- Email: [email protected]
Integrated Circuits: The Coming Battle
28
Long Range Planning Vol. 12
June 1979
Integrated Circuits: The Coming Battle I. M. Mackintosh*
Computerization has been hailed on the one hand as a liberating force for mankind, to, on the other, a destroyer of the human role in many manufacturing and commercial activities. There is little doubt that the possible impacts of future trends in the computer industry upon commercial and industrial activity will be the subject of continuing debate in the next 25 years. However, many commentators have expressed the view that the 'rate' of change could well be unacceptable and create disturbing imbalances in an already disturbingly imbalanced society. This paper examines the impact of the rapidly changing computer technology upon the computer manufacturing industry itself. These changes are of immense importance since they will affect both the distribution of technology as well as economic power. The author, in a detailed and persuasive analysis, plots the trends of ~future shock'. These trends have impacts not only for the computer industry but for the activities of large scale manufacturing and commercial enterprises throughout the world.
Ever since the inventions of the integrated circuit and the silicon planar process, American semiconductor producers have dominated the world markets for ICs,t especially at the leading-edge of each new generation of technology. This is certainly the case today for products ofLSit complexity. But there are increasing signs that things might be changing. For example, competitive memory products are now being produced in substantial quantities in Japan, and there is a growing interest by European companies in acquiring a stake in U.S. semiconductor companies. A key question, therefore, is whether the global domination by American IC companies will be seriously threatened, particularly as the VLSI era approaches, and it is to an examination of the major facets of this question that this paper is directed. The popular but simplistic view-widely held outside the United States-is that the American domination of this industry has been based primarily on very substantial and *The author is Chairman of Mackintosh Consultants Ltd., 33 Bruton Street, London W1. tFor IC read Integrated Circuit, and for LSI read Large Scale Integration.
continuous financial support from the U.S. government. Whereas this funding has obviously been important, the real foundations of this American success are far more complex and need to be understood in detail before the outcome of the impending intercontinental LSI/VLSI battle can be correctly forecast. Indeed, any prognosis of future structural developments in this industry must begin with an understanding of the principal historical forces which have moulded the industry into its present form. The reason, quite obviously, is that whereas important new strategic influences are now emerging, these historical forces will continue to have significant effect for the foreseeable future. This paper therefore comprises two main parts. The first looks briefly at some aspects of the historical development of the semiconductor industry in general, and ICs in particular, in the U.S., Europe and Japan, giving due attention to the various factors which produced the different pace and direction of development in each geographic region. It is then possible to identify a number of key factors which have led to the current American domination of world IC markets. From this foundation, the second part goes on to examine the relevance of these factors, plus the new strategic forces now gathering momentum, to the future chances of success ofiC producers in each geographical region, and to offer a prognosis of the future capabilities of each of these regions to compete in the VLSI era.
The Origins of the Semiconductor Industry As is well-known, the 1950s were characterized by the germanium transistor first going into high-volume production. Its most immediate application was in cheap, portable radios, and the nation which seized this opportunity most effectively was Japan, possessing at that time the considerable advantage of low labour costs. However, Europe (in particular, Philips) was not left far behind.
Integrated Circuits: The Corning Battle In the U.S., however, the gleam in the electronic industry's collective eye was not caused by radios, largely because of the high American labour costs (the idea of moving labour-intensive assembly operations off-shore had not yet been tried). Instead, the greatest need (what we might call the 'user-pull', as distinct from the 'makerpush' effect) for the transistor was mainly in the defence and aerospace sectors-1957 being the year of the Sputnik-and in the infant computer industry. The demands of these military and industrial sectors for devices of higher performance and reliability thus led in time to the emergence of the silicon transistor and later to integrated circuits-both developed, of course, by American companies. The final effects of these original, basic reactions to the advent of the transistor were as follows: {1) The Europeans and Japanese became strong in germanium technology and in the main types of equipments (i.e. consumer electronic products) which were based at that time on germanium transistors. (2) The Americans became pre-eminent in silicon technology and in the main types of equipment {military, computers, communications, etc.) based on this technology. (3) These distinctive postures, originally taken up 1520 years ago, still pertain today, with the Europeans and Japanese leading by a significant margin in most aspects of consumer electronics and the Americans continuing to dominate every other sector of the electronics industry. What can be learned, therefore, from this briefhistorical review is that industrial synergism-the mutual interdependence of different industrial sectors and, in particular, of the equipment and component sectors of the electronics industry-has been a signiftcant factor in shaping the development of the electronics industry in different geographic regions. This is explored later.
The Principal Strategic Factors Affecting the Development of the International Integrated Circuit Industry With the advantage of this historical perspective, the principal strategic factors which have affected-and in most cases will continue to affect-the development of the global IC industry can be identified. There are, obviously, many such strategic factors, and considerable analysis of data, case histories, etc. is necessary if their relative importance is to be correctly assessed, and this work allows certain principal conclusions to be drawn, although it is obviously impossible to present the mass of detailed evidence on which these conclusions are based. The ~ain strategic factors which need to be discussed are:
29
Support * Government Industrial Synergism * Technological * Market FactorsInnovation * Industrial Structure * Management and People. *
The Role of Governments It is beyond question that, almost from its beginning, the American semiconductor industry as a whole has received substantial and broadly-based support from various U.S. government agencies. It has been estimated1 that this totalled about $900m for R & D alone between 1958 and 1974, representing a subsidy of the cost of American semiconductor innovation to the tune of about $55m per annum (in the terms of, say, 1965 average dollar values). Now, it is clear that financial support from public funds on such a large scale has grossly distorted normal competitive conditions and commercial criteria in this industry, and it is difficult to imagine, therefore, how any non-American nation could succeed without providing support to its own indigenous industry on a comparable scale.
The Benefits of Industrial Synergism The general, theoretical benefits of industrial synergism are familiar. However, in a paper of this type it could be useful to emphasize the particular importance of synergism in the development of the global IC industry. Figure 1 represents the well-known Learning Curves for silicon transistors and ICs, showing the systematic relationship between price and cumulative production experience. From sources such as these it is possible to generate the price trend curves, shown in Figure 2, for three silicon devices of progressively increasing complexity-the diffused silicon transistor, the 'average' IC as defined by the Learning Curve, and the more recent IK-bit dynamic RAM. However, following the example of Moore, 2 if these data are normalized in time so that the prices are related to the number of years since product introduction, the effect is to shift the points for the higher-complexity devices to the left, giving the result illustrated in Figure 3. This shows fairly conclusively that there is a general (and very steep) price-reduction curve which can reasonably be assumed to apply to all kinds of high-volume silicon devices, and this is considered again later. First, however, it is illuminating to examine the changes in the performance: cost ratio of computers over the past 15 years or so. Figure 4 is a reproduction of the original curve developed by Turn (Rand Corporation Report) 3 showing the cost in dollars per million instructions per
Long Range Planning Vol. 12
30
June 1979
100
:;;;:
10
(/)
::) Q)
.g 0..
..... ·c: ;::j
---~--~
-~----~~--~~--~--0.1~----~----~----~~--~----~----_.~8
10
10 6
{Units)! Production
Figure 1. Learning curve
100
100
e1K- bit Dynamic RAM xAverage IC 0 0iffused Silicon Transistor •
• 1K - bit Dynamic RAM x Average IC • Diffused Silicon Transistor
X
X
\•
~ Cll
10
10
=s0
Q)
Cii ...
c.> ·.::
co
"5
•• •
0
Q)
.\:1
a:
0.. Ol
~ 1P (/) Q)
Ol
e ~
<(
1.0
Years
1.0
Years After Product Introduction
Figure 2. Typical price trends of silicon devices
Figure 3. Normalized silicon device price trends
second (MIPs) for high-performa nce general-purpose computers, covering the years 1960-1990. It is necessary to note, however, that there is, so far, only one working version of the Illiac IV computer, and the 1000 MIP and 10,000 MIP Processors have yet to be developed.
