INDUSTRIAL DYNAMICS AND THE EVOLUTION

OF FIRMS AND NATIONS COMPETITIVE CAPABILITIES

IN THE WORLD COMPUTER INDUSTRY

Timothy Bresnahan* and Franco Malerba**

* Stanford University, USA

** University of Brescia and CESPRI, Bocconi University, Italy
April 1997
We would like to thank Salvatore Torrisi for sharing with us his knowledge about the U.K. industry; Simona
Heidemperger, Lu-Leng Chua, and Victoria Danilchouk for much assistance, and David Mowery, Richard Nelson,
Richard Rosenbloom, Giovanni Dosi, Martin Fransman, Steve Klepper, Sidney Winter, Peter Murmann and the
participants of the CCC project for useful comments and suggestions. We thank the Italian CNR and the Sloan
Foundation for research support.

1. The major issues

The evolution of the computer industry over the long term illustrates many of the broad themes
addressed by this book. In particular, it highlights the coevolution of technology, market structure,
and institutions, and addresses the sources of international competitive advantage. Since its inception
in the early 1950s, the computer industry has been characterized by rapid and sustained technical
change, continuous product innovation punctuated by a few major breakthroughs, creation of new
uses for computers and new markets and coexistence between

established actors and new

entrepreneurial firms. And since the beginning of the industry, one country -- the United States -- has
been the world technological and competitive leader.
From these remarks drawn from the history of the world computer industry, some questions
emerge. A first set of questions refer to the relationship between radical change and competition
among incumbents and new firms. How is it that old and new firms coexisted during the history of
the computer industry? Is there a link between radical innovations opening up new markets and the
competition between old and new actors? A second set of questions relate to the specific relationship
between technological change, market structure, and institutions during the history of the industry.
Was there a unique type of coevolutionary process during the whole history of the industry? Or was
there more than one coevolutionary process? In either case, why? A final set of questions relate to
the persistence of international technological and competitive advantages of one country during the
whole history of the industry. Why was the United States, and not other countries, able to profit from
the opportunities to become world technological and competitive leader? How was the United States
able to persist in that role? This chapter is going to try to answer these questions using a historical
and analytical perspective.
A first look at the major features of the computer industry identifies some aspects that are going
to be relevant for the analysis presented in this paper. First, computer hardware has advanced very
rapidly in price/performance measures, fueled by rapid advances in the underlying electronic
components as well as in computers themselves.1 A wide variety of hardware categories has
emerged: large and powerful computers such as mainframes, intermediate classes such as
minicomputers and workstations, and classes with less expensive products such as personal

For an extensive review of measurement studies of computer price/performance ratios, including extensive
discussion of alternative definitions of performance, see Gordon (1989). On any definition of price/performance, 2025% improvements have been sustained over four decades. For a key class of electronic components, semiconductors,
see the chapter in this volume by Langlois and Steinmueller, and Malerba (1985).

computers. Technical progress has made the largest computers much more powerful, the smallest
more affordable, and increased choice and variety in between. Computer hardware was once supplied
by a few pioneering firms; now there are hundreds of suppliers. The impact of performance increases
and price decreases, together with dramatic improvements in complementary technologies such as
software, storage devices, and telecommunications, and with considerable innovations and learningby-using by customers, has been to build a multi billion dollar worldwide industry.
Second, widespread adoption of computers in business and among consumers has contributed to
this ongoing growth. Three very different kinds of demand are important here. First are the buyers
of large computers for business data processing. These demanders are professionalized computer
specialists in large organizations.2 These sites have absorbed dramatic increases in computer power.
Thus, while mainframe computing sites number only in the tens of thousands, their total market size
is very large. A second kind of demand is that for individual productivity applications on PCs.
This is a newer body of applications, first reaching measurable commercial importance about fifteen
years ago. Most use depends on mass-market software, such as word processing or spreadsheet
programs.3 This market has seen much growth in two ways: by replacement and upgrading, and by
diffusion, as more and more (especially) white-collar workers have seen their work at least partially
computerized. In this segment, individual customers tend to buy hardware and software from a wide
variety of vendors in arms-length, market relationships. Unit sales of successful products run to the
millions. Individual demanders are small, however, so that through the late 1980s the market sizes
for the first two types of computing were roughly equal. The third demand is composed by scientific,
engineering, and other technical computation. Served by supercomputers, by minicomputers and
later by workstations, factories, laboratories, and design centers do a tremendous volume of
arithmetic.4 In total, technical computing market size is roughly as large as each of the two kinds of
commercial computing described above.
This variety in demand has permitted the emergence of different suppliers and markets. Market

They have close bilateral working relationships with the most important vendors. The computer business
systems are complex pieces of computer software, entailing an equally complex innovation process. In the industrialized
countries, most of the sites doing this kind of computing have been in operation for decades. A process of learning by
using, plus ever cheaper large computers, has led to considerable replacement and upgrading of facilities
3

See Mowery in this volume for a detailed treatment of software.

Here the users are technically sophisticated, and the applications have crisp technical goals. Buyer-seller
relationships are more like arms-length markets than close bilateral links.

segmentation has meant, for much of the history of the industry, that mainframes, minicomputers, and
PCs have served distinct kinds of demands. More importantly, demand variety has permitted the
emergence of new, entrepreneurial firms in parallel to established ones. Even as the oldest segment,
mainframe computers, was consolidating around a dominant firm (IBM) and dominant design in the
early 1960s, other firms (notably DEC) were creating the minicomputer segment. Later, other
segments such as personal computers, workstations, and superminicomputers would offer yet more
entry opportunities. This observation leads us to our first analytical distinction. We first treat the
industry dynamic surrounding the creation and persistence of IBMs leadership in mainframe
computing from the late 1940s to the late 1980s. We treat the industry dynamic which created new
markets by new entry separately, as its coevolutionary processes are fundamentally different.
Industry Dynamic 1: Creation and Persistence of IBMs Leadership in Mainframes
(late 1940s - late 1980s)
For our purposes, mainframe computers are systems used for large departmental or companywide applications.5 We shall cover mainframes from the very early period, the time before a clear
definition of a business computer or a computer company emerged, up to the late 1980s. IBM
emerged from an early competitive struggle to dominate supply, in the process determining the
technologies needed for computing, the marketing capabilities needed to make computers
commercially useful, and the management structures that could link technology and its use.
Competitors, customers, and even national governments have defined their computer strategies in
relationship to IBM. IBM was the manager of both the cumulative and the disruptive/radical parts
of technical change. Customers learning by using and IBM engineers learning by doing were
focused on the same IBM computer architectures. When an established technology aged, IBM was
not only its owner but also the innovator of the new.
Industry Dynamic 2: Creation of New Market Segments and Entry (late 1950s-late 1980s).
The second industry dynamic saw the founding and evolution of new computer segments and
markets. Minicomputers are machines intended for scientific and engineering use; when used in

The boundaries of the mainframe segment are not clear. Commercial minicomputers eventually became much
like mainframes. We do not treat the development of the commercial minicomputer segment. For our (international
comparison) purposes, the commercial minicomputer segment can be thought of as an extension of the mainframe
segment.

commerce they fall into support roles such as communications controllers. Microcomputers (personal
computers) are low price, small systems for individual applications, both in business sites and at
home.

Workstations are used by individual engineers in graphics and computation-intensive

applications such as design. In this dynamic, a series of new markets were opened up by
entrepreneurial startup firms. While there was some sharing of fundamental technical advance, each
new segments founding was characterized by considerable innovation and entry. As a result, the
technologies, as engineers use that term, of different segments were distinct. Those new segment
foundings that led to viable markets brought computing to new kinds of demanders. Successful firms
tended to be specialized. Buyers were departments of firms or individuals. Each of these segments
saw some maturation toward a dominant computer design, and toward a dominant model of the
appropriate supplying firm for the segment.
Industry Dynamic 3: Entry into the mainframe market by networks of small computers and rentdestroying challenge to IBMs leadership (1990s)
The 1990s have seen a third industry dynamic. Reversing the longstanding trends of Industry
Dynamics 1 and 2, this era saw competitive convergence of computers of all sizes in the 1990s.
Existing types of small computers were networked together and offered to IBMs traditional
customers. 6 The new technical and competitive importance of networks of small computers has
eroded the earlier market segmentation between mainframes, mini, and micro. After a decade of
stable segmentation, the distinct kinds of computers that had evolved in Industry Dynamics 1 and 2
came into direct competition with one another. Client/server platforms use computer networking
to link together user-friendly clients (such as PCs) with powerful servers (bigger PCs, workstations,
minicomputers or mainframes). The computer network consists of clients and servers, quite likely
sold by different hardware and software companies, and networking infrastructure, likely sold by yet
others. The networked computer became the platform on which large applications can be built. The
buyers in this area are a complex mix of individuals, departments, and enterprises. As we write,
neither the dominant design for a network of computers nor for a computer company in this
environment is clear.
The three industry dynamics have been characterized by coevolution of firms capabilities,

See Bresnahan and Greenstein (1995) for an explanation of this reversal.

