily at the silicon level, of what they think may be the next generation of their products. In this sense, Intel’s efforts are increasingly centered around platforms, particularly focused in the areas of advanced computing, communications, and wireless technologies, while continuing to invest in new manufacturing, testing, and packaging processes. Moreover, the company has been involved in a substantial effort in product development outside the U.S., having established facilities in several countries, including Israel, Malasya, China, India and Russia. 5. THE FUTURE There is too much ambiguity about the future of the industry to artificially recommend Intel a deterministic strategy. However, from the innumerous dimensions of uncertainty, three are particularly relevant for designing Intel’s tech strategy: buyer power, technological disruption and technological lead. Technological Disruption in the Microprocessor Industry The understanding that demand homogeneity makes the microprocessor industry very susceptible to technological disruptions, bring us to the next question: “Is a disruption in microprocessors likely to happen in the near future?” On one hand, most experts agree that a disruption will happen soon and that the emerging winner is likely to derive the advent of microphotonics technology. Micro Photonics – the experts claim – will transcend current natural limits, since the transmission of light eliminates heat generation and transmission delays. If this technology can be perfected, processing speed can increase many fold. On the other hand, many analysts, including several Intel’s researchers, believe that the current technology can be improved at least another 10 years. In a nutshell, while most experts agree on the eventuality of a technology disruption, there is great uncertainty about its timing – to the point that its effect in the industry may not be relevant. Industry Structure & Value Change Disruption While elements such as homogeneity of demand, strong learning curves and economies of scale are intrinsic to the microprocessor industry, the industry structure results from a combination of forces very susceptible to change. Companies left out of the industry by the well devised Intel strategy, are making substantial investments in order to leapfrog Intel’s technological lead. Even today, firms like IBM, Sony and Toshiba have publicly unveiled details of a new microprocessor called Cell that supposedly has 10-times the performance capacity of Intel’s current chips, and may potentially substitute them in the near future. Other firms, like Pirelli , IBM, Lucent, Nortel Networks and several startups, are indeed betting in microphotonics and investing heavily to be first to market if this disruption materializes. On the other hand, the PC manufacturers are likely to fight hard – maybe together – for more market power 6 in order to increase their share of PIE . The potential emergence of a dominant PC manufacturer, perhaps Dell, could alter the bargaining power balance in the value chain. It would be much more diffult for players in the microprocessor industry to build complementary assets, as Intel has done with its branding campaign (Figure 3, Appendix 4). Technological Lead As we discussed earlier, technology is the nature of competition in the industry. The technological competition not only affects the rivalry in the industry but also affects the whole industry structure. Whether Intel ends up being a leader or a follower deeply impacts Intel’s tech strategy in either technological disruption scenarios. What Does the Future Look Like? Picturing future scenarios according the three dimensions discussed above results in the eight scenarios described in exhibit 1 bellow. While all of these are feasible scenarios, some are more likely to happen (1, 2, 3 and 4), some are of particular interest from a strategic perspective (5, 7 and 8) and one is very improbable – reason for which it is called “Dell’s Dream”. Given the main uncertainties surrounding the future of the industry a contingent approach to strategy, based on a scenario analysis, is used in the next section to address the question of “What should Intel do?”. In particular, scenarios 2, 3, 4, 5, 7 and 8 are discussed in more detailed given their relevance in designing Intel’s technology strategy. SCENARIOS 1. It can’t get better 4 2. Official Threat 1 3. Brave New World Several Buyers 4. Tech Game 2 3 5. Moving Down in the Food Chain 6. Dell’s Dream 7. Intel Outside 8 8. Fight of Titans 5 Leader One Big 7 6 “Dell” Micro Follower No disrupt Photonics Exhibit 1 – Scenario Analysis 6. WHAT SHOULD INTEL DO? In recommending Intel’s strategy we will consider the potential future scenarios to answer three questions: (1) where in the value chain Intel should play, (2) what can Intel do to maximize its probability of success after the next disruption in the industry and (3) how should it structure its R&D efforts in order to remain successful in the new technology order. The first question is critical in order to understand if expected changes in industry structure might make vertically integrating attractive for Intel. We analyzed this potential strategic move for a few potential future scenarios. The rest of the chapter is focused in Intel’s largest challenge to come. If most analysts are correct the most significant disruption in the industry since vacuum tubes is at most 20 years away. What will be Intel’s challenges and what should the company do about them in order to survive, if possible at all? Last, but not the least, given Intel’s challenges and the actions that it must take to overcome them, how should Intel implement its strategy? How should they organize their R&D structure? How can they keep the focus on the current business but still leverage their capabilities to transition to the new technology? How do they build new necessary capabilities? Should Intel Vertically Integrate?7 At this point in time, Intel has no incentives to vertically integrate, mainly because: 1. There are no evident synergies to be achieved by forward integrating, nor does the double marginalization greatly affect the chip maker’s profits. 