Figure 5, therefore, shows Turn's curve redrawn to give greatest weight to actual, well-documen ted cost data, and this is compared with the device price curve shown earlier in Figure 3. The correlation is striking and seems to prove what every IC engineer has always instinctively
Integrated Circuits: The Coming Battle
31
100Mr-------~---------r--------~--------~------~~------~
10M •"Stretch
•'BM709411 • CDC6600
1M
en a..
• CDC 7600
•IBM 360/195
~
...
Q)
a..
• CDC Star
ILLIACIV•~
lOOK
1000 MIP • Processor
~
10K
•
1K~--------~--------~--------L---------L---------L-------~
1970
1900
1980
1990
Year
Figure 4. Cost trends in high-performance general application computers
believed: that the computer industry's spectacular growth has been due mainly to its ability to produce equipment which could compute at ever-increasing speeds and reliability levels, and ever-decreasing cost and size, and that essentially all of these attributes have stemmed from advances in silicon technology.
100M
100
• x • o
1K - bit Dynamic RAM Average IC Diffused Silicon Transistor High Performance Computer Costs
X\ '§ .!!! 0
10
0
0
~ Cl
c:
Q) (/) Q)
Cl Q)
>
\
a.. ~
'~\
a..
X~,.~
Q)
~
\\
10M
1.0
Cii ~
Computiog Co"'
f(.,..' ' ~
Slllcoo Devke Price Trends
.!!!
0 0
1M
0
~ •
'-
• lOOK
Years (Normalised)
Figure 5. Comparison of computing costs and silicon device costs
But there is another side to the coin. Figure 6 shows the growth of the total U.S. consumption of monolithic silicon ICs over the period 1964-1976, with digital ICs indicated separately. Since most of these digital ICs, needless to say, were necessarily used in computing equipment of all kinds (including military), the conclusion is inescapable. Just as the American computer industry's growth has been critically dependent on the availability of ever-increasing numbers of ever-improving ICs, so has the spectacular growth of the American IC industry depended to a very high degree on having a large, innovative and 'local' computer market avid to make use of its rapidly-developing semiconductor capabilities. This must be regarded as a particularly convincing example of the benefits of industrial synergism and leaves no doubt that the simultaneous American domination of the integrated circuit, computer and professional electronics sectors are all part of the same basic phenomenon, and that this is the main-not the only, but the mainreason for today's domination by the United States of the world-wide IC business.
32
Long Range Planning Vol. 12
June 1979
1400
X
X
1200
In Europe and Japan, however, the process ofin11ovation in the field of advanced semiconductor components has been hindered by the relative absence until recently of innovative 'user-pull' markets and, especially in Europe, by too much emphasis on basic research and too little on development and marketing. A recent development of enormous significance, and an excellent example of the innovative strengths of the American IC industry, is the microprocessor. Its strategic importance stems mainly from its great commonality of application, as illustrated in Figure 7, thus allowing
X X
High
SSI Microprocessor
X
r--
Digital
1
c
0 ·.;:;
co .!::!
X
Q.
I I
c.
<(
0
X
X/
.~ a;
·-·
/ ·-·-1. ~·~·--r
1964
1966
1968
c 0
E E 0
1970
1972
1974
1976
u
Year
Figure 6. U.S. consumption of monolithic integrated circuits
And the corollary, of course, is that the absence until recently of such synergetic user industries outside America has been the principal reason for the early lack of success of the European and Japanese IC producers.
The Impact of Technological Innovation It is clear that innovation has been, and will for the foreseeable future remain, a major strategic factor in the growth. of the international semiconductor industry. Analysis of case histories shows, however, that the key elements in the total process of innovation are development and marketing-not basic research. In fact, there has been no correlation whatsoever in the U.S. between the commercial success of an IC company and the quality of its basic research programme, an ability to recruit key personnel having historically been much more important. Nevertheless, disciplined in-house development of processes and products has been, and will remain, a key factor in any semiconductor company's success, but for product development work to be relevant it is essential for the company to compete actively in the world's most innovative markets-wherever they may be-for those products.
Low L - - - - - - - - - - - - - - - ; - ; : - : ; - High Low IC Complexity
Figure 7. Complexity versus commonality LSI products the opportunity to break out of the vicious circle of greater complexity: fewer applications: higher cost. It is clear that the microprocessor offers as big a step forward for digital systems as did the original integrated circuit, and it is symptomatic that in this product area Europe as yet has an almost insignificant capability, whereas Japan is already beginning to produce microprocessors, albeit on a modest scale.
Market Factors In light of earlier comments on industrial synergism, it is clear that access to large and innovative (user-pull) markets is a key factor influencing success in the IC business. Today, both Japan and Europe have finally developed large and innovative markets in the consumer electronics sector, but not yet in other industrial sectors, and hence these non-American producers have no choice but to aim for the maximum possible degree of penetration of export markets appropriate to their product spectrum. By and large, this means attacking the U.S. market-which is already giving cause for concern to some U.S. IC producers. By way of comparison, Figure 8 shows Mackintosh estimates of the 1965, 1975 and 1985 total markets for
Integrated Circuits: The Coming Battle TotallC Consumption ($M)
Per Capita IC Consumption ($)
1965
1975
1985
1965
1975
1985
U.S.A.
c60
c1200
c3500
0.3
5.7
15.9
EEC
c4
c480
c2200
-
1.9
8.5
Japan
c7
c480
c1900
-
4.4
16.2
Figure 8. Comparison ofiC usage trends ICs in Europe, Japan and the United States, in both absolute and per capita terms. The preferential future rate of growth of the Japanese and European markets is worth noting. It will be understood, of course, that the high per capita usage in Japan is related to the high production of electronic goods containing ICs.
Industrial Structure It is obvious that major differences exist between the structure of non-American and American IC companies. In both Europe and Japan, the bulk of the current IC capability resides within vertically-integrated and highly-structured companies, whereas in the U.S.-with the exception of organizations such as IBM and Western Electric-most of the IC capability resides in companies in which the semiconductor activity is a major part of its total industrial commitment. Some blurring of these historically sharp differences has recently occurred, however, through the various vertical integration moves which several U.S. component and equipment companies have made.
33
offer a brief summary of certain overall conclusions which are relevant to the main theme of this paper. In general, it is my view that the vogue for vertical integration is an irrelevant diversion in the long-term development of the electronics industry. We live in an age of specialization, and it has for a long time been difficult to accept that IC companies can sell watches, for example, better than the established specialists, or that the minicomputer companies, as another example, will succeed in establishing and maintaining a cost-effective semiconductor capability over the long term. Of course, there are the notable exceptions of IBM, Texas Instruments and one or two others which seem to disprove the Mackintosh General Theory of Vertical Disintegration, i.e. the general hypothesis that vertical integration, by and large, is a snare and delusion. However, apart from some special circumstances mainly centred on the microprocessor, I believe that of all the vertical integration activities-upwards and downwards -now going on in many parts of the world, only a few will turn out to be successful in the long term.
Management and People Anyone who has worked extensively in the electronics industry both inside and outside the United States, will recognize that there often exists in other countries a real sense of inferiority about American management skills. It is not just the general aura of infallibility surrounding a Harvard MBA, for example, but the sheer bewilderment with which the typical non-American electronics executive compares his apparent performance with that of his American counterpart. This, however, is a misconception and to support this view Figure 9 shows a listing of the top ten American companies: (a) in the electron tube and transistor businesses in 1955;
Clearly, the whole question of vertical integration is of great importance, but it is, however, a topic of enormous complexity. All that can be included here is to
(b) in the semiconductor business in 1960 and 1965; (c) in the IC business in 1975.
The Leading U.S. Manufacturers 1955 Tubes
Transistors
1960 Semiconductors
1965 Semiconductors
1975 Integrated Circuits
1.
RCA
Hughes
Tl
Tl
Tl
2.
Sylvania
Transitron
Transitron
Fairchild
Fairchild
3.