5

strategies and organizations, technologies, and market structures, and (often) by a changing
relationship between the industry, public policy, and national institutions. As we will show more in
detail later in this paper, at the initial stage of each industry dynamic the coevolutionary process was
caused by the introduction of a new technology developed by an inventor or a firm. The new
technology spurred entry of new or established firms which added modifications and changes to the
original technology. In our three industry dynamics, the new technology addressed a new demand
and new types of users, or old types of users in a radically new way. This created specific types of
user-producer relationships, which led firms to develop new competencies and organizations. Over
time, firms developed appropriate competencies, strategies, and organizations more suitable for the
new technology. These firms made new innovations and fostered the rate of technical change in
specific directions. Relatedly, public policy and institutions were relevant at various stages of the
evolution of the computer industry: some policies and institutions remained unchanged over time,
while others evolved in tune with the specific industry dynamic and changed in various degrees and
forms in different countries. As we will show later on, however, this coevolution of technology,
firmscapabilities, strategies and organizations, industry structure and public policy and institutions
proceeded differently in each industry dynamic, with congruence between firmsstrategies, industry
characteristics, and countriesperformance.7
A first look at the three industry dynamics shows that the United States has been persistently the
innovative and commercial world leader in the computer industry. In Industry Dynamic 1, IBM
emerged as the world leader. IBM, and therefore the U.S., persisted in its leadership despite
competitive attacks from individual companies, strategic alliances, and even whole national computer
industries such as Japans. In Industry Dynamic 2, new successful American firms entered emerging
market segments while IBM continued to maintain its dominance in mainframes. Finally, in Industry
Dynamic 3, the challenge to IBMs leadership came mainly from American firms, as a result of the
convergence between mainframes and networks of smaller computers. We examine the different
reasons behind American success in each of the three periods.
The paper is organized in the following way. In sections 2 to 6 we discuss industry coevolution
and the international competitive advantages in a comparative way (the United States, Europe, and
Japan). In section 2 we analyze the birth of the new industry during the late 1940s and early 1950s
and the emergence and persistence of IBM world leadership in mainframes for more than four
7

A segment not treated is workstations. The history of this segment has many of the same elements of the
microcomputer history.

decades. In sections 3, 4, and 5, we discuss the growth of new market segments (mini and micro
computers respectively) and the entry, success and growth of new American actors who became
world leaders. In section 6 we examine the challenge to IBM domination in mainframes coming from
computer networks and other existing American firms. Finally, in section 7 we draw some general
conclusions from the analysis.
2. The birth of the new industry: major similarities in the United States and Europe
The early period of many pioneering efforts leading to the emergence of a mainframe computer
industry was characterized by extreme similarity in the initial conditions and types of entrants between
the United States and Europe, with Japan lagging behind. The scientific capabilities of the universities
in the United States and Europe and the structure and capabilities of the connected industries--office
equipment and the electrical-electronic industry--were very similar.
2.1 The major role played by universities around the world in the pre-commercial period
In the very early years of the industry (1940s) universities in both the United States and Europe
were active at the scientific and prototype levels. In the United States, scientists in universities
worked in cooperation with the government which acted both as a source of funds and as a major
potential consumer of technology, notably in the military and the Census Bureau. In addition, a few
private firms also funded basic research in computers (IBM, for example, supported the development
of the Mark I computer of Aiken at Harvard in 1944).
The scientific commitment by various American and European universities to develop digital
electromechanical computers led to the development of several machines during the 1940s. In the
United States, for example, ENIAC, the first digital electromechanical computer, was developed by
Eckert and Mauchly in 1946 with the support of the Army at the University of Pennsylvania. The
military also sponsored Project Whirlwind at MIT leading to the development of magnetic core
memory.
European research in computers was at the world scientific frontier during the 1940s. In the UK,
after the development of the Colossus (1945) (a computer similar to ENIAC) and ACE (1945),
Manchester and Cambridge Universities moved to the forefront of research. Manchester University
developed the Mark I (1948) (the first computer to use a magnetic drum memory) and Manchester
Universitys Digital Machine (MADM) (1949), and Cambridge University developed the EDSAC
(1949). In Germany, Konrad Zuse, of Berlin University, was very active in this field. During the
7

1930s, he built a series of electromechanical calculators (Z-1, Z-2, Z-3, and Z-4). The Z-3 was the
first one fully operational, while Z-4 was installed in a German V-2 rocket plant and at the end of the
second world war was leased to the Swiss Polytechnic Institute.
In the United States, the origin of the commercial computer industry came with the move of the
two inventors of ENIAC from university into the business world. In 1946 Eckert and Mauchly
established their own firm, the Eckert-Mauchly Corporation, to develop general purpose commercial
computers for scientific as well as for business accounting uses. The Eckert Mauchly enterprise
failed, however, because external finance (particularly venture capital) was limited.8
Later on, in the 1950s, the role of universities expanded to include participation in the
development of technologies that would be used in government projects or embedded in existing
firmsproducts.9
2.2 Three types of entrants around the world
Mainframes were the first commercial products developed and sold in the computer industry
during the 1950s. They made extensive use of magnetic drum memories and, later on, of transistors.
Early computers functionality focused on scientific uses, then expanded to business uses such as
accounting. The first mainframe producers maintained links with universities and continued to have
government (military) support. They also opened up extensive linkages with business firms who
became major users of computers.
The success of IBMs model of a computer company, characterized by a Chandlerian three
pronged investment in technology, marketing, and management, is now very familiar. We should
remember, however, that this model was not obvious to market participants in the early period.
Three distinct types of entrants entered the industry: office equipment producers, electronics
firms, and new firms. Among office equipment firms, the most prominent were IBM, RemingtonRand (later Sperry Rand), Burroughs, NCR (which later on bought CRC), Olivetti, Bull, and BTM
(an IBM distribution partner). Among electronics firms, General Electric, Honeywell, RCA, Siemens,
Standard Electrik Lorenz, Telefunken, GEC, and Ferranti.

New firms were less important

competitively but included CDC, SDS, ERA, CRC, Eckert-Mauchly (sold later to Remington Rand),
8

Some (albeit insignificant) financial support came from firms such as Prudential, Nielson and Northrop
Aircraft. These financial sources were not sufficient and in the 1950 the Eckert-Mauchly Corporation was acquired by
Remington Rand.
9

See OECD (1966) and Flamm (1988) for a description of the American and European efforts.

Nixdorf, and Zuse. All three types of entrants were in both the United States and Europe.
At this stage, Japanese firms did not have an important commercial presence in computing
(Flamm (1987)). By the late 1950's, however, there were several Japanese technology initiatives
under way. What was to be the largest Japanese computer company for some time, IBM Japan,
was engaged in protracted negotiations with Japanese government agencies. These negotiations
resulted in substantial technology transfer, including favorable licensing of IBM products to the
Japanese industry (Anchordoguy, (1989)).
In Japan, entry occurred later than in the United States or Europe. Heavy electric equipment firms
(Toshiba, Mitsubishi) entered alongside consumer electronics firms (Matsushita, NEC, and Hitachi)
which had capabilities in both areas. Japanese regulatory treatment of telecommunications permitted
NTT an active role, encouraging its suppliers of communications equipment, notably Oki Electric and
Fujitsu, to enter as well. No Japanese business equipment firms (like IBM or BTM overseas) entered.
Thus, Japanese entry differed in being later and purely electronically based.
The three types of entrants reflect the specific technological and market features of the new
worldwide industry. Computers were a new electronics good which attracted several existing
electronics producers already active in other electronics fields. Similarly, some of the first applications
of computers were in business, attracting firms with established connections to business data
processing. Interestingly enough, the tension between technology-based and market-oriented firm
organizations and competencies was going to reappear throughout the history of the industry.
(Davidow (1986)).
Despite the early enthusiasm, there remained fundamental uncertainty on the technological
development of the industry, the range of applications, and the potential size of the future market
(Rosenberg (1995)). The nature of commercial use of computing and the potential size of the market
were unclear. Similarly, the appropriate business model and strategy for a computer company were
not known. In particular, it was not well established whether the primary usages for computers
would be in making calculations or in processing data. Neither was it certain whether the largest
demand segments would be military, scientific/engineering, commercial, or other. These uncertainties
meant that the most important directions for technical progress, such as the relative importance of
calculation speed vs. storage access, were unclear. They also meant that the nature of buyer/seller
relationships and of commercialization efforts were unsettled.10
10

See Katz and Phillips (1982) and Usselman (1993) for a discussion of this uncertainty with particular regard
to how it impacted different firms in the industry.

During the 1950s, the three types of entrants-office equipment producers, electronic firms, and
new firms-had distinct capabilities and distinct strategies, which were similar in Europe and the United
States for each group of entrants. The electronics-based firms faced the challenge of either building
or acquiring a business-equipment marketing capability--including a substantial field sales force--or
finding a way to succeed without it.11 Many found the potential profitability attractive enough to
overcome this barrier.12 Firms with business equipment capabilities quickly recognized the need to
add technological ones. Some, like Remington-Rand or National Cash Register (which bought CRC),
attempted to obtain these capabilities by acquisition. Internal development of technical capability was
the more common strategy, undertaken by IBM and Burroughs in the U.S. and by Olivetti and BTM
in Europe.13
To succeed, startups such as CDS, SDS, DEC, or Eckert-Mauchly would need to develop both
technical and marketing capabilities. Many startups focused on the more technical side, producing
computers only for specialized uses. Firms of any type would need remarkable financial resources
because neither the technical nor the marketing capability could be built without considerable capital.
In Europe, some firms focused on niche strategies: Nixdorf, Zuse, Telefunken, and so on. From
the beginning, Nixdorf had the goal of producing small computers for specific uses. In 1952, Heinz
Nixdorf founded the Labor fur Impulstechnik (which only in 1968 changed its name to Nixdorf).
Nixdorf developed the first vacuum tube calculator for accounting for RWE and then built several
computers for Bull and Wanderer. By this route, a capable computer firm in the small-user niche was
built.
Among office equipment firms, a major tension arose between the established mechanical and
electromechanical core competencies and the new emerging electronic ones. For example, the office
machinery producer Olivetti entered electromechanical calculators in 1949 by creating (with Bull) a
commercialization company for the distribution of Bulls tabulating machinery. In the early 1950s,
it started doing R&D on electronic computers in various locations in Italy and the United States.