2. Chip makers don’t have the experience to compete in efficiency with a company like Dell, and the other PC manufacturers (potential targets for an acquisition) have higher costs, with some even exiting the industry (e.g. IBM). 3. From a strategic standpoint, we can see that chip manufacturers don’t have a strong dependence on any particular PC assembler, nor should they be worried at this moment about an increase in buyer power that requires the creation of artificial competition. However, Intel should pay attention to a few changes in the industry structure that might make it interesting to integrate forward. ¾ “The Official Threat” (Scenario 2): a joint-venture of an IBM, Motorola et al. to produce microprocessors was recently announced. This is a risk for incumbents in the industry – particularly Intel – since the entry of technical capable players may lead the industry to “commoditization”. A third player may destabilize the “follow the leader” game being current played in the industry, breaking the coordination between players and, therefore, leading to the loss of bargaining power against buyers (computer manufacturer). In this scenario, forward integrating would be an alternative to keep a direct communication channel with the customer. Intel could use forward integration to defend its “Intel Inside” brand. The brand equity (complementary asset) would be quintessential for Intel’s defensive strategy (see figure 4, appendix 2). ¾ “Moving Down in the Food Chain” (Scenario 5): The emergence of a market leader with strong complementary assets (supply chain distribution and brand recognition) would significantly alter the power distribution between microprocessor and PC manufacturers. By controlling a significant share of the PC market a player such as Dell would be able to exert pressure over its supplier and capture value that up until now has been captured by Intel or AMD. The alternative for those companies, in this case, would be to forward integrate to i) nurture strong competitors in the market ii) capture the value by avoiding double marginalization. ¾ “Intel Outside” (Scenario 7): Such a disruption may significantly change the industry structure in ways that are very difficult to predict. One of the risks, however, for incumbents is that the new technology will level even more the technological field (homogeneity of demand) and, on top of that, destroy the brand equity nurtured throughout years. In this scenario, there would be very high buyer power and significant rivalry; resulting in a much less attractive industry. Current players would have an incentive, therefore, to i) exit the industry altogether; ii) either forward integrate or move upstream in the value-chain (becoming perhaps a service provider by leveraging their R&D capabilities and their brands (Intel) – similarly to IBM’s recently move). Intel, however, would be in a good position to leverage its complementary assets, particularly its ability to incrementally improve the technology and its brand equity, not only to survive in the industry but also to quickly reshape competition in the same way it has done before – eventually migrating to scenario 8 or scenario 4 . What Does Intel Need To Survive The Microphotonics Disruption? In designing Intel’s strategy careful consideration has to be given to the technological leadership. In this section we discuss the impact of being a leader or a follower in technology development and for each of the most relevant scenarios, we analyzed the challenges that Intel will face and what it can do (if anything) in order to remain successful. If the IP of a new technology works in much the same way as the electronic microprocessor, the technology will probably evolve so fast in ferment the first versions of the technology and their IP will be quickly turned obsolete. Therefore, we believe that IP protection will help generate uniqueness, but only for a short period of time. “Tech Game” (Scenario 4): Intel is first to market with microphotonics technology: This scenario assumes that Intel leads innovation and is able to hold its uniqueness at least in the short term. Even in this scenario, where Intel has uniqueness on its side, it will face certain challenges in order to maintain a high level of value capture. 1.1 Slow innovation of complementors: When microphotonics eliminate limits of processing speed, other computer parts will become bottlenecks of processing power, such as memory, drivers, etc. In fact, memory is already a bottleneck, as clockspeed has been evolving much faster than memory capacity. This bottleneck may slow down the speed of adoption of the new technology, giving extra time for competitors to catch up. The strategy here will depend on the complementor. For example, in the case of memory, since Intel is already a manufacturer, it can develop R&D internally. In other cases, Intel may subsidize R&D in some of the companies. 