GE
Philco
Philco
Motorola
NSC
4.
Raytheon
Sylvania
GE
Gl
Intel Motorola
5.
Westinghouse
Tl
RCA
GE
6.
Amperex
GE
Motorola
RCA
Rockwell
7.
National Video
RCA
Clevite
Sprague
Gl
8.
Ranland
Westinghouse
Fairchild
Philco/Ford
RCA
9.
EIMAC
Motorola
Hughes
Transitron
Signetics (Philips)
10.
Lansdale Tube
Clevite
Sylvania
Raytheon
AMI
Figure 9. The leading U.S. manufacturers of vacuum tubes, semiconductors and ICs for the years 1955, 1960, 1965 and 1975
34
Long Range Planning Vol. 12
June 1979
This table, by and large, reflects the abilities of U.S. managements to bridge the gap between radically different technologies (i.e. vacuum tubes to germanium; germanium to silicon) and, in fact, only one of the leading 10 U.S. tube manufacturers in 1955 has survived as a significant IC producer today (RCA). The inescapable conclusion is that, popular opm10n sometimes to the contrary, American companies in general do not have a good track record in the management of electronics technology. A few, however, have obviously exhibited very impressive skills, and it is these successful managements, of course, on which the U.S. domination (and reputation) is based. What it all adds up to is that America's overwhelming success in the IC business has been, not surprisingly, something of a statistical phenomenon. With so many companies starting up, in such favourable conditions, some at least were likely to succeed in a big way. And they certainly did. There are some other advantages which American managements have enjoyed. The distinguishing organizational feature of the successful U.S. semiconductor companies is that they are geared to react swiftly to new developments. They are also, in many cases, led by an impressive new breed of technical entrepreneurs who are skilled at this particular trade. By comparison, European and Japanese semiconductor companies have often been too ponderous in their decision-making and have sometimes been managed by individuals whose understanding of the complexities of the semiconductor industry was sometimes less than perfect. In the last analysis, the IC industry-like every otherdepends on the right people being motivated in the right way to do the right job. In this sense, America has so far had most of the advantages since the evidence is very strong that entrepreneurial drive and freedom are essential conditions for success in the IC industry, and these are qualities that seem to thrive preferentially in America's relatively laissez-faire economy. The United States has also had a major advantage in th.:: fortuitous combination of a high rate of personnel mobility with the existence of several large and highly capable research laboratories which have acted as national generators of technology and technologists. Thus, in America the diffusion of technology has occurred mainly through the diffusion of people, and the commercial exploitation of new techniques has rarely been inhibited for long because the possessor of know-how and the would-be exploiter could not make common cause. By comparison, both Europe and Japan exhibit a considerably lower degree of personnel mobility, with the result that companies must rely on-and pay for-a proportionately much larger in-house programme of research and development. Another, indirect consequence of America's high personnel mobility-particularly of technical and mana-
gerial personnel-has been the emergence of what can be called 'skili dusters'. So far as electronics is concerned, this can be applied to centres such as Boston's Route 128 and, of particular relevance today, to areas of semiconductor expertise such as 'Silicon Valley' in California. (This phenomenon of skili clusters, by the way, is by no means new, London's Savile Row tailors having probably started the fashion about 200 years ago.) It is abundantly clear that the existence of Silicon Valley confers important advantages on the IC companies which operate within it, particularly in regard to the high (but informal) level of localized communication and debate, and to the availability of the strong commonservices industry which has developed in that area.
Summary of the Present State of the Industry Summarizing the present state of the world-wide IC industry, including one or two strategic points in addition to those discussed above, it can be concluded that the American domination has so far been founded on the following 10 inherent advantages.
(1) Critically-important synergism with major user industries.
(2) (3) (4) (5)
Large and innovative domestic markets.
(6)
En~ugh
Substantial government support over many years. Business climate suitable for entrepreneurs. Availability of substantial amounts of venture capital. good management in enough good com-
pames.
(7) Existence of large, capable research laboratories. (8) High mobility of technical and managerial personnel. (9) Skiii clusters (e.g. Silicon Valley). (10) Good fortune-including cheap energy and enormous international economic strength. The future importance of these factors wili obviously have an important bearing on any changes which may occur in the competitiveness of the American IC industry.
Future Prospects It is not the aim here to provide a detailed forecast ofiC
technology per se, but it is clear, nevertheless, that future technological developments will profoundly influence any strategic forecast of the IC industry. To this end, the evidence is now strong that a new high-resolution lithographic technology will emerge within the next few years, probably based on electron beam techniques, 4 and which will be very expensive to establish. This VLSI technology will find application first in circuits requiring vast numbers of components (principally, of course,
Integrated Circuits: The Coming Battle memory, microcomputers and imaging), but will later be used also for numerically more mundane applications since by then the production economics will strongly favour VLSI against more classical technologies. One major change which will therefore occur within a few years is a substantial increase in the investment required to complete at the 'leading edge'. Thus the industry will begin to move into an era in which the sheer size of the initial financial commitment will provide a stabilizing feedback effect. There will therefore be reduced opportunities for spin-offs to leapfrog into prominence by means of some astute technical, marketting or economic stratagem. With this general scenario, then, the probable future changes in the 10 key strategic factors already identified can be considered (Figure 10).
•••• •• ••• ••• • •• U.S.
1. Industrial Synergism
2. Market Availability 3. Government Support
4. Business Climate 5. Venture Capital
6. Management Qualities 7. Research Capability 8. Personnel Mobility
Japan
Europe
~
9. Skill Clusters 10. Economic Conditions
11111111111
Figure 10. Future prospects in LSI/VLSI Key Relative future prospects from bad to good: Bad
k;' <:I
MW!fff{f[[NJ
I. .I
Good.