11

One class of technically capable potential entrants, telephone companies, did not enter. In the United States,
AT&Ts entry was blocked by an antitrust consent decree.
12

Siemens, for example, decided on a full scale entry in 1954, focusing on mainframes for commercial and
scientific uses. At the end of the 1950s it produced the first 2002 computer. It must be noted that during the 1950s
Siemens contacted IBM for cooperation in computers. Actual cooperation, however, was limited to the supply of
readers of punched card and magnetic tape machinery and did not include mainframes (Malerba (1985)).
13

For a discussion of BTM and Ferranti, see Freeman (1965), Malerba (1985), and Usselman (1993).

10

Conflicts between the dominant mechanical competence and culture of Olivetti and the emerging
electronic one took place since the beginning. The new Electronic Division and the new factory that
was going to produce the first electronic calculator, the ELEA 9003 (developed in 1959), were
located near Milan, far away from the central headquarters in Ivrea in order to be isolated from the
prevailing mechanical culture.
2.3 American Government Encouragement of (Domestic) Industry Development
Substantial government backing for the early U.S. computer industry offered advantages to firms
in the United States. There is little support, however, for the view that the U. S. government bought
success for IBM and no support whatsoever for a strategic trade policy view of U.S. government
actions.
Some of the purely technical capabilities needed to build computers were backed by Federal,
especially military, research funding. Technologies such as transistors and core memories were
developed in laboratories - AT&Ts and MITs - before the semiconductor industry took off in the
mid-1950s. Some of these developments were heavily dependent on Federal money (Flamm (1987)).
Further, many early U.S. computer systems were themselves directly supported by federal research
funds. In this manner, the Army supported ENIAC; the Navy and Air Force supported Project
Whirlwind at MIT; the Census Bureau supported UNIVAC, and so on.14 Further, it was clear that
the military was going to purchase many computers from domestic suppliers. At a broad, general
level it is clear that this environment worked to the advantage of IBM. Government support of
technical capability development meant that IBMs existing marketing capability could be integrated
into a full Chandlerian three pronged investment more effectively. Indeed, many of the technical
developments for defense computing were commercially useful. Ultimately, the Defense Calculator
became the IBM 701; much of SAGE was valuable in SABRE; and so on. Similarly, reduction in
uncertainty and increases in general technical knowledge obtained from defense computers may have
been valuable commercially.
Yet this dual-use or spillover story was not, in fact, an important factor in IBMs success. The
U.S. government actions were far removed from intentional strategic trade policy aimed at creating
a national champion: IBM. Defense Department agencies supported the development of a domestic

14

Flamm in chapter 3 tabulates military research support in detail.

11

computer industry.15 National defense goals, however, could easily have been met by a largely
technical computer industry, with marketing capabilities focused on defense procurement procedures.
Indeed defense support for computer systems development at IBM (including the Naval Ordnance
Research Computer-NORC) and for the 701 Defense Calculator (which was financed by IBM but
pre-sold to defense customers) was a small fraction of total IBM's effort. Furthermore, government
funding at Eckert-Mauchly Computer Corporation and Engineering Research Associates, both
purchased by Remington Rand, was actually intended to put IBM at a competitive disadvantage.16
Defense Department, however, did act like a well-funded demander with a real need for computerbased weapons systems, and let the supplying industry structure emerge in the marketplace.17
Another branch of the same government, the U.S. Department of Justice, worked actively to
prevent the emergence of IBM as the computer industry dominant firm. In particular, the U.S.
Department of Justice was systematically against IBMs strategy of strengthening marketing and
technical capabilities within the same firm. It therefore worked directly against government support
of the three pronged investment. In two antitrust lawsuits, the U.S. Department of Justice sought
to characterize IBMs marketing capability as anticompetitive.18 The 1956 consent decree between
IBM and the government had, among other provisions, stark limitations on IBMs use of service
bureaus as a sales device. The second antitrust lawsuit, brought in 1965 and contested for over a
decade, viewed IBMs service, sales, and support efforts as anticompetitive lockin devices.
The legislative branch was also anti-IBM, tilting procurement policy against IBM. Here the issue
was more political, having to do with spreading out government procurement funds to states where

15

Soviet production of nuclear arms led to very considerable investment in anti-aircraft defenses, whose C3I
part--realized in the 1960's in the SAGE system--was heavily computer based.
16

The IBM eye-view of early developments was that they constituted Government Funded Competition.
Only because of this threat did IBM reverse its standing policy against Federal research collaborations, using Navy
money to design the NORC (Pugh (1995)).
17

Indeed, Usselman offers a very interesting argument that U.S. procurement policy favored IBM only because
it took this form (Usselman (1993)). IBM would not likely have been chosen as the national champion in the critical
early phases, nor would a supply side procurement policy have led to the development of the IBM commercialization
capabilities.
18

The 1956 consent decree between IBM and the government had, among other provisions, stark limitations on
IBMs use of service bureaus as a sales device. The second antitrust lawsuit, brought in 1965 and contested for over a
decade, viewed IBMs service, sales, and support efforts as anticompetitive lockin devices.

12

IBM was not as important an employer as other system companies.19 Besides, IBM did not initially
understand the power of a single computer architecture which could serve both military and business
clients and did not expect its Federal support to provide spillouts to its commercial data-processing
business. In the critical early stages, IBM had separate development efforts and built separate
capabilities to pursue new military/scientific customers and to supply its traditional
business/commercial customers because it conceived these two lines of business as distinct.20 It
therefore appeared to believe that electronics offered good opportunities for both commercial and
governmental business.21
Only ex post was it clear that the industry was one in which technology was general purpose.
Spillouts from one use to another eventually proved to be significant but not anticipated at the time.22
In conclusion, government policy was only accidentally favorable to the creation of IBM as a
worldwide leader. Parts of the government opposed IBM's most critical investments. The supportive
parts of the government were passively pursuing their own policy goals and were not following any
type of strategic trade policy. No one, not even IBM, saw the implications of the positive role of
government support for business applications.
3. The rise and persistence of IBM world dominance in mainframes
In the late 1950s and early 1960s, IBM became the world leader in mainframes, and remained so
for thirty years. IBM was able to be highly innovative in mainframes, addressing the demand coming
from big or medium size users, government, and universities. Three stages can be identified: the rise
to worldwide leadership through a high commitment to the Chandlerian three pronged investments;
the consolidation of leadership through innovation in a modular product line (IBM System 360); and
the continuation of its dominance through waves of highly successful products (IBM System 370,

19

There is an active debate on whether government agencies were able to evade the law and procure IBM as
they saw fit. See Greenstein (1992). There is no debate about the anti-IBM policy itself.
20

See Usselman (1993) on the distinction between IBMs facilities at Poughkeepsie and at Endicott.

21

See Pugh (1995) on the discussion about IBM initiatives in making its tabulating machines more electronics
in the late 1940s, based on beliefs about the future of the tabulation business. See also Katz and Phillips (1982) on the
observation that IBM understood that commercial as well as defense applications might be useful early on.
22

Indeed, spillouts to IBMs commercial business from its defense business were important. See Katz and
Phillips (1982) or Flamm (1988) for an accounting of some very important technical developments first appearing in
defense sector products.

13

etc.).
3.1 The generation of IBM world leadership: the role of the three pronged investments
In the 1950s and early 1960s, IBM introduced highly successful families of computers such as the
701 (1952), 650 (1954) and the 1401 (1960).23 Each of these new families involved considerable
development of new computing technology, not only processing power but also peripherals. Moving
from defense calculation to business data processing, IBM supported business systems very
effectively at customers sites, and built a capability to address customer wants and needs. This
combination of technical drive and customer focus was difficult to achieve, requiring new
management structures. IBM's success sprang from its major R&D investments and more generally,
the Chandlerian three pronged investments (managerial capabilities, technology, and marketing).24
IBM rapidly became the world market leader because of its continuous R&D effort in developing new
products, coupled with advanced manufacturing capabilities, excellent marketing competence, and
management structures keeping technology and market aligned.
Though much of the attention of the early computer industry was placed on technical calculations
for military and scientific purposes, commercially-oriented companies like IBM and Remington Rand
were quick to move toward business uses. This movement was not simple, however. Within IBM
management, there was wide resistance to building a general purpose computer on the grounds that
demand was limited. In the very early 1950s, creation of a computer such as the 701 would call for
a large financial commitment and much R&D. This was to be a stored-program computer, able to
use the binary number system instead of the decimal, and would be pre-assembled by IBM for rapid
setup (and later rapid upgrades) at the customers site. All of these features called for new, and
expensive, technological and production capabilities. The resistance was overcome by IBMs new
leader, Thomas Watson, Jr., and IBMs 701 quickly overtook Remington Rands Univac, the leading
computer at the time. The 701's combination of advanced technology and customer-aware features
like rapid setup were well matched to IBMs strong marketing presence (already built with tabulating
card machines) in business data processing.
The resistance within IBM continued with the IBM 650 Magnetic Drum Calculator, which was

23

This section draws heavily on Fisher et al. (1983).

24

See Usselman (1993), Sobel (1986).