1.2 Supplier hold-up problem: The complexity of this new technology will probably demand the development of specific components and equipment. If Intel is the only large company who will be producing these chips, suppliers may not be willing to make specific capital investments for fear of being held up later. A contract would be very hard to implement here, so Intel would have two options: develop the R&D internally or a make credible commitment that they won’t hold suppliers up, such as licensing the technology. Again, the decision will depend on each specific case. 1.3 “Fight of Titans” (Scenario 8) - Dell’s power: Dell has become a major player in the computer industry and arguably they are far from the ceiling. Dell might be afraid that with the new technology, Intel will capture an even larger share of the industry’s PIE. Intel has created a reputation for shrewdly getting most of the PIE. This could cause Dell to delay the adoption of the technology. Intel has once helped failing companies downstream to survive in order to keep it competitive. Maybe this will be the moment to repeat the play. As our analysis has shown, even with IP protection, there are significant challenges that Intel will have to face. They are not trivial and will probably imply a reduction in Intel’s share of the PIE compared to the current scenario. However, under this scenario Intel may still be able to overcome these challenges and leverage their key complementary assets, such as their brand, their manufacturing excellence and their capabilities in applied research – the ability to bring innovation quickly to the production lines. The brand will again be an important asset that will give credibility to the new chips, which will be critical in the early adoption phase. The manufacturing excellence will allow it to keep costs low, which reduces the incentives of competitors to enter. “Brave New World” (Scenario 3): Another company is first to market In this case, the main question is whether Intel’s complementary assets will allow it to survive despite another company’s uniqueness advantage at least in the short term. The answer is that probably not. The brand would probably quickly vanish as soon as another company appears with a superior technology (unless it presented significant problems in the beginning). The manufacturing abilities would also be useless if the company had no competitive products to produce. Theoretically, Intel could license the technology from the owner of the IP, in order to stay in the game in the short term as it prepares to uses its manufacturing excellence and its applied research capabilities to eventually leapfrog the new leader. However, given Intel’s reputation and the risk of it becoming again the leader in a second generation of the technology would probably lead the innovator not to license to Intel. In this case, Intel would be in a very complicated situation, and its survival could be jeopardized. Therefore, we conclude that it is critical for Intel to be the first to market with a microphotonics processing chip. This will lead to manageable challenges, which Intel will be able to deal with. The Future of R&D at Intel Since we have shown that it is crucial for Intel to be first to market with microphotonics processing, an appropriate technology strategy has to answer two questions. The first one is how Intel should allocate its R&D efforts between the current technology and the new potential disruptive technology, and the second one is how should those resources be allocated along time in order for Intel to be the first to market. Interestingly, companies investing in microphotonics research are mostly in the telecommunications and microprocessor industry. There seems to be a logical explanation why research is concentrated in this industry. Manufacturers of PC micro-processors have little incentive to develop this technology, since the one currently available delivers performance already way beyond the computing needs of most the consumer base. However, microphotonics technology promises to be the new Holy Grail for the telecommunications industry. There is currently a bottleneck problem in the industry because light signals from optic cables need to be translated into electronic ones, which is what computers and TVs “understand”. This causes a bottleneck problem that reduces the effectiveness of the infrastructure significantly. Installed at first in phone lines and later perhaps in digital appliances themselves, the new-generation chips would help direct traffic on fiberoptic networks, zapping data and video directly into home computers and TV sets at the speed of 1 gigabit per second--more than 10,000 times the 56 kilobits per second that is now standard. That kind of speed could turn desktops at home and in the office into smoothly operating interactive media centers. Although the optical component companies are more interested in specific application for their industry, there is a risk to Intel that once these companies have a working chip for telecommunications equipment, it might be only another stretch until they enter the micro processing business. In fact, the article “Pirelli: On the Trail of the Holy Grail” (Businessweek Online, March 4, 2002) suggests that Pirelli, the largest investor in the technology, might become the next Intel a few decades from now. Intel’s only incentive to invest in this new technology is the threat that other company will do it, since its current position in the market can’t get much better – “It can’t get better” (Scenario 1). Also, Intel has a great incentive to free-ride in technology advancements in telecommunication companies and universities, waiting until it is reasonably well developed to adapt it to micro-processors. On the other hand, as we saw in the previous section, the downside of being second to market is very large. Therefore, Intel has to strike the equilibrium between the cost savings that free riding will allow and the risk of being second to market in the micro-processing innovation. Intel must factor in that it tends to be faster bringing the product from design to production, giving them extra time to catch up if needed. As a matter of fact, the “Brave New World” (Scenario 3) may actually not be as terrible for Intel if the company follows closely the new technology’s developments. We believe that Intel should still spend most of its R&D budget