Taking the first two together (Industrial Synergism and Market Availability), much of the future growth in the world's markets for electronic functions is likely to be in domestic and personal product sectors, not industrial. Despite the maker-push effect which the American IC industry has so far exerted in products such as calculators, electronic watches and video games, the fact is that it is in just these areas of consumer, automotive and personal electronics that Europe and Japan are strong. Their budding IC industries therefore have the prospect of the kind of critically-important synergistic relationship with
35
major user industries which the American IC industry has so far enjoyed with its data processing customers. In addition, there is certain to be substantial growth, in both Europe and Japan, in 'protected' applications such as telecommunications and the 'national' computer industries. Overall, therefore, there will be selective growth of the IC markets in Japan and Europe and much of this growth will be in user companies which, by corporate inclination or national preference, will tend to select 'local' suppliers, all other things being equal. For these first two items, therefore, in the future Japan and Europe will enjoy much greater parity with the U.S. than in the past. So far as Government Support is concerned it seems that until quite recently, the importance of this factor had not been understood properly by the governments of many advanced nations, although things are now changing rapidly. In Europe, for example, the British, French, German and Italian governments are all beginning to talk about-and in some cases activate-plans to provide support typically in the range $50-100m spread over 4-5 years . And in Japan, there is, of course, the famous VLSI programme, 5 about which it is very difficult to obtain hard facts. Our own best estimate in Mackintosh Consultants is that the purely government funding for this VLSI project is running at an annual rate of about $6Sm (in 1978 dollars) and there is little doubt that it will continue at about this level well into the 1980s. In any event, several non-American governments are now beginning to support their indigenous IC industries with meaningful sums of money and, therefore, the longstanding American advantage in this respect will diminish, although substantial support of the U.S . industry can be expected to continue. Taking the fourth and fifth factors together (i.e. The Business Climate and Venture Capital), so far as the business climate is concerned, despite the probable stabilizing influence of the advent of VLSI technology, the entrepreneurial touch will remain an important ingredient of success in the IC industry in which long-term success will only go to him who can afford, has control of, and knows how to fully exploit the most complex industrial technology man has yet devised. The multi-sector conglomerate, therefore, with too many of its eggs not in the semiconductor basket, will tend to lack the total commitment to success of the entrepreneurially-led specialist IC companies. Thus America will continue to have the edge over both Japan and Europe, where the large corporations are unlikely to allow their semiconductor managers the same freedom of decision which their American counterparts will enjoy. Nevertheless, the opportunities for new American entrepreneurs will diminish as the size of the risks exceeds acceptable limits for conventional American venture
36
Long Range Planning VoL 12
June 1979
capitalists. (Indeed, the diminution of classical U.S. venture capital has already been observable for some time, with a sharp reduction in the number of new semiconductor start-up companies.) In both Japan and Europe, the financial community has historically been markedly unadventurous about providing venture capital, and this situation is unlikely to change in the foreseeable future. On the other hand, the fact that the IC capability in these countries is mainly controlled by large companies, could turn out to be an advantage from this point ofview since most of these will be capable of funding VLSI technology-especially with the aforementioned government support. The resource question, therefore, seems likely to become fairly evenly balanced in the future between the U.S., Europe and Japan. The question of the Quality of Management can be dealt with very briefly. In my assessment there is no significant difference between the inherent capabilities of executives in these different countries. Moreover, evidence has been offered already which strongly suggests that Americans do not have a monopoly of competence when it comes to the management of technology. The conclusion can be drawn that the managements in all of the relevant countries are, within experimental accuracy, about equally skilful (or incompetent) at creating commercial success from this esoteric semiconductor technology. The next three factors-Research Capability, Personnel Mobility and Skill Clusters-can also conveniently be discussed together, since they are inter-related. Taking the U.S. first, it is unlikely that any significant diminution in its strength will occur in any of these areas. While there may well be some reduction in the amount of basic research carried out, this will be more than offset by increases in applied research in areas such as VLSI techniques, product testing and software problems. Personnel mobility will certainly remain high, and it is very unlikely that any of the important American skill clusters-whether they are called Silicon Valley, TI, IBM, Bell Labs. or whatever-will disappear. (Parenthetically, it should be observed that there are substantial differences between a geographical skill cluster such as Silicon Valley, and a corporate skill cluster such as TI, but again this is a lengthy and complex matter which it is not appropriate to discuss here.) By and large, therefore, America can be expected to continue to enjoy great strength in all three of these areas. Turning to Japan, a systematic build-up of the national research capability has been under way for many years and this process will undoubtedly continue. For reasons which are well known, personnel mobility is low in Japan, due primarily to the high level of loyalty which tends to exist between company and employee. However, there will be some increase in mobility as joint ventures, company mergers and government policies slowly blur individual corporate identities, and increasingly permeate the Japanese way oflife. On the question ofskill clusters, Japan also enjoys a strong
position since the Japanese electronics industry is already mainly confined to the two metropolitan areas of Tokyo and Osaka. This clustering will be reinforced by an in'creasing number of co-operative industrial R & D activities, such as can already be seen in the VLSI programme. Overall, therefore, Japan is either already strong, or will soon become so, in each of these three strategic areas. In Europe, however, things look distinctly worse. In the first place, although it is a fact so patently obvious as to seem hardly worth mentioning, it is nevertheless sometimes forgotten that whereas the U.S. and Japan are single nations, each with a single language, national sense of commitment, set of laws, customs and cultural attitudes, etc., Europe represents a set of highly individualistic nations, each with its own language, national objectives and way of doing things. The net result-hardly surprisingly-is that there is not yet such a thing as a truly 'Common Market'. Thus even neighbouring markets (e.g. France in relation to West Germany) can often represent as great a problem as a far-distant market such as the United States. Echoing this general situation, the European semiconductor industry basically consists (with the exception of Philips) of a number of predominantly national producers, each organized principally to serve the needs of its own national markets. One of the consequent liabilities is that Europe has nothing remotely to compare with America's Silicon Valley, nor is it likely that any meaningful geographical skill clusters will ever develop in the European IC industry. For the same general reasons, personnel mobility in Europe is also low, inhibited both by employment traditions and national boundaries, and is unlikely to increase significantly. The European research capability in solid-state electronics has always been high, although often commercially ineffective through an inability to bridge the gap between science and sales. However, this research capability will improve due to increasing government support, and increasing co-operation both between European laboratories and outwith Europe. Overall, therefore, in these three categories Europe comes behind Japan which, in turn, will be deficient in relation to America only in respect of personnel mobility. The final category-Economic Conditions-really warrants a complete paper in its own right if its relative importance is to be accurately assessed. Some aspects have already been covered earlier in this report but with regard to the relationship of the U.S.A. with other world nations the key points can very briefly be summarized as follows. In the years since the end of the Second World War, America has dominated the economic state of health of
Integrated Circuits: The Coming Battle the OECD nations. This strength has primarily been founded on cheap energy, abundant natural resources and a large enough population for the producers of manufactured goods to enjoy the benefits of considerable economies of scale. Meanwhile, however, other nations-in particular, Germany and Japan-have been recovering from the ravages of war and one of the pillars of American economic strength has eroded as the dramatic increase of oil prices has coincided with the gradual depletion of America's oil resources. For the future, therefore, there is likely to be much more balance of economic strength between the U.S.,EuropeandJapan which is the situation reflected in Figure 10.
Conclusions As a summary of the foregoing discussion Figure 11 shows the changes which will probably take place in the principal strategic factors on which America's ascendancy in the integrated circuits industry has so far been based, and on which the future domination ofVLSI technology may depend. The Past
U.S.Japan EuropeU.S.JapanEurope
IJJ
Jl ~ ~ H
~~
2. Market Availability
~~
IJI Ill)
3. Government Support 4. Business Climate 5. Venture Capital 6. Management Qualities 7. Research Capability 8. Personnel Mobility 9. Skill Clusters 10. Economic Conditions
~
~~
IJ t!ill ~ &J ~~ [81 ~~ ~ rEill IIJ tfJill [i!ll IJ tiEJ (2B ES) ffiill ~ ~~ ~~ B ~~ ~~ -~~ fl l\S)j [3j llj ~~ ~~~ B ~ ~ Jl [Sill ~ ~~ tuj 1m] liJ ~~ ~ • flliJj f£ill Ill • Ill
Figure 11. Summary: the shifting balance of advantage Key Relative advantage from bad to good: Bad ~
[;j II II
Good.
In brief, even allowing for the many remaining strengths of the American IC industry-such as its immensely strong technological base, its position on the learning curve, management-in-depth, etc;-and the fact that the European and Japanese managements will still suffer from important liabilities such as the relative absence of entrepreneurial freedom, there can be no doubt that the balance of advantage is now beginning to swing away from the United States. For this reason, in the VLSI era, American producers will face problems of daunting magnitude in maintaining global market share and innovatory leads against escalating transatlantic and transpacific competition, and the most probable prognosis is that the U.S. domination of this criticallyimportant industrial asector will eventually disappear, to be replaced first by a condition of approximate parity between America and Japan, who will possibly be joined somewhat later by Europe. It hardly needs to be said that the effects of these changes -on both component and equipment companies-will be immense and will need to be understood in detail if the changing competitive circumstances are to be digested without acute corporate discomfort.
The Future
1. Industrial Syngerism
[£]
37
Acknowledgments-Although the views expressed herein are strictly personal, some of the analytical work on which this paper is based was carried out during a major semiconductor study carried out by the author's consultancy company for a number of European governments, principal among them the West German Bundesministerium fur Forschung und Technologie, to whom thanks are due for permission to draw on some of the data collected during that project.
References (1) C. L. Hogan, An Address to Semicon/Europe 75, Zurich; November 4 (1975). (2) G. E. Moore, Progress in Digital Electronics, Proc. lnt'l Electron Devices Meeting; December (1977). (3) R. Turn, Computers in the 1980's, Columbia University Press, New York (1974). (4) I. M. Mackintosh, Applications of the Scanning Electron Microscope to Solid-State Devices, Proc. IEEE 53, p. 370, April (1965). (5) Y. Tarui, LSI and VLSI Research in Japan, Proc. lnt'l Devices Meeting; December (1977).