14

to become the Model T of computers (Cuthbert Hurd).25 Many IBM managers thought that the
computers rental price, over $1000/month, would limit demand too sharply to cover the large
development costs. IBM was making progress on a broad technical front: the 650 had 1000 10-digit
words of memory, and vastly improved diagnostics, for example. Optional features made it more
suitable for calculation or accounting. Accounting reports called for developments in a wide variety
of peripherals, including printers, magnetic tape storage, and magnetic disk drives. IBM could offer
customers an integrated solution, having made progress on the computer and the variety of
peripherals. After being pushed forward into the marketplace, the 650 became the most successful
computer of the 1950s, with about 1800 units shipped.
The importance of integration and of wide technical progress continued to be important
throughout the decade. The IBM 1401 was, indeed, much faster and more reliable than the previous
generation 650. Perhaps more important for its use as an accounting machine was the breadth of
peripherals. It could use both punch cards and magnetic tape, worked with the simultaneously
introduced 1403 chain printer. Sales of this machine were approximately one quarter of storedprogram electronic computers.
On the marketing side as well, IBM pushed ahead. The company made an early and expensive
commitment to field service and support, and to customer education. This took place in an era of
widespread ignorance about the potential uses of computers. This marketing commitment was not
only domestic.

IBM succeeded in dominating the European as well as the Japanese mainframe

markets. In Europe, IBMs superiority in products and customer assistance was coupled with a local
presence on the main markets. For example, IBM UK and IBM Germany tried very hard (succeeding
to a large extent) in being considered respectively a British and a German company. In Japan, IBM
Japan was for a long time first in revenues among "Japanese" computer companies. The strategy
reflected powerful marketing forces. IBM used the "IBM World Trade" model, making itself as local
a company as possible. This meant involving nationals in almost all roles, including senior
management. The point of this extreme localization was to ensure that relationship selling efforts
worked.26 The scope of IBMs three pronged investment was global, with worldwide exploitation of
scale economies in designing and building computers linked to a local marketing organization.

25

Interview, Winter 1995.

26

In the mid 1980s, IBM Japan was replaced by IBM Asia/Pacific, a less Japanese entity headed up the U.S.
expatriates.

15

IBM certainly did make large mistakes. The STRETCH committee, an effort to advance its
technical capabilities and to build a high-speed computer, turned out to be a commercial failure. IBM
used many of the technical lessons it learned in STRETCH in later products, but did not recover its
development expense from the project itself.27 Yet overall it succeeded in developing a very strong
technological and marketing potential.
The computer business in its early phases took place in a competitive environment under
tremendous increasing returns to scale. All of marketing, technology, demand and company business
model were still to be invented. Since the first two were quite expensive and the last two very
uncertain, investments were highly uncertain. From a company perspective, IBM pushed forward
with very expensive investments in technology, marketing, and management, thereby putting
tremendous pressure on rivalscapabilities. As a consequence a very concentrated structure with a
single dominant firm emerged.
3.2 The consolidation of IBM world leadership: IBM system 360
By the early 1960s, the main technical features of a commercially useful computer were clear and
many of these features were already embodied in products. Important problems remained, however,
notably incompatibilities between computers in different families. The final stage in this evolution was
the introduction and implementation of the completely modular, compatible computer family. This
was the IBM System 360. This product line, and the organization that improved and sold it, became
the dominant industry model.
Incompatibilities between different computers forced costs and delay on using companies. In the
early 1960s, computers were designed to be used either for commercial or for scientific uses, not for
both. If a user wanted to run both applications, he needed two separate machines, support staffs that
understood two different bodies of technical knowledge, and so on. In addition, IBM's machines
themselves were not compatible across the broad family of processors: a program run on one
processor could not be used on another. Users whose needs grew over time could exhaust the
capacity of their existing computer and need to reprogram for the next larger model.
IBM's System 360 was designed to solve these user problems. It was, however, neither an easy
decision to make nor a straightforward technical feat. Technically, compatibility over a wide range
of sizes and uses of computers changed the design process. Designers could not optimize for specific
27

Some of these technologies were quite important, including transistors, a parallel architecture, and
manufacturing processes needed to integrate many different components.

16

purposes, but instead needed to be coordinated in the production of general purpose components that
could be specialized. Such coordination raised costs and slowed the invention process. IBM
appointed a task force called the SPREAD Committee to develop and supervise the project.
Developing the new machine required such a big commitment that the venture came to be popularly
known as a "bet the company" initiative. IBM virtually employed all its financial, technological, and
human capital. Overall, it devoted a larger percentage of its revenue than it (or any of its competitors)
had ever spent on any of their projects.
The decision itself was difficult. Designing a new family of computers meant that there would be
future compatibility, but implied incompatibility in the present. IBM executives responsible for the
1401, for example, argued that competitive pressures forced an investment in improved 1401s.28 The
decision also meant abandoning investments in products under development, even successful ones.29
The IBM decision to end the proliferation of product lines and put resources into a single
development effort was risky and expensive. More importantly, it showed that the management
control structures needed for the three pronged strategy were in place.
The 360 made existing machines obsolete. Even though it was not directly compatible with the
existing processors, users found switching irresistible because of the tremendous benefits of the new
system.30 Among the many distinguishing features of the 360 was price/performance superiority not
only over IBM's own computers but over the systems of all its competitors. The success of the 360
was immediate and gigantic: more than a thousand orders were made in the first month after
announcement, and many more thousands followed.
The introduction of the IBM System 360 drastically changed the structure of IBM and of the
overall computer industry. With System 360 the standardization of components and software
allowed the exploitation of economies of scale in component production and the consequent upstream
vertical integration of computer firms, the supply of a broad line of compatible computers, and the
introduction of modularity and incremental modifications in computer design. From a company

28

After Honeywell introduced its H-200 computer, IBM experienced 196 lost 1401 sales in 8 weeks. At that
time, the chairman of SPREAD was also head of the General Products Division, responsible for the 1401. The
argument for further 1401 investment lost to the new System 360 concept despite such powerful opposition.
29

The almost finished 8000 project was labeled a "wrong" approach despite the enormous financial resources
already devoted to it and was discontinued.
30

The irony is that to get its customers into the circumstances where they would not need to switch computer
families for over twenty-five years, IBM had to get them to switch at the beginning.

17

perspective, this event showed the realization, after some early confusion, of the importance of scale
and scope economies in computer design and commercialization. From an industry perspective, the
event was even more important. Success of this strategy at IBM defined a level of innovation and
commercialization capability to which other firms, worldwide, were going to be compared by
customers.
3.3 Continued innovation by the dominant firm: the IBM system 370
The 1970s saw the continued dominance by IBM in the mainframe segment. IBM introduced a
new family of computers called System 370. These computers represented considerable technical
advance, much of its own-competence destroying. As with the System 360, IBM was prepared to
give up the rents on existing product lines and on the technologies embodied in them. Backward
compatibility of the System 370 helped to preserve IBMs position, for customers did not see much
destruction of their own computer-using competencies. They could and did continue to use IBM
machines predominately.
3.4 Strategies and evolution of the losers to IBM around the globe
In the era of the IBM System 370, the scope of geographic product markets was once again
worldwide. IBM was setting the standard in two senses. First, competitive products needed to
respond to IBM price/performance, software availability, and service and support levels. Second, a
critical technology choice was whether to establish an IBM-compatible product line or to attempt a
separate, incompatible platform. The IBM-compatible strategy would involve development of
expensive R&D and manufacturing facilities. IBM kept these costs high for rivals, pushing the
technical frontier with repeated own-competence destroying innovations. The 1970 unbundling of
IBM hardware and system software, under pressure from U.S. antitrust authorities, made the IBMcompatible strategy easier in the System 370 era. Yet rivals found themselves followers to IBM,
producing specific components for some customers but never gaining control of the system/370
architecture. The IBM-incompatible strategy was even more expensive, for it involved departures
from IBM complementary software and support infrastructure. To purse the IBM-incompatible
strategy, a competitor would need substantial marketing expenditures for a field sales and service
force, software, and so on, as well as technical capabilities comparable to the leaders. Firms also
faced marketing choices: would they compete for mainstream large-company data center customers
or find smaller market niches?
18

Facing IBM success, American, European, and Japanese firms adopted variants of these follower
strategies. In the following pages, we show how a strategy of direct competition with IBM using a
competing platform failed, but how niche strategies and IBM-compatibility strategies permitted
survival without threatening IBM dominance.
a. Head to head confrontation with IBM: always a losing strategy
Some rivals introduced a competing platform, trying to replicate the IBM three pronged
investment. This expensive strategy involved the development of a number of computers, each
compatible with the others, and development or encouragement of correspondingly compatible
peripherals, software, and end-user knowledge. Effective competition by this method involved
building two very distinct assets. One was technological: the capability to design and manufacture
computers and close technological complements. The other was marketing: the ability to help
customers turn computer technology into useful solutions. Such attempts failed all around the world
because the electric and electronics companies were not able to develop effective marketing
capabilities, while business equipment companies were not able to move the capabilities of the
company fully into electronics and to bet completely on the electronic technologies.
This tension between business equipment companiesexisting core competencies and electronic
technologies is well illustrated by the case of Olivetti. Olivettis increasing commitment to mechanical
office equipment technologies, particularly after the purchase of Underwood in 1959, drained a
massive amount of resources and did not allow committing considerable resources to mainframes.
Consequently, Olivetti decided to focus on mechanical technologies and traditional office equipment
products rather than on electronics and mainframes and sold its Electronic Division to GE (later on
sold to Honeywell) in 1964 (Malerba, 1985).
b. Compatibility with IBM: a losing strategy in the 1960s becomes a survival strategy in
the 1970s
Another strategy available to competitor companies in these circumstances was the acceptance
of the IBM System 360 platform standard and the sale of IBM-compatible computers. A related
strategy was building IBM-compatible peripherals such as storage devices, taken up by plugcompatible manufacturers (PCMs).
Most of the initial attempts to introduce IBM-compatible computers failed. For example, RCA