on its current technology making further incremental improvements. Many analysts, including several Intel’s researchers, believe that it will be able to extend the Moore’s law for another 10 years. Intel should only make sure that it is not too behind other potential competitors. Since the development of microphotonics technology demands very large investments, and since it will most likely first make it to telecom applications, it will be very visible which give Intel a better control of when to jump in and embrace the new technology. However, as the technology evolves the incentive for a company to develop an application for microprocessors increases. Any company that is first to market has a chance to lever its uniqueness into the future. Therefore, Intel must follow closely the evolution of the technology in order to decide the timing for a more significant investment in the new technology, since as long as it is first to market it has a great chance to stay at the top of the game. How to structure R&D in the future at Intel? In the previous section we concluded that when the technology is sufficiently evolved, Intel must invest heavily in it in order to gain the necessary knowledge quickly so as to become the first to market in computer chips with microphotonics technology. We believe that Intel should start immediately investing in basic technology and ramp up as the technology evolves. This poses an interesting problem to Intel’s R&D structure. While concerned with anticipating and reacting to what many believe will be the next technological disruption in the industry, Intel cannot afford lose focus on developing its products according to the current technological paradigm. In fact, this is a very typical problem in organizations facing a potential disruption – How and when to move into the next wave without losing its technical leadership and market position in the current technology? We will discuss in this section how this R&D should be structured given that Intel has not developed an expertise in basic research and the company’s culture is built around strong applied and process R&D. There is an additional problem that incentives are difficult to align in these two areas. In many ways, this problem appears to be very similar to the strategic problem Intel has faced when investing in the development of “its” own new standard for photo lithography. In order to start basic research, it is not necessary to have a new R&D department. As mentioned earlier, the skills, culture and incentives will still be focused in incremental improvements of its well-established technology. Therefore, it makes sense – at least initially – to use a structure already in place, such as the one found at universities like MIT. Indeed, such an approach would fit quite well with Intel’s current model of distributed R&D. As the technology evolves and the first applications appear in telecommunications, it will be time for Intel to get more serious about research in microphotonics applied to micro-processing. Intel will have to develop the skills needed for a different and faster pace of technological advancement. The structure in place now does not have the adequate rewards structure, culture and people to develop this technology. The rewards system and the culture need to allow for greater risk taking, and creative thinking. The potential cultural and organizational conflict that rises from the environment required for more basic research suggests that in the long term Intel should have a separate R&D structure for the microphotonics laboratory. Coordination will be specially challenging. On one hand, an independent structure seems more appropriate for the new R&D efforts. One the other hand, it is very important for Intel to be able to leverage its complementary assets, such as manufacturing excellence and the brand. While obviously involving the definition of common goals and incentive, the design of linking mechanisms such as cross-departmental teams, the implementation of a dual R&D structure is very complex and will not be addressed in this paper. 7. CONCLUSION The objective of this paper was to describe the evolution of the microprocessor industry and Intel in particular, analyze what the next potential technological disruption may be, and propose some potential scenarios of how this disruption may take place and what strategic actions Intel may take under each of them. Our sense is that the next big disruption will come from the microphotonics technology. Despite the current prediction that this technology still has at least a decade of perfection ahead, there are several firms investing heavily and betting on being the first to market. Intel has done some research efforts in this area, but still sees no fundamental barrier to extend Moore’s law (in other words, its current technology) into the next decade. In this paper, we argue that Intel should hedge against the risk of being left out of this potential disruption and begin investing today in a close follow-up of the evolution of this new technology, as well as be prepared to invest heavily and move fast if microphotonics shows signs of becoming commercially feasible. Based on our analysis of the potential scenarios, given by who owns the IP and our prediction of the future industry characteristics, it is imperative for Intel to be first to market if microphotonics indeed proves to be the future of microprocessors. APPENDIX 1 – DEFINITIONS Transmission delays Transmission delays occur in the wires that connect things together on a chip. The "wires" on a chip are incredibly small aluminum or copper strips etched onto the silicon. A chip is nothing more than a collection of transistors and wires that hook them together, and a transistor is nothing but an on/off switch. When a switch changes its state from on