Long Range Planning Vol. 12
June 1979
Integrated Circuits: The Coming Battle I. M. Mackintosh*
Computerization has been hailed on the one hand as a liberating force for mankind, to, on the other, a destroyer of the human role in many manufacturing and commercial activities. There is little doubt that the possible impacts of future trends in the computer industry upon commercial and industrial activity will be the subject of continuing debate in the next 25 years. However, many commentators have expressed the view that the 'rate' of change could well be unacceptable and create disturbing imbalances in an already disturbingly imbalanced society. This paper examines the impact of the rapidly changing computer technology upon the computer manufacturing industry itself. These changes are of immense importance since they will affect both the distribution of technology as well as economic power. The author, in a detailed and persuasive analysis, plots the trends of ~future shock'. These trends have impacts not only for the computer industry but for the activities of large scale manufacturing and commercial enterprises throughout the world.
Ever since the inventions of the integrated circuit and the silicon planar process, American semiconductor producers have dominated the world markets for ICs,t especially at the leading-edge of each new generation of technology. This is certainly the case today for products ofLSit complexity. But there are increasing signs that things might be changing. For example, competitive memory products are now being produced in substantial quantities in Japan, and there is a growing interest by European companies in acquiring a stake in U.S. semiconductor companies. A key question, therefore, is whether the global domination by American IC companies will be seriously threatened, particularly as the VLSI era approaches, and it is to an examination of the major facets of this question that this paper is directed. The popular but simplistic view-widely held outside the United States-is that the American domination of this industry has been based primarily on very substantial and *The author is Chairman of Mackintosh Consultants Ltd., 33 Bruton Street, London W1. tFor IC read Integrated Circuit, and for LSI read Large Scale Integration.
continuous financial support from the U.S. government. Whereas this funding has obviously been important, the real foundations of this American success are far more complex and need to be understood in detail before the outcome of the impending intercontinental LSI/VLSI battle can be correctly forecast. Indeed, any prognosis of future structural developments in this industry must begin with an understanding of the principal historical forces which have moulded the industry into its present form. The reason, quite obviously, is that whereas important new strategic influences are now emerging, these historical forces will continue to have significant effect for the foreseeable future. This paper therefore comprises two main parts. The first looks briefly at some aspects of the historical development of the semiconductor industry in general, and ICs in particular, in the U.S., Europe and Japan, giving due attention to the various factors which produced the different pace and direction of development in each geographic region. It is then possible to identify a number of key factors which have led to the current American domination of world IC markets. From this foundation, the second part goes on to examine the relevance of these factors, plus the new strategic forces now gathering momentum, to the future chances of success ofiC producers in each geographical region, and to offer a prognosis of the future capabilities of each of these regions to compete in the VLSI era.
The Origins of the Semiconductor Industry As is well-known, the 1950s were characterized by the germanium transistor first going into high-volume production. Its most immediate application was in cheap, portable radios, and the nation which seized this opportunity most effectively was Japan, possessing at that time the considerable advantage of low labour costs. However, Europe (in particular, Philips) was not left far behind.
Integrated Circuits: The Corning Battle In the U.S., however, the gleam in the electronic industry's collective eye was not caused by radios, largely because of the high American labour costs (the idea of moving labour-intensive assembly operations off-shore had not yet been tried). Instead, the greatest need (what we might call the 'user-pull', as distinct from the 'makerpush' effect) for the transistor was mainly in the defence and aerospace sectors-1957 being the year of the Sputnik-and in the infant computer industry. The demands of these military and industrial sectors for devices of higher performance and reliability thus led in time to the emergence of the silicon transistor and later to integrated circuits-both developed, of course, by American companies. The final effects of these original, basic reactions to the advent of the transistor were as follows: {1) The Europeans and Japanese became strong in germanium technology and in the main types of equipments (i.e. consumer electronic products) which were based at that time on germanium transistors. (2) The Americans became pre-eminent in silicon technology and in the main types of equipment {military, computers, communications, etc.) based on this technology. (3) These distinctive postures, originally taken up 1520 years ago, still pertain today, with the Europeans and Japanese leading by a significant margin in most aspects of consumer electronics and the Americans continuing to dominate every other sector of the electronics industry. What can be learned, therefore, from this briefhistorical review is that industrial synergism-the mutual interdependence of different industrial sectors and, in particular, of the equipment and component sectors of the electronics industry-has been a signiftcant factor in shaping the development of the electronics industry in different geographic regions. This is explored later.
The Principal Strategic Factors Affecting the Development of the International Integrated Circuit Industry With the advantage of this historical perspective, the principal strategic factors which have affected-and in most cases will continue to affect-the development of the global IC industry can be identified. There are, obviously, many such strategic factors, and considerable analysis of data, case histories, etc. is necessary if their relative importance is to be correctly assessed, and this work allows certain principal conclusions to be drawn, although it is obviously impossible to present the mass of detailed evidence on which these conclusions are based. The ~ain strategic factors which need to be discussed are:
29
Support * Government Industrial Synergism * Technological * Market FactorsInnovation * Industrial Structure * Management and People. *
The Role of Governments It is beyond question that, almost from its beginning, the American semiconductor industry as a whole has received substantial and broadly-based support from various U.S. government agencies. It has been estimated1 that this totalled about $900m for R & D alone between 1958 and 1974, representing a subsidy of the cost of American semiconductor innovation to the tune of about $55m per annum (in the terms of, say, 1965 average dollar values). Now, it is clear that financial support from public funds on such a large scale has grossly distorted normal competitive conditions and commercial criteria in this industry, and it is difficult to imagine, therefore, how any non-American nation could succeed without providing support to its own indigenous industry on a comparable scale.
The Benefits of Industrial Synergism The general, theoretical benefits of industrial synergism are familiar. However, in a paper of this type it could be useful to emphasize the particular importance of synergism in the development of the global IC industry. Figure 1 represents the well-known Learning Curves for silicon transistors and ICs, showing the systematic relationship between price and cumulative production experience. From sources such as these it is possible to generate the price trend curves, shown in Figure 2, for three silicon devices of progressively increasing complexity-the diffused silicon transistor, the 'average' IC as defined by the Learning Curve, and the more recent IK-bit dynamic RAM. However, following the example of Moore, 2 if these data are normalized in time so that the prices are related to the number of years since product introduction, the effect is to shift the points for the higher-complexity devices to the left, giving the result illustrated in Figure 3. This shows fairly conclusively that there is a general (and very steep) price-reduction curve which can reasonably be assumed to apply to all kinds of high-volume silicon devices, and this is considered again later. First, however, it is illuminating to examine the changes in the performance: cost ratio of computers over the past 15 years or so. Figure 4 is a reproduction of the original curve developed by Turn (Rand Corporation Report) 3 showing the cost in dollars per million instructions per
Long Range Planning Vol. 12
30
June 1979
100
:;;;:
10
(/)
::) Q)
.g 0..
..... ·c: ;::j
---~--~
-~----~~--~~--~--0.1~----~----~----~~--~----~----_.~8
10
10 6
{Units)! Production
Figure 1. Learning curve
100
100
e1K- bit Dynamic RAM xAverage IC 0 0iffused Silicon Transistor •
• 1K - bit Dynamic RAM x Average IC • Diffused Silicon Transistor
X
X
\•
~ Cll
10
10
=s0
Q)
Cii ...
c.> ·.::
co
"5
•• •
0
Q)
.\:1
a:
0.. Ol
~ 1P (/) Q)
Ol
e ~
<(
1.0
Years
1.0
Years After Product Introduction
Figure 2. Typical price trends of silicon devices
Figure 3. Normalized silicon device price trends
second (MIPs) for high-performa nce general-purpose computers, covering the years 1960-1990. It is necessary to note, however, that there is, so far, only one working version of the Illiac IV computer, and the 1000 MIP and 10,000 MIP Processors have yet to be developed.