19

built completely IBM-compatible computers, the Spectra Series.31 RCAs initial marketing plan was
built around availability and superior technical features at the time of IBM order backlogs. Reliability
problems and other technical difficulties plagued the Spectra through the System 360 era. Though
RCAs effort was profitable for a period, the operations were never really large, and were ultimately
sold to Univac.
RCA's failure also involved its collaborative partners around the world. In 1964 Siemens, after
realizing that its new 3003 Series was already obsolete, started a collaboration with RCA, which was
developing the new Spectra 70 series. RCA gave Siemens a license and technical assistance for the
Spectra 70 series, which was shipped as the Siemens 4004 in 1965. In the meantime, the IBM System
360 was overtaking the European as well as American markets. For Siemens, the collaboration with
RCA proved initially successful at the commercial level: during the 1960s and early 1970s, Siemens
increased its market shares (in Germany from 5% in 1965 to 16% in 1972). This was also due to the
introduction of a line of smaller processors (302, 304, 305) and the acquisition of Zuse (1966). Later
on, however, RCAs withdrawal from computer production (1971) left Siemens without a partner
with advanced technological capabilities.
Later on (1970s), with the unbundling decision, the compatibility strategy became more
successful. The compatibility strategy was pursued by a group of American, European, and Japanese
companies, often related through agreements of various types.
In particular, the IBM-compatibility strategy of Amdahl, Fujitsu, and Hitachi became successful
for a long time in the world market. Amdahl, founded by an ex-IBM designer, was in the IBMcompatible business from the beginning. It had considerable success as a market follower. Japanese
firms, which had been investing through domestic consortia in computer technologies, began to make
IBM-compatible exports. International linkages were important here. Hitachi had a partnership with
RCA, and had learned much about the IBM-compatible business from that experience. Fujitsu was
lending money to Amdahl from 1972 and later acquired an ownership stake. In an era of IBM
dominance of world markets, only Japanese and Japanese-cooperating sellers succeeded in becoming
effective followers through the IBM-compatible strategy.
After an initial period, these new compatible entrants made IBM-compatible computers less
expensive. This transferred some profits from IBM to customers. However, control of the System

31

Another example was Honeywell's earlier offer of IBM 1401-compatible machines unfortunately introduced
at the time of the System 360 rollout.

20

370 architecture, the operating system, and other key software components, as well as the world
dominant market share in mainframe computers remained with IBM.
Unbundling by IBM and the monitoring role of American antitrust authorities also made
peripheral sales by PCMs somewhat easier. The same forces also made third-part leasing and other
financing more widely available. This had the effect of lowering returns to IBM. Makers of
compatible computers and peripherals limited IBMs market power and its ability to price
discriminate. However, it is difficult to conclude that any of this direct, compatible competition had
any immediate impact on industry evolution. IBM retained control of the mainframe architecture and
IBMs profitability was not destroyed. Actually, its broader competitive effect was to enhance the
attractiveness of the IBM platform to customers.
c. Niche strategy: always a survival strategy
Another strategy was to avoid head-to-head competition with IBM entirely, seeking out a body
of customers left unserved by the 360 platform. Three kinds of niches emerged: specialized
commercial market niches, governments, and protected domestic markets for the non-U.S. producers.
c.1 Market niches
In the United States, the most successful versions of this strategy were niche efforts. CDC was
very successful with scientific users. NCR found industry-specific niches, notably with smaller
computers, in retail trade and banking. In Europe, contrary to mainframe suppliers such as Standard
Electrik Lorenz and Telefunken which exited completely the computer market, some firms decided
to enter the market for mid-range systems (comparable with IBMs systems S/32, S/34, S/36 and later
AS400). For example, during the 1960s and the 1970s in Germany several firms, such as Nixdorf ,
Konstanz, Triumph Adler, Kienzle, Dietz and Krantz, started to produce mid-range systems. These
mid-range systems were all proprietary, focused on sector-specific applications and with specific
software. These companies (particularly Nixdorf) experienced major success until the 1980s. The
introduction and diffusion of microcomputers and the emergence of standard bundles led to the rapid
disappearance of these firms (see later on the discussion on the crisis of Nixdorf).
c.2 Government niches
A very important niche was procurement by the government, for both military and civilian uses.
In the United States, procurement policy, under pressure from congressman Jack Brooks
21

government operations subcommittee, was anti-IBM in intention and effect. Honeywell and Univac
took advantage of this, and had much more success in the niche market of Federal government
computer purchases than in the broader commercial computer market. In the other countries, every
government protected its weak domestic firms from IBM by use of procurement policy.
c.3 Niches by leading domestic customers
In addition, large domestic customers had close relations with large domestic mainframe
producers and would buy domestic. A buy-German or buy-French attitude among large firms and
large institutions was present during the 1960s. In Japan, large Japanese buyers, whether the
government or NTT, were influenced to buy Japanese computers. Throughout the 1960's, import
market share in Japan fell steadily as Japanese firmscapabilities grew.32
3.5 Coevolution in Europe: the decline of the industry and the role of European national
champions policies as exit barriers
Faced by the mounting competitive challenge from IBM, European producers declined in
competitiveness. As previously mentioned, European producers either did not invest enough in R&D
or did not have advanced manufacturing and marketing competencies. Some of them (particularly
old office equipment producers such as Olivetti) continued to have difficulties in absorbing the new
electronic technology and culture and did not want to abandon mechanical or electromechanical
technologies. Other producers had major productive and coordination problems. For example, ICL
inherited two incompatible mainframe product lines, one from ITC and the other from English
Electric, and kept them incompatible during the first part of the 1970s, while developing a third
(incompatible) line. All three were also IBM-incompatible. The severe crisis faced in the late 1970s
and early 1980s led ICL to move to IBM-compatible strategies, reduce its R & D expenditures on
mainframes and increase it on other computer types, while focusing on specific markets: defense,
retailing, and financial services. Specialization in these vertical markets still characterizes ICL.
Moreover, in their search for international alliances, some European firms chose big losers in
the technological and commercial race. We have already mentioned Siemens' links with RCA. Also

32

Above and beyond the IBM patents, most Japanese firms sought overseas partners. Oki Electric had Sperry
as a partner from 1963, Hitachi-RCA from 1961, Toshiba-GE from 1964; Mitsubishi-TRW; NEC-Honeywell from
1962. Matsushita never had an overseas partner but had negotiated with Philips before exiting the computer business in
1964. The important exception was Fujitsu, which had no foreign partner at this stage.

22

during the 1970s the main European mainframe producers continued to foster their links with foreign
firms, without much success: Siemens with Fujitsu, ICL with Fujitsu, and CII with Honeywell.
During the 1980s ICL set up a cooperation with Fujitsu (in which Fujitsu supplied components and
semiconductor design tools, while mainframe architecture, design and software were developed by
ICL). In 1991 Fujitsu acquired 80% of ICL. Fujitsu, however, did not decrease ICL's R & D
expenditures and let ICL follow quite autonomous marketing strategies in Europe. Siemens
continued to follow a strategy of international alliances with key producers in mainframes. After the
failures of the alliance with RCA and the UNIDATA experience, Siemens' search for a partner led
the company to start a cooperation with Fujitsu (1977) by selling Fujitsu's large mainframes under
Siemens' label. Moreover, Siemens founded Comparex with BASF (which was already marketing
Hitachi's computers in Europe) for the production of IBM-compatible computers. While cooperation
with Fujitsu is still on at the present time, Siemens' involvement in Comparex was strongly reduced
in 1988.
In Europe, public policies represented a major exit barrier from mainframe production for
national champions such as Siemens, CII, and ICL. First, governments intervened by supporting
directly or indirectly the mergers between unsuccessful companies in order to create national
champions. For example, in 1968 ICL was formed from the merger of International Computers and
Tabulators (ICT) (already incorporating the computer operations of BTM, Ferranti, General Electric
Powers and EMI) and English Electric Computers (EEC) (already incorporating the computer
operations of Elliot Automation, English Electric, Leo Computers and Marconi). In the same period;
in France CSF/CGE and SEA of the Schneider Group merged to form CII. Second, increasingly
large programs of support were launched in the various countries. The French Plan Calculs (1967-71
and 1971-75), the British Advanced Computer Technology Project (1969), the German First and
Second Data Processing Programs (respectively in 1967-69 and 1969-70) channeled a large part of
public policy support to CII, ICL and Siemens, respectively (Malerba (1985)). Third, governments
protected national champions' markets through public procurement. For example, in 1969 ICL
received 94% of central governments orders for computers and benefited by the official government
statement to buy from British firms wherever reasonably possible(Torrisi (1995)).
There was even an attempt to increase the size of IBM competitors through a pan-European
company. Siemens, CII, and Philips formed the UNIDATA joint venture. Because of conflicts
among the partners and the struggle over the control of the joint venture, real cooperation never took
off and the joint venture ended in 1975.
23