to off or off to on, it has to either charge up or drain the wire that connects the transistor to the next transistor down the line. Imagine that a transistor is currently "on." The wire it is driving is filled with electrons. When the switch changes to "off," it has to drain off those electrons, and that takes time. The bigger the wire, the longer it takes. As the size of the wires has gotten smaller over the years, the time required to change states has gotten smaller, too. But there is some limit -- charging and draining the wires takes time. That limit imposes a speed limit on the chip. There is also a minimum amount of time that a transistor takes to flip states. Transistors are chained together in strings, so the transistor delays add up. On a complex chip like the G5, there are likely to be longer chains, and the length of the longest chain limits the maximum speed of the entire chip. Heat Build-up Finally, there is heat. Every time the transistors in a gate change state, they leak a little electricity. This electricity creates heat. As transistor sizes shrink, the amount of wasted current (and therefore heat) has declined, but there is still heat being created. The faster a chip goes, the more heat it generates. Heat buildup puts another limit on speed. APPENDIX 2 – GRAPHS AND TABLES Figure 1: The S-Curve EVOLUTION OF INTEL MICROPROCESSORS' PERFORMANCE Maximum Historic Clock Speed (MHz) 4000 Pentium 4 HT 560 Pentium 4 HT Extreme 3500 Pentium 4 HT 3000 Pentium 4 2500 2000 1500 Pentium III 1000 Pentium II 500 8008 8080 8086 4004 486 DX DX4 80286 386 DX Pentium Pentium MMX Pentium Pro 0 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Time (Introduction Date) Figure Figure 2: Basis of competition - evolution of the Microprocessor industry Complementary Assets Weak Uniqueness High Strong •Supplier of components are unknown at the customer level •Non-reliable manufacturing process • Strong patents •Significant technological lead •Strong Brand Intel Inside •Excellence in manufacturing • Technological advancements don’t last long • Several firms compete with similar technologies Low 1: Figure 3: Product Innovation, Market share and Market Power Positive Reinforcing Loop Price + Market Power R + + Sales Market Share R + + Revenue + Product Attractiveness R + + Premium for superior technology Technology Performance + Investment in R&D + Figure 4: Selection of Type of Ownership and Technology Standards Technology Open Closed Ownership Public Private • The adoption of an open-standard technology stimulates the dissemination of the technology • Rivalry is expected to be intense • Enable companies to increase margins to quasimonopolistic levels • Technology disruption offers a great risk to incumbents Figure 5. Forward Integration Niche Strategy Forward Integration (defend the brand) Niche strategy Suppliers Microprocessor Manufacturers Buyers Table 1. R&D expenditures (as % of 2000 revenues) for key industry players FIRM % 3Com Advanced Micro Devices Apple Computer Compaq Computer Intel Micron Technology Palm Silicon Graphics Sun Microsystems Texas Instruments 22.3% 13.3% 7.1% 4.0% 14.9% 12.3% 9.3% 12.0% 11.9% 17.6% Total Average 10.5% Source: The Standard From Bloomberg And Sec Filings Table 2. Operating Margins for PC and Chip Manufacturers (%, 2002) PC manufacturers* (12.30) Gateway 5.70 Dell 9.70 IBM 4.20 HP Chip manufacturers 16.4 Intel 3.8 AMD APPENDIX 3.– PORTER’S FIVE FORCES ANALYSIS OF THE MICROPROCESSOR INDUSTRY The dynamics of the microprocessor Industry structure and the evolution of its value chain have also played an important role in determining the amount of value captured by firms such as Intel and AMD. An industry analysis – according to Porter’s model – offers some insight to understand the ability of these companies to capture value: BARRIERS TO ENTRY • The huge sunk costs involved in build a “fab” and the speed of innovation makes much harder for other firms to enter this market. SUPPLIERS • Raw materials are cheap and easily available. • There are many suppliers of equipments, which fosters competition among them, increasing even more the microprocessor manufacturer’s bargaining power. RIVALRY BUYERS • There is certainly competition in this industry, but it is softened by the constant technological innovations and the development of “branded” processors • However, competition still prevents firms’ from capturing a higher share of the value they create. SUBSTITUTES • There are not clear substitutes for microprocessors. • The development of alternatives to the current technology is one of the biggest threats to the incumbents in the industry. • Relationship has evolved from a configuration in which there were a few huge buyers (as a percentage of the firms production capacity) to another in which microprocessor producers have dozens of clients, each accounting for a much lower share of the firms’ production capacity. • Players such as Intel and AMD have gained market power and increased the value captured. We can infer from the diagram that the current industry structure favors microprocessor manufacturer’s value capture. While competition amongst key players reduces their ability to extract higher profit margins, the evolution of the value chain structure has increased their bargaining power against the buyers. APPENDIX 4 – VALUE CHAIN STRUCTURE IN THE MICROPROCESSOR INDUSTRY Figure 1. Evolution Of Relationship With Buyers Suppliers Microprocessor Manufacturers Buyers Figure 2. The Microprocessor Value Chain Raw material providers Chip manufacturers PC assemblers Wholesalers Retailers Figure 3. Proliferation of PC Manufacturers and Importance of Microprocessor Brand R Importance of Microprocessor Brand Proliferation of -PC Manufacturers R Differentiation Amongst PC brands - + Investment in Marketing (building complementary assets) -