Figure 5, therefore, shows Turn's curve redrawn to give greatest weight to actual, well-documen ted cost data, and this is compared with the device price curve shown earlier in Figure 3. The correlation is striking and seems to prove what every IC engineer has always instinctively
Integrated Circuits: The Coming Battle
31
100Mr-------~---------r--------~--------~------~~------~
10M •"Stretch
•'BM709411 • CDC6600
1M
en a..
• CDC 7600
•IBM 360/195
~
...
Q)
a..
• CDC Star
ILLIACIV•~
lOOK
1000 MIP • Processor
~
10K
•
1K~--------~--------~--------L---------L---------L-------~
1970
1900
1980
1990
Year
Figure 4. Cost trends in high-performance general application computers
believed: that the computer industry's spectacular growth has been due mainly to its ability to produce equipment which could compute at ever-increasing speeds and reliability levels, and ever-decreasing cost and size, and that essentially all of these attributes have stemmed from advances in silicon technology.
100M
100
• x • o
1K - bit Dynamic RAM Average IC Diffused Silicon Transistor High Performance Computer Costs
X\ '§ .!!! 0
10
0
0
~ Cl
c:
Q) (/) Q)
Cl Q)
>
\
a.. ~
'~\
a..
X~,.~
Q)
~
\\
10M
1.0
Cii ~
Computiog Co"'
f(.,..' ' ~
Slllcoo Devke Price Trends
.!!!
0 0
1M
0
~ •
'-
• lOOK
Years (Normalised)
Figure 5. Comparison of computing costs and silicon device costs
But there is another side to the coin. Figure 6 shows the growth of the total U.S. consumption of monolithic silicon ICs over the period 1964-1976, with digital ICs indicated separately. Since most of these digital ICs, needless to say, were necessarily used in computing equipment of all kinds (including military), the conclusion is inescapable. Just as the American computer industry's growth has been critically dependent on the availability of ever-increasing numbers of ever-improving ICs, so has the spectacular growth of the American IC industry depended to a very high degree on having a large, innovative and 'local' computer market avid to make use of its rapidly-developing semiconductor capabilities. This must be regarded as a particularly convincing example of the benefits of industrial synergism and leaves no doubt that the simultaneous American domination of the integrated circuit, computer and professional electronics sectors are all part of the same basic phenomenon, and that this is the main-not the only, but the mainreason for today's domination by the United States of the world-wide IC business.
32
Long Range Planning Vol. 12
June 1979
1400
X
X
1200
In Europe and Japan, however, the process ofin11ovation in the field of advanced semiconductor components has been hindered by the relative absence until recently of innovative 'user-pull' markets and, especially in Europe, by too much emphasis on basic research and too little on development and marketing. A recent development of enormous significance, and an excellent example of the innovative strengths of the American IC industry, is the microprocessor. Its strategic importance stems mainly from its great commonality of application, as illustrated in Figure 7, thus allowing
X X
High
SSI Microprocessor
X
r--
Digital
1
c
0 ·.;:;
co .!::!
X
Q.
I I
c.
<(
0
X
X/
.~ a;
·-·
/ ·-·-1. ~·~·--r
1964
1966
1968
c 0
E E 0
1970
1972
1974
1976
u
Year
Figure 6. U.S. consumption of monolithic integrated circuits
And the corollary, of course, is that the absence until recently of such synergetic user industries outside America has been the principal reason for the early lack of success of the European and Japanese IC producers.
The Impact of Technological Innovation It is clear that innovation has been, and will for the foreseeable future remain, a major strategic factor in the growth. of the international semiconductor industry. Analysis of case histories shows, however, that the key elements in the total process of innovation are development and marketing-not basic research. In fact, there has been no correlation whatsoever in the U.S. between the commercial success of an IC company and the quality of its basic research programme, an ability to recruit key personnel having historically been much more important. Nevertheless, disciplined in-house development of processes and products has been, and will remain, a key factor in any semiconductor company's success, but for product development work to be relevant it is essential for the company to compete actively in the world's most innovative markets-wherever they may be-for those products.
Low L - - - - - - - - - - - - - - - ; - ; : - : ; - High Low IC Complexity
Figure 7. Complexity versus commonality LSI products the opportunity to break out of the vicious circle of greater complexity: fewer applications: higher cost. It is clear that the microprocessor offers as big a step forward for digital systems as did the original integrated circuit, and it is symptomatic that in this product area Europe as yet has an almost insignificant capability, whereas Japan is already beginning to produce microprocessors, albeit on a modest scale.
Market Factors In light of earlier comments on industrial synergism, it is clear that access to large and innovative (user-pull) markets is a key factor influencing success in the IC business. Today, both Japan and Europe have finally developed large and innovative markets in the consumer electronics sector, but not yet in other industrial sectors, and hence these non-American producers have no choice but to aim for the maximum possible degree of penetration of export markets appropriate to their product spectrum. By and large, this means attacking the U.S. market-which is already giving cause for concern to some U.S. IC producers. By way of comparison, Figure 8 shows Mackintosh estimates of the 1965, 1975 and 1985 total markets for
Integrated Circuits: The Coming Battle TotallC Consumption ($M)
Per Capita IC Consumption ($)
1965
1975
1985
1965
1975
1985
U.S.A.
c60
c1200
c3500
0.3
5.7
15.9
EEC
c4
c480
c2200
-
1.9
8.5
Japan
c7
c480
c1900
-
4.4
16.2
Figure 8. Comparison ofiC usage trends ICs in Europe, Japan and the United States, in both absolute and per capita terms. The preferential future rate of growth of the Japanese and European markets is worth noting. It will be understood, of course, that the high per capita usage in Japan is related to the high production of electronic goods containing ICs.
Industrial Structure It is obvious that major differences exist between the structure of non-American and American IC companies. In both Europe and Japan, the bulk of the current IC capability resides within vertically-integrated and highly-structured companies, whereas in the U.S.-with the exception of organizations such as IBM and Western Electric-most of the IC capability resides in companies in which the semiconductor activity is a major part of its total industrial commitment. Some blurring of these historically sharp differences has recently occurred, however, through the various vertical integration moves which several U.S. component and equipment companies have made.
33
offer a brief summary of certain overall conclusions which are relevant to the main theme of this paper. In general, it is my view that the vogue for vertical integration is an irrelevant diversion in the long-term development of the electronics industry. We live in an age of specialization, and it has for a long time been difficult to accept that IC companies can sell watches, for example, better than the established specialists, or that the minicomputer companies, as another example, will succeed in establishing and maintaining a cost-effective semiconductor capability over the long term. Of course, there are the notable exceptions of IBM, Texas Instruments and one or two others which seem to disprove the Mackintosh General Theory of Vertical Disintegration, i.e. the general hypothesis that vertical integration, by and large, is a snare and delusion. However, apart from some special circumstances mainly centred on the microprocessor, I believe that of all the vertical integration activities-upwards and downwards -now going on in many parts of the world, only a few will turn out to be successful in the long term.
Management and People Anyone who has worked extensively in the electronics industry both inside and outside the United States, will recognize that there often exists in other countries a real sense of inferiority about American management skills. It is not just the general aura of infallibility surrounding a Harvard MBA, for example, but the sheer bewilderment with which the typical non-American electronics executive compares his apparent performance with that of his American counterpart. This, however, is a misconception and to support this view Figure 9 shows a listing of the top ten American companies: (a) in the electron tube and transistor businesses in 1955;
Clearly, the whole question of vertical integration is of great importance, but it is, however, a topic of enormous complexity. All that can be included here is to
(b) in the semiconductor business in 1960 and 1965; (c) in the IC business in 1975.
The Leading U.S. Manufacturers 1955 Tubes
Transistors
1960 Semiconductors
1965 Semiconductors
1975 Integrated Circuits
1.
RCA
Hughes
Tl
Tl
Tl
2.
Sylvania
Transitron
Transitron
Fairchild
Fairchild
3.
GE
Philco
Philco
Motorola
NSC
4.