In spite of their niche strategies and the national champions policies, market shares of European
firms remained quite low. For example, in France in 1972 IBM controlled 58% of the installed base,
while CII, Siemens and Philips claimed 12% and Honeywell and Bull 18%; in 1980 IBM still had 52%
of the installed base of computers (in France), while CII-Honeywell-Bull controlled 31%. In the
U.K., ICL's market share declined from 41% in 1968 to 31% in 1985
The effect of protection by individual European governments was to keep an uncompetitive
European computer industry alive and sheltered from being destroyed by IBM. These barriers to exit,
however, did not lead European firms to launch major policies and investments able to increase their
innovativeness and competitiveness, internationally.
3.6 Coevolution in Japan: catching up in the 1960s
Japanese firms, and government policy in that country, behaved very differently. In the early
1960s the Japanese computer industry also lagged behind IBM. In Japan, this situation led to
consortia. With help from government procurement policies and large organizations like the
telephone company NTT, firms met with considerable success selling computers within Japan. This
was not the case in the overseas markets. As a result the country continued to be a net importer of
computers. An example of one of the first consortia was the FONTAC project.33 Planned as an IBM
1401-killer, this 1962-1964 project was late to market, completion coinciding with the next
generation (System 360) announcement from IBM. Both government and private firms participated
in it.
The second half of the 1960s saw a concerted Japanese attempt to catch up in computer
technology. Technology initiatives were not pointed at the turbulent, unclear, and unstandardized
early computer market. Instead, the establishment of broadly compatible platforms in the System 360
had made the importance of scale, standardization, and compatibility clear. Japanese companies
technical initiative was coordinated by MITI in the Super High-Performance Computer Project.
The Project had the goal of building specifically Japanese technical capabilities in hardware and
software. With IBM patents (extracted as a condition of IBM presence in Japan) and a budget, the
Ministry of International Trade and Industry had the resources to encourage cooperation. Six
computer firms participated. Firms participated in order to achieve more rapid time-to-market, and
sought quickly-commercializable machines. MITI favored a higher technological standard

33

See Anchordoguy (1989) and Fransman (1995).

24

(Anchordoguy, 1989).
In addition to government sponsored R&D, other Japanese institutions supported this early
development. From 1968, NTT was active as a buyer of computers and as a coordinator of computer
company developments of systems, e.g. the DIPS-1 system. In particular, NTT supported computers
development of three major producers (Fujitsu, Hitachi, and NEC) which were also major
telecommunications equipment suppliers.34 Government subsidized low interest loans to rent
Japanese computers through the Japan Electronic Computer Corporation and encouraged many users
to select Japanese brand computers (Anchordoguy (1989)). Other users, with both national and
private commercial goals, were encouraged to make this selection as well.
The effects of these early initiatives were to build a very substantial technological capability within
some Japanese firms,

partially catching up to IBM. This, plus market advantage in the home

Japanese market, led to substantial import substitution. Japanese vendors share of the Japanese
computers market grew steadily over the 1960's, reaching almost 60% (Anchordoguy (1989)). After
IBMs introduction of System /370 this share, however, was substantially reduced to about 50%.
Market success of Japanese vendors was weaker for computer software than for hardware. The
creation of the Japan Software Company, a joint venture of the Industrial Bank of Japan, NEC,
Fujitsu, and Hitachi, had considerable government subsidy. The firm was not able to ship such key
products as the common language that would let applications run on any Japanese computer. It was
dissolved in 1972.
At the time of the worldwide transition from System/360 to System/370, Japan had built strong
but not yet worldwide-competitive computer hardware competencies. Incomplete protection from
external competition such as from IBM and no protection from domestic competition among Japanese
producers had spurred firms development efforts. Note that both MITI and NTT avoided
European-style national champion policies. Each worked with a number of companies which
protected only one firm and de facto created a barrier to exit in that single firm. Japanese policies,
on the contrary, worked with a number of companies supporting their cooperation in some
technologies, but keeping them in competition. The presence of both cooperation and competition
has been a major reason for the moderate success of Japanese policies compared to the failure of
European ones.

34

See Fransman (1990) on Japanese developments, especially NTTs role.

25

3.7 Japanese advantages consolidated in the 1970s

Overtime, changes in the computer market played to Japanese firms strengths. When IBM
unbundled operating system software from computer hardware. Japanese companiesability to build
strong hardware competencies increased in importance, and their software weakness became less
important (due to the opportunity to sell hardware alone).
During the IBM System/370 era, Japanese firms held a large share of the domestic market through
government, NTT, and Keiretsu customer connections. But they lagged in both hardware and
software technologies. In the late 1970's two VLSI (integrated circuit) joint-ventures were launched
by several Japanese computer firms, one sponsored by NTT, the other by MITI. These projects were
focused on the creation of a Japanese hardware engineering and manufacturing capability. The scope
was ambitious, including manufacturing equipment for integrated circuit devices, integrated circuits,
and computer hardware. Separate initiatives dealt with the problem of computer software. In this
period software initiatives met with little success. As a result major attempts to establish an
integrated computer systems company, with both hardware and key software, plus worldwide service,
sales and support systems, was largely unsuccessful as measured by world market standards.
The hardware initiatives, by contrast, brought some Japanese firms to the worldwide competitive
level. This was reflected in domestic and export markets. Domestically, Fujitsu passed IBM Japan
in sales in 1979; by 1982, Fujitsu had 22% of the market compared to IBM's 20%. NEC and Hitachi
were not far behind with 17% and 15%, respectively. By the early 1980's, Japan became a net
exporter of computer equipment. The exports were largely IBM-compatible mainframes.
As the 1970s ended, the Japanese were by far the most effective followers of IBM. Traditional
national competencies, such as relationship selling at home and efficient volume manufacturing for
export, were being well exploited.
The next round of consortia technology development efforts were more ambitious. The goal was
establishing a computer architecture independent of IBM.35 This would take Japanese competencies
into direct head-to-head competition with IBM, of the form seen above to be very difficult.
Accordingly, Japanese efforts involved an attempt not merely to catch up but to leapfrog IBM
technologically.
This effort played out in the Fifth Generation Computer Project and in the Supercomputer Project.
Again, these were multicompany government- or NTT-sponsored collaborative research efforts. The
35

The 1982 secret-stealing incident, in which Hitachi was caught attempting to buy IBM trade secrets from a
consultant, was often cited as proof of the importance of these efforts.

26

Supercomputer Project, begun in 1981 with nearly a decade-long time horizon, was a hardware
technology-capability project.

Involving nine large Japanese computer firms in a variety of

subprojects, it attempted to (inter alia) push forward integrated circuit technology in breakthrough
areas such as gallium arsenide instead of silicon. While that and other technology-capability goals
were clear, the exact design concept for a future supercomputer was left unresolved. Over the same
time period, the Fifth Generation Computer Project had more of a computer system development
flavor. Its goal was an artificial-intelligence based thinking machine.36
These collaborative research efforts have led to much discussion of "Japan, Inc." It is worth
stressing the breadth of Japanese mainframe computer companies activities in this period. While
these long term collaborative development projects were part of their activities, they were by far not
the most important part. Matching product capabilities with IBM, with Amdahl, and with one
another meant that there was plenty of ordinary technological competition. Once again, the
appropriate view of Japan, Inc. involves both some coordination by the government and
considerable independence and competition among firms.37
The Japanese leapfrogging efforts, and Japanese firms technology development efforts,
were once again of mixed success. Some of the longer-term technology initiatives, such as gallium
arsenide chips and artificial intelligence software, turned out to be far less fruitful directions than
anticipated. Yet the overall effort led to the development of more fully realized three pronged
investments--of a Japanese form--by the late 1980s. Many observers saw this development as a very
real threat to IBMs dominance. Others saw less of a threat.38
The Japanese initiatives never had a real market test, and so offer little hard evidence resolving
the debate about the effectiveness of public/private consortia at this more ambitious level. At the end

36

The SIGMA project, to build Japanese Unix computers and workstations, had a shorter payback span, equal
IBM-independence, but still depended on international standards.
37

Some observers doubt the efficacy of the central coordination . See Callon (1993) for the view that the
consortia largely pursued the firms' separate goals, rather than uniting them in a government-led way.
38

See Ferguson (1993) and Anchordoguy (1989) as well as Sobel (1986) for the view that it was a real threat.
Ferguson and Anchordoguy are particularly interested in the government policy issues. Since IBMs strengths were
integrative, they argue, could not a whole country coordinate the three pronged investments if it had the right policies?
But see Callon (1994) for the view that MITI did not coordinate. Also, see Fransman (1995) for the view that MITI did
coordinate but suffered from vision failure. While MITI has been quite effective in catching-up programs in
mainframes through cooperative programs, MITI has been unsuccessful in the perception of developments alternative to
mainframes (such as computer networks) or in the creation of radically new technologies.

27

of the 1980s, the Japanese efforts were coming to fruition. At the same time, the first industry
dynamics was coming to an end. The era of IBM dominance of large-systems computing did end,
but it was not the Japanese threat that ended it. Instead, it was competition from networked small
computers that attached the large mainframes. To see the origins of that threat, we now move
backward in time and examine the second, entrepreneurial, industry dynamics.
4. Industry Dynamics 2: Entrepreneurship and Entry
The second industry dynamics, that of the founding and evolution of new firms, new markets, and
new technological capabilities, went forward in parallel to the first. It saw, however, radically
different coevolutionary processes. We now return to the early 1960s to follow this very distinct
coevolution.
4.1 Minicomputers: the sources of American competitive advantage change
The introduction of the first real time interactive general purpose minicomputer-the PDP8 by
DEC in 1965- opened up new types of demand for computers in research laboratories and
manufacturing plants (the monitoring and control of industrial processes). In addition, computers
were used for technical problem solving activities, and focused on specific applications. Integrated
circuits, and not transistors were used from the beginning as basic semiconductor components.
The appropriate seller marketing model for small business and scientific minicomputers was
built on the fact that the relevant buyers were technically fluent.39 The features of computers that
mattered to buyers in these segments could be described quickly in objective, technical language.
Institutions for direct communication among buyers about products sprang up. These had a strong
engineering fraternity flavor, but played some of the same roles as the sales force in the commercial
segments. No extensive software support was provided by minicomputer producers, so new
intermediate actors emerged between the minicomputer producers and the customers: system houses
and value added retailers.
4.2 High entry rates and rapid firm growth in the American industry
Minicomputers had a major effect on the structure of the American industry. A large number
of new specialized minicomputer firms entered the industry. DEC, the largest of them (with about
39

This contrasts with the mainframe marketing model, with its extensive field sales forces, customer support
and service, and relationships with senior financial or operations executives in customer companies.