Raytheon
Sylvania
GE
Gl
Intel Motorola
5.
Westinghouse
Tl
RCA
GE
6.
Amperex
GE
Motorola
RCA
Rockwell
7.
National Video
RCA
Clevite
Sprague
Gl
8.
Ranland
Westinghouse
Fairchild
Philco/Ford
RCA
9.
EIMAC
Motorola
Hughes
Transitron
Signetics (Philips)
10.
Lansdale Tube
Clevite
Sylvania
Raytheon
AMI
Figure 9. The leading U.S. manufacturers of vacuum tubes, semiconductors and ICs for the years 1955, 1960, 1965 and 1975
34
Long Range Planning Vol. 12
June 1979
This table, by and large, reflects the abilities of U.S. managements to bridge the gap between radically different technologies (i.e. vacuum tubes to germanium; germanium to silicon) and, in fact, only one of the leading 10 U.S. tube manufacturers in 1955 has survived as a significant IC producer today (RCA). The inescapable conclusion is that, popular opm10n sometimes to the contrary, American companies in general do not have a good track record in the management of electronics technology. A few, however, have obviously exhibited very impressive skills, and it is these successful managements, of course, on which the U.S. domination (and reputation) is based. What it all adds up to is that America's overwhelming success in the IC business has been, not surprisingly, something of a statistical phenomenon. With so many companies starting up, in such favourable conditions, some at least were likely to succeed in a big way. And they certainly did. There are some other advantages which American managements have enjoyed. The distinguishing organizational feature of the successful U.S. semiconductor companies is that they are geared to react swiftly to new developments. They are also, in many cases, led by an impressive new breed of technical entrepreneurs who are skilled at this particular trade. By comparison, European and Japanese semiconductor companies have often been too ponderous in their decision-making and have sometimes been managed by individuals whose understanding of the complexities of the semiconductor industry was sometimes less than perfect. In the last analysis, the IC industry-like every otherdepends on the right people being motivated in the right way to do the right job. In this sense, America has so far had most of the advantages since the evidence is very strong that entrepreneurial drive and freedom are essential conditions for success in the IC industry, and these are qualities that seem to thrive preferentially in America's relatively laissez-faire economy. The United States has also had a major advantage in th.:: fortuitous combination of a high rate of personnel mobility with the existence of several large and highly capable research laboratories which have acted as national generators of technology and technologists. Thus, in America the diffusion of technology has occurred mainly through the diffusion of people, and the commercial exploitation of new techniques has rarely been inhibited for long because the possessor of know-how and the would-be exploiter could not make common cause. By comparison, both Europe and Japan exhibit a considerably lower degree of personnel mobility, with the result that companies must rely on-and pay for-a proportionately much larger in-house programme of research and development. Another, indirect consequence of America's high personnel mobility-particularly of technical and mana-
gerial personnel-has been the emergence of what can be called 'skili dusters'. So far as electronics is concerned, this can be applied to centres such as Boston's Route 128 and, of particular relevance today, to areas of semiconductor expertise such as 'Silicon Valley' in California. (This phenomenon of skili clusters, by the way, is by no means new, London's Savile Row tailors having probably started the fashion about 200 years ago.) It is abundantly clear that the existence of Silicon Valley confers important advantages on the IC companies which operate within it, particularly in regard to the high (but informal) level of localized communication and debate, and to the availability of the strong commonservices industry which has developed in that area.
Summary of the Present State of the Industry Summarizing the present state of the world-wide IC industry, including one or two strategic points in addition to those discussed above, it can be concluded that the American domination has so far been founded on the following 10 inherent advantages.
(1) Critically-important synergism with major user industries.
(2) (3) (4) (5)
Large and innovative domestic markets.
(6)
En~ugh
Substantial government support over many years. Business climate suitable for entrepreneurs. Availability of substantial amounts of venture capital. good management in enough good com-
pames.
(7) Existence of large, capable research laboratories. (8) High mobility of technical and managerial personnel. (9) Skiii clusters (e.g. Silicon Valley). (10) Good fortune-including cheap energy and enormous international economic strength. The future importance of these factors wili obviously have an important bearing on any changes which may occur in the competitiveness of the American IC industry.
Future Prospects It is not the aim here to provide a detailed forecast ofiC
technology per se, but it is clear, nevertheless, that future technological developments will profoundly influence any strategic forecast of the IC industry. To this end, the evidence is now strong that a new high-resolution lithographic technology will emerge within the next few years, probably based on electron beam techniques, 4 and which will be very expensive to establish. This VLSI technology will find application first in circuits requiring vast numbers of components (principally, of course,
Integrated Circuits: The Coming Battle memory, microcomputers and imaging), but will later be used also for numerically more mundane applications since by then the production economics will strongly favour VLSI against more classical technologies. One major change which will therefore occur within a few years is a substantial increase in the investment required to complete at the 'leading edge'. Thus the industry will begin to move into an era in which the sheer size of the initial financial commitment will provide a stabilizing feedback effect. There will therefore be reduced opportunities for spin-offs to leapfrog into prominence by means of some astute technical, marketting or economic stratagem. With this general scenario, then, the probable future changes in the 10 key strategic factors already identified can be considered (Figure 10).
•••• •• ••• ••• • •• U.S.
1. Industrial Synergism
2. Market Availability 3. Government Support
4. Business Climate 5. Venture Capital
6. Management Qualities 7. Research Capability 8. Personnel Mobility
Japan
Europe
~
9. Skill Clusters 10. Economic Conditions
11111111111
Figure 10. Future prospects in LSI/VLSI Key Relative future prospects from bad to good: Bad
k;' <:I
MW!fff{f[[NJ
I. .I
Good.