28

one-third of minicomputer sales over many years), was an entrepreneurial startup with roots at MITs
Lincoln Laboratory. Other new firms included CCC, Microdata, General Automation and Computer
Automation. Because of the importance of minicomputers in scientific instrumentation, many
instrument firms, including Hewlett-Packard, Varian, Perkin-Elmer, and Gould, entered the
minicomputer market.40
Existing computer firms entered minicomputers late and with mixed success: IBM had a
small mainframe marketing strategy, not much blessed in the marketplace, while Honeywell
(another instruments and controls company) did well. The last source of new minicomputer
companies were spinoffs. Data General was formed by entrepreneurs leaving DEC; Prime Computer
by executives from Honeywell; Tandem (which might be classified as either a minicomputer or
mainframe firm) by an HP marketing executive. Thus, in the early period (through roughly 1975) of
the minicomputer market the sources of entrants were different than they had been in the mainframe
market.
4.3 Limited entry and slow firm growth in Europe and Japan
In Europe, few new minicomputer firms entered the industry, and for several reasons. First,
American producers had a first mover advantage and rapidly entered the European market. Second,
similarly to American mainframe producers, established European mainframe producers did not move
or moved too late and unsuccessfully into minicomputers. Third, limited spin off from universities
took place. Fourth, lack of venture capital unpaired the financial support for new ventures. Fifth,
the protectionist measures used by the European governments for mainframes (such as public
procurement) could not be extended to a market formed by small and medium enterprises and
research laboratories.
As previously mentioned, during this period in Germany several producers thrived in the
market for mid range systems for specific applications. The case of Nixdorf is highly illustrative of
this phenomenon. In 1965, Nixdorf introduced the Universal Computer 820, a small computer with
a cash register based on semiconductors. Later, it developed several sector specific software
solutions which targeted small and medium size firms, mainly for banks and retailing (POS and cash
registers) and followed a strategy of direct distribution. For its 8870 family (1973), Nixdorf
developed the software Comet which proved to be a major success. During the 1970s, however, the
40

While all of these were entrepreneurial companies, they had been founded to make instruments, not

computers.

29

first weaknesses emerged: there was no integration or modularity among the various product lines,
internal technology development was limited and product quality was low.
Japanese efforts in the minicomputer business have been largely unsuccessful.

The

government sponsored SIGMA project of the 1980s, for example, which sought to develop a UNIXbased workstation platform for application software development. It did not reach market success.
5. Microcomputers: the sources of American competitive advantage persist
The availability of a new technology at the component level - the microprocessor-allowed
firms to develop products more user friendly and more decentralized in terms of computing capacity
and to satisfy the needs of new types of demand: family, hobby, educational uses and small business.
Personal computers were far less powerful but also far cheaper than the machines discussed so far.
This spread computing power through organizations (beyond centralized MIS) leading to the creation
of many new applications.
5.1 High entry rates and rapid firm growth in the American industry
In the United States, the market developed first on the basis of hobbyist demand, with
suppliers typically adopting the marketing model of minicomputer producers. In the late 1970s, there
were two main de facto standards for personal computers, CP/M and the Apple II. Although the
CP/M operating system was itself the proprietary product of Digital Research, it was available on
dozens of different brands of computers, most running the Intel 8080 or the competitive nearly
compatible Zilog Z80 microprocessor. In these circumstances, the market was quite open to entry.
Most CP/M computer firms were entrepreneurial startups. Even an English expatriate, Adam
Osborne, was able to found an American startup firm. Osborne computer's strategy of portability was
for some time very successful as the firm had a leading CP/M role. The Apple II system had a
proprietary architecture and operating system but, was like CP/M, an open one. Software developers
could rely on the Apple or CP/M environment to provide a stable platform for applications or utilities
development.

A large number of independent software companies came into being, again,

overwhelmingly entrepreneurial in origin.

Some software products were exceedingly successful and affected the overall demand for
microcomputers and the development of the marketplace. Of these, the most important example is
VISICALC, a spreadsheet program.

Developed by entrepreneurs when they were students,

VISICALC provided a strong motivation for accounting or similar number-crunching workers to

30

acquire microcomputers for direct use in work. Market forces transformed the hobbyist personal
computer into the business personal computer.
Entrants resembled those in minicomputers. They consisted largely of established electronics
(but not computer) firms, and de novo entrants. They may be divided into specialized computer firms
(such as Apple, Commodore, Tandy and Compaq) and clones.
5.2 The reaction of IBM
Established mainframe and minicomputer producers had a demand perception lag compared
to new personal computer producers. When IBM decided to move in, it did through external linkages
with competent firms: Microsoft for operating system software and Intel for microprocessors. The
IBM PC was an advance over CP/M and the Apple II. More importantly, it was a product from longestablished IBM rather than an entrepreneurial startup. Other experienced electronics companies
entered in the early 1980's as well. DEC and HP made PCs that were IBM-compatible to varying
degrees; ultimately the power of the PC standard in the marketplace compelled them to be fully
compatible. AT&T also entered with an IBM PC work alike. These entry efforts by minicomputer
and telephone firms were not linked to any direct marketing connection between the firms' existing
product lines. Rather they were attempts to use general electronics design, manufacturing and
marketing capabilities in a growing new area. Mainframe sellers also entered the PC market place,
notably Wang, Burroughs, and Honeywell. Their machines can best be understood from a marketing
perspective: as a courtesy to existing large-computer customers, the firms offered the new small
computers as well.
5.3 Competing approaches in PCs and the emergence of the PC platform
There were other initiatives technically comparable to the IBM PC in the sense that they used
16-bit rather than the previously standard 8-bit microprocessors. Apple, by then an established
company, introduced the Apple III. This system, which is much more PC-like than Macintosh-like,
gained considerable early enthusiasm in the marketplace. Apple had great difficulties, however,
building reliable Apple III systems. Another initiative also had elements of standards continuity:
CPM/86. The "86" label here means that the operating system ran on Intel 8086 microprocessor: it
was a 16-bit system with some multiprocessing capabilities. Other entrepreneurial startups had, for
some time, success in the 16-bit world: consider, for example, Cromemco (a startup named after a
college dormitory).
31

As extensively documented by Langlois (1990), the potential for many 16-bit initiatives were
swept away by the IBM PC standard. This product was both architecturally open and affiliated with
the IBM brand name. Only Apple's third effort at more advanced personal computing, the Macintosh,
(technically very different from Apple III, and much cheaper than the little-demanded Lisa system it
replaced) was successful in the marketplace.

Macintosh enjoyed very substantial product

differentiation advantages. Though it was designed as "the computer for the rest of us," that is, as
a machine that would have a broad mass market of unsophisticated users, the Macintosh succeeded
initially as a niche product for users (such as marketing departments running desktop publishing
software) who valued its graphical capabilities.
5.4 Firms strategies in the PC platform world
The shift from entrepreneurial companies supply to an IBM branded supply reflected the
changing nature of demand, and was widely bemoaned. The relevant chapter of Freiberger (1984)
calls this era "The Arrival of the Suits", that is, the replacement of technologists with businessmen
as the suppliers of personal computers.
IBM's decision to open the PC architecture traded off future competition for present speed
in reaching the market and the standard-setting benefits of openness. Future competition was going
to come from other manufacturers of PCs themselves. For a period, however, neither other brandname PCs from other electronics producers nor "clones" from startups were an effective source of
competition for IBM
a. The branded clone strategy
New successful strategies emerged during the second half of the 1980s. The branded clone
is one. After 1986 and the move of Compaq into branded competition, entrepreneurial startups began
to compete with IBM on a more effective basis: firms with strong technical bases such as ALR and
AST were, like Compaq, able to shift to having a brand presence. Entry with specifically marketing
or distribution advantages, such as that of Dell, was another route. Many overseas firms followed
this same path. ACER, for example, first built technical competence in Taiwan as a producer of
"clone" computers, often with other firms' brand names on them. Only later did ACER attempt to
build its own export brand name. The extent and pace of this increase in competition in the PC
segment was difficult to foresee at the outset, if only because the size and growth of the segment,
which drew much of the entry, were significantly larger than anticipated.
32

b. Competition from complementary technologies

Perhaps, a more important source of competition, and certainly unanticipated at the time IBM
chose the open architecture, came from makers of complementary PC components. In particular, the
technological leadership determining the direction of technical advance of the PC came to be divided
among
(1) makers of computers, of which IBM was the largest and most influential;
(2) Intel, maker of the microprocessors in the PCs,41
(3) Microsoft, maker of the operating system for PCs.42
Divided technical leadership reduced IBM's ability to steer the direction of PC technical
change in connection with broader strategic goals.43
Despite competition from the other technological leaders and from other PC computer
manufacturers that limited IBM's market power as a PC firm, the PC platform was an effective
competitor against other ways of supporting first "individual productivity applications and later
small business and small department multiperson applications. The market, not any individual firm,
determined most of the rate and direction of technical change in the platform. The market had on its
side great economies of scale, not only in the production of individual components, but also in the
(external) economies arising from network externalities (Langlois (1990)). In all computer markets
seen to date, success for a platform meant levering up suppliers' efforts through customers'
complementary investments. In personal computers, this took a new form. Instead of tight bilateral
marketing relationships between a platform vendor and the customers, external economies flew
through the compatibility of independent software vendors' (ISVs) products with a platform, and with
the pool of knowledge of how to use the platform and its software. Literally millions of users
collectively participate in these external economies, each having only a weak direct link to suppliers.
The aggregate strength of the leverage for the platform is at least as large here as in the mainframe
case.