Taking the first two together (Industrial Synergism and Market Availability), much of the future growth in the world's markets for electronic functions is likely to be in domestic and personal product sectors, not industrial. Despite the maker-push effect which the American IC industry has so far exerted in products such as calculators, electronic watches and video games, the fact is that it is in just these areas of consumer, automotive and personal electronics that Europe and Japan are strong. Their budding IC industries therefore have the prospect of the kind of critically-important synergistic relationship with
35
major user industries which the American IC industry has so far enjoyed with its data processing customers. In addition, there is certain to be substantial growth, in both Europe and Japan, in 'protected' applications such as telecommunications and the 'national' computer industries. Overall, therefore, there will be selective growth of the IC markets in Japan and Europe and much of this growth will be in user companies which, by corporate inclination or national preference, will tend to select 'local' suppliers, all other things being equal. For these first two items, therefore, in the future Japan and Europe will enjoy much greater parity with the U.S. than in the past. So far as Government Support is concerned it seems that until quite recently, the importance of this factor had not been understood properly by the governments of many advanced nations, although things are now changing rapidly. In Europe, for example, the British, French, German and Italian governments are all beginning to talk about-and in some cases activate-plans to provide support typically in the range $50-100m spread over 4-5 years . And in Japan, there is, of course, the famous VLSI programme, 5 about which it is very difficult to obtain hard facts. Our own best estimate in Mackintosh Consultants is that the purely government funding for this VLSI project is running at an annual rate of about $6Sm (in 1978 dollars) and there is little doubt that it will continue at about this level well into the 1980s. In any event, several non-American governments are now beginning to support their indigenous IC industries with meaningful sums of money and, therefore, the longstanding American advantage in this respect will diminish, although substantial support of the U.S . industry can be expected to continue. Taking the fourth and fifth factors together (i.e. The Business Climate and Venture Capital), so far as the business climate is concerned, despite the probable stabilizing influence of the advent of VLSI technology, the entrepreneurial touch will remain an important ingredient of success in the IC industry in which long-term success will only go to him who can afford, has control of, and knows how to fully exploit the most complex industrial technology man has yet devised. The multi-sector conglomerate, therefore, with too many of its eggs not in the semiconductor basket, will tend to lack the total commitment to success of the entrepreneurially-led specialist IC companies. Thus America will continue to have the edge over both Japan and Europe, where the large corporations are unlikely to allow their semiconductor managers the same freedom of decision which their American counterparts will enjoy. Nevertheless, the opportunities for new American entrepreneurs will diminish as the size of the risks exceeds acceptable limits for conventional American venture
36
Long Range Planning VoL 12
June 1979
capitalists. (Indeed, the diminution of classical U.S. venture capital has already been observable for some time, with a sharp reduction in the number of new semiconductor start-up companies.) In both Japan and Europe, the financial community has historically been markedly unadventurous about providing venture capital, and this situation is unlikely to change in the foreseeable future. On the other hand, the fact that the IC capability in these countries is mainly controlled by large companies, could turn out to be an advantage from this point ofview since most of these will be capable of funding VLSI technology-especially with the aforementioned government support. The resource question, therefore, seems likely to become fairly evenly balanced in the future between the U.S., Europe and Japan. The question of the Quality of Management can be dealt with very briefly. In my assessment there is no significant difference between the inherent capabilities of executives in these different countries. Moreover, evidence has been offered already which strongly suggests that Americans do not have a monopoly of competence when it comes to the management of technology. The conclusion can be drawn that the managements in all of the relevant countries are, within experimental accuracy, about equally skilful (or incompetent) at creating commercial success from this esoteric semiconductor technology. The next three factors-Research Capability, Personnel Mobility and Skill Clusters-can also conveniently be discussed together, since they are inter-related. Taking the U.S. first, it is unlikely that any significant diminution in its strength will occur in any of these areas. While there may well be some reduction in the amount of basic research carried out, this will be more than offset by increases in applied research in areas such as VLSI techniques, product testing and software problems. Personnel mobility will certainly remain high, and it is very unlikely that any of the important American skill clusters-whether they are called Silicon Valley, TI, IBM, Bell Labs. or whatever-will disappear. (Parenthetically, it should be observed that there are substantial differences between a geographical skill cluster such as Silicon Valley, and a corporate skill cluster such as TI, but again this is a lengthy and complex matter which it is not appropriate to discuss here.) By and large, therefore, America can be expected to continue to enjoy great strength in all three of these areas. Turning to Japan, a systematic build-up of the national research capability has been under way for many years and this process will undoubtedly continue. For reasons which are well known, personnel mobility is low in Japan, due primarily to the high level of loyalty which tends to exist between company and employee. However, there will be some increase in mobility as joint ventures, company mergers and government policies slowly blur individual corporate identities, and increasingly permeate the Japanese way oflife. On the question ofskill clusters, Japan also enjoys a strong
position since the Japanese electronics industry is already mainly confined to the two metropolitan areas of Tokyo and Osaka. This clustering will be reinforced by an in'creasing number of co-operative industrial R & D activities, such as can already be seen in the VLSI programme. Overall, therefore, Japan is either already strong, or will soon become so, in each of these three strategic areas. In Europe, however, things look distinctly worse. In the first place, although it is a fact so patently obvious as to seem hardly worth mentioning, it is nevertheless sometimes forgotten that whereas the U.S. and Japan are single nations, each with a single language, national sense of commitment, set of laws, customs and cultural attitudes, etc., Europe represents a set of highly individualistic nations, each with its own language, national objectives and way of doing things. The net result-hardly surprisingly-is that there is not yet such a thing as a truly 'Common Market'. Thus even neighbouring markets (e.g. France in relation to West Germany) can often represent as great a problem as a far-distant market such as the United States. Echoing this general situation, the European semiconductor industry basically consists (with the exception of Philips) of a number of predominantly national producers, each organized principally to serve the needs of its own national markets. One of the consequent liabilities is that Europe has nothing remotely to compare with America's Silicon Valley, nor is it likely that any meaningful geographical skill clusters will ever develop in the European IC industry. For the same general reasons, personnel mobility in Europe is also low, inhibited both by employment traditions and national boundaries, and is unlikely to increase significantly. The European research capability in solid-state electronics has always been high, although often commercially ineffective through an inability to bridge the gap between science and sales. However, this research capability will improve due to increasing government support, and increasing co-operation both between European laboratories and outwith Europe. Overall, therefore, in these three categories Europe comes behind Japan which, in turn, will be deficient in relation to America only in respect of personnel mobility. The final category-Economic Conditions-really warrants a complete paper in its own right if its relative importance is to be accurately assessed. Some aspects have already been covered earlier in this report but with regard to the relationship of the U.S.A. with other world nations the key points can very briefly be summarized as follows. In the years since the end of the Second World War, America has dominated the economic state of health of
Integrated Circuits: The Coming Battle the OECD nations. This strength has primarily been founded on cheap energy, abundant natural resources and a large enough population for the producers of manufactured goods to enjoy the benefits of considerable economies of scale. Meanwhile, however, other nations-in particular, Germany and Japan-have been recovering from the ravages of war and one of the pillars of American economic strength has eroded as the dramatic increase of oil prices has coincided with the gradual depletion of America's oil resources. For the future, therefore, there is likely to be much more balance of economic strength between the U.S.,EuropeandJapan which is the situation reflected in Figure 10.
Conclusions As a summary of the foregoing discussion Figure 11 shows the changes which will probably take place in the principal strategic factors on which America's ascendancy in the integrated circuits industry has so far been based, and on which the future domination ofVLSI technology may depend. The Past
U.S.Japan EuropeU.S.JapanEurope
IJJ
Jl ~ ~ H
~~
2. Market Availability
~~
IJI Ill)
3. Government Support 4. Business Climate 5. Venture Capital 6. Management Qualities 7. Research Capability 8. Personnel Mobility 9. Skill Clusters 10. Economic Conditions
~
~~
IJ t!ill ~ &J ~~ [81 ~~ ~ rEill IIJ tfJill [i!ll IJ tiEJ (2B ES) ffiill ~ ~~ ~~ B ~~ ~~ -~~ fl l\S)j [3j llj ~~ ~~~ B ~ ~ Jl [Sill ~ ~~ tuj 1m] liJ ~~ ~ • flliJj f£ill Ill • Ill
Figure 11. Summary: the shifting balance of advantage Key Relative advantage from bad to good: Bad ~
[;j II II
Good.
In brief, even allowing for the many remaining strengths of the American IC industry-such as its immensely strong technological base, its position on the learning curve, management-in-depth, etc;-and the fact that the European and Japanese managements will still suffer from important liabilities such as the relative absence of entrepreneurial freedom, there can be no doubt that the balance of advantage is now beginning to swing away from the United States. For this reason, in the VLSI era, American producers will face problems of daunting magnitude in maintaining global market share and innovatory leads against escalating transatlantic and transpacific competition, and the most probable prognosis is that the U.S. domination of this criticallyimportant industrial asector will eventually disappear, to be replaced first by a condition of approximate parity between America and Japan, who will possibly be joined somewhat later by Europe. It hardly needs to be said that the effects of these changes -on both component and equipment companies-will be immense and will need to be understood in detail if the changing competitive circumstances are to be digested without acute corporate discomfort.
The Future
1. Industrial Syngerism
[£]
37
Acknowledgments-Although the views expressed herein are strictly personal, some of the analytical work on which this paper is based was carried out during a major semiconductor study carried out by the author's consultancy company for a number of European governments, principal among them the West German Bundesministerium fur Forschung und Technologie, to whom thanks are due for permission to draw on some of the data collected during that project.
References (1) C. L. Hogan, An Address to Semicon/Europe 75, Zurich; November 4 (1975). (2) G. E. Moore, Progress in Digital Electronics, Proc. lnt'l Electron Devices Meeting; December (1977). (3) R. Turn, Computers in the 1980's, Columbia University Press, New York (1974). (4) I. M. Mackintosh, Applications of the Scanning Electron Microscope to Solid-State Devices, Proc. IEEE 53, p. 370, April (1965). (5) Y. Tarui, LSI and VLSI Research in Japan, Proc. lnt'l Devices Meeting; December (1977).