41

Competition from licensed clones of Intel chips, such as AMD, was largely irrelevant to the determination of
the technical direction of the platform. Competition from almost-clones, such as the NEC V-series, was important in
some local markets but never as influential as the Z80 had been earlier on.
42

Again, competition from other terms which were licensed to make versions of the operating system, such as
IBM, was largely irrelevant in terms of Microsoft's role in setting technical direction.
43

As IBM's failure to establish MCA or OS/2 clearly show.

33

5.5 European industry: limited entry and survival only in niches

In personal computers the success of European firms has been limited for reasons similar to
the ones discussed for minicomputers. American microcomputer producers had a first mover
advantage and rapidly entered the European market, while established European mainframe producers
moved late and unsuccessfully into microcomputers. Moreover, outside the U.K. the lack of venture
capital and the low spin-off from universities limited the rate of European entry.
The entry of American microcomputer producers in Europe eliminated all the German firms
which had entered midrange systems during the 1960s and 1970s. Nixdorf is a case in point. The
introduction of personal computers substituted midrange systems in several applications. In addition,
Nixdorf's proprietary systems contrasted with open systems or IBM-compatible strategies of several
PC producers. Moreover, some of the weaknesses of Nixdorf (already discussed above) became
more relevant during the 1980s: a late attempt to enter the PC market, a presence in too many vertical
markets (cash-registers, banks, manufacturing, EDP-centers and communications) with too limited
internal technological competencies and high reliance on external technology, and a limited renewal
of the most successful products.
The main European mainframe producers, such as Siemens and ICL, entered the PC market
late and have not been highly successful. Siemens entered personal computer production in 1985,
while only in the 1987, ICL (after the unsuccessful development and production of CPM and CDOS
small and medium systems during the first part of the 1980s) decided to move to MS-DOS by
developing MS-DOS PCs in collaboration with ACER. Both Siemens and ICL undertook acquisition
processes in order to strengthen their competencies in PCs and in related fields (software and
communications). In 1989 Siemens bought Rolm from IBM, while in 1990 it acquired ailing Nixdorf.
In 1991, ICL-Fujitsu acquired Nokia Data System Division.
As in the minicomputer case, few new entries into the microcomputer market came from
European startups. The U.K. represents an exception (Torrisi (1995)). Here the presence of
Cambridge University, Cambridge Science Park, technical societies such as the Cambridge Computer
Club, high rates of spin off from established British computer producers and foreign-owned firms
produced relatively high entry of new firms (such as Amstrad, Acorn, Sinclair, Cambridge Computers,
Apricot and Psion) in the markets for personal computers, home computers, school computers and
palm-top computers.
Niche strategies have been followed by most new European entrants (Torrisi (1995)).
Amstrad represents a case of successful diversification from consumer electronics into low-price
34

products (Thomas (1990)). In 1984, Amstrad, a consumer electronics firm founded in 1968, entered
the personal computer business by introducing manufacturing, marketing, and sales techniques that
were popular in the consumer electronics business. Amstrad followed a very aggressive export
strategy and targeted the European market by offering low-price, advanced design systems with
standard components. Recently, Amstrad's strategy of low-price products with low R&D and lack
of a dealer network faced difficulties with the entries of clone producers and of high quality-low cost
producers such as Atari and Compaq. Psion is an example of innovative entry by a scientist.
Contrary to Amstrad, Psion was founded by a scientist from Imperial College in London. After
developing software games for home computers, the firm introduced the first hand-held computer-the
Organizer-in 1984. The firm grew internationally by specializing in hand-held portable computers and
software for the home, professional, and retail markets. Psion was able to develop advanced
competencies in design, engineering, logistic and manufacturing. It used direct marketing, sales
forces, and retail channels. Psion is now pursuing the integration between hand-held computers and
radio communications. Finally, Acorn represents a case of rapid identification of a growth area in
a protected niche market. Acorn was established in 1982 as a consequence of the launch of BBC's
new educational programs and the Department of Education's decision to start IT diffusion programs
in the national school system. After a period of success, the increasing competition in the home
computer market led to a period of crisis for Acorn, which was then acquired by Olivetti. Later on,
Acorn developed a new RISC microprocessor for personal computers.
Only one major office equipment producer successfully entered the standard PC market:
Olivetti. In the Olivetti case, however, this entry into PCs represented a re-entry into the computer
industry. After its exit from mainframes in the mid-1960s, Olivetti's electronic competencies did not
disappear completely with the sale of its Electronic Division to General Electric. In 1965, the small
group of electronic researchers who remained in Olivetti developed a desk-top electronic calculator
(Programma 101) for business and technical applications. The P101 was innovative because it had
a low price (1,000 dollars), was programmable, and did not require specialized personnel for its
working. The P101 did not use integrated circuits, but instead automatically assembled ad-hoc logic
components. The P101 had such a tremendous commercial success that Hewlett Packard paid
royalties to Olivetti for the development of its HP 9100, very similar to the P101. The P101 remained
an isolated case in the stream of Olivetti mechanical desk-top calculators of the 1960s. Actually, in
1968, a new advanced mechanical calculator, Logos 27, was introduced by Olivetti replacing the
P101 (Torrisi (1996)). Due to the collapse of the mechanical and electromechanical office machinery
35

market of the 1970s, Olivetti reentered in the computer industry in 1978. The advantage of not being
a mainframe producer (and of having some internal electronics competencies) produced a successful
entry into microcomputers. After having produced the first electronic typewriter (1978), Olivetti
moved to automatic teller machines and photocopiers, and finally to personal computers. In 1983,
it introduced the M24. In order to support and strengthen its entry into microcomputers during the
1980s, Olivetti followed a strategy of international alliances and opened up a wide range of
cooperative agreements. Two major ones with AT&T and DEC were later on terminated.
In this period, traditional European policies could not be effectively used in an attempt to
maintain unsuccessful producers in the market, as they were in mainframes. National champions
policies in fact proved quite ineffective as barriers to exit, because demand was controllable only to
a very limited extent by public procurement and domestic supply could not easily be sheltered by
protectionist measures.
5.6 Japanese industry: limited entry in a fragmented domestic market
The Japanese PC industry was characterized by Japanese firms' efforts focused on the localmarket in a time of worldwide standards. As a result, Japanese PC hardware exports were small and
PC software exports near zero. In 1992, PC domestic shipments were roughly two million units,
exports roughly a million. Many of the exports were based on manufacturing and design capabilities
for complex electronic artifacts, such as the Toshiba notebooks and laptops.
The Japanese PC market has yet (early 1995) to adopt the IBM/Microsoft/Intel PC standard.
A NEC almost-world-standard design has had the largest market share. In the early 1990s, NEC held
a very large share of the Japanese PC market (53% in 1992) (Dataquest in Fransman (1995), p.270).
The reasons for this success are several: NEC developed its PC in a decentralized way, kept its core
competencies in-house, used new marketing channels (such as Bit-Inns and Microcomputer Shops),
developed new advanced PC software applications, and pushed compatibility across products such
as NEC PC-8000 and NEC PC-9800. Being also a major telecommunication and semiconductor
producer, in the process of PC development NEC was favored by the in-house lack of a dominant
mainframe culture (see Fransman (1995) for a detailed description of NEC's involvement in PCs).
Based on almost-Intel microprocessors, and an almost-IBM-PC architecture, these computers neither
gained much connection to worldwide external economies nor obtained the production scale
economies of the standard PC. A strong competitor, possibly because it is more graphical and thus

36

more linguistically flexible, has been Apple Japan.44

Another very interesting contender for the Japanese market standard during the 1980s was
TRON. A collaborative effort, neither government nor NTT sponsored, TRON was intended to be
an all-Japanese microcomputer system. From a new microprocessor through to a Japanese-character
set (kanji) user interface, TRON would permit complete separation from the world market.
Language differences, and specific market conditions, mattered. For the evolution of the
Japanese industry until the 32-bit era, PCs could only accommodate Kanji in very clumsy ways. Also,
a separate software market did not develop in Japan. The domination of the mainframe market by
Fujitsu, Hitachi ,and NEC led to the development of customized software which then hampered the
development of standard package software. Even microcomputers were marketed in Japan the way
mainframes were in most other places: they came with bundled software, or specific application
software written for the individual user site.
Whatever the origins, the effects were to have a fragmented, specifically Japanese domestic
market (Cottrell (1994)). Software and hardware producers in Japan faced little competition from
international computers. But with a domestic market substantially smaller than the world market, the
Japanese industry was unable to participate in worldwide scale economies and substantial external
economies associated with microcomputers.
6. Computer systems and networks: the sources of American advantages change again
The 1990s saw three linked changes in computer industry structure and the workings of
competition. The process of vertical disintegration, which had been historically confined to making
each new market segment less integrated than the last, spread to all the segments. The locus of rent
generation shifted downstream to software and applications developments. Computer hardware itself
became more of a commodity.

Finally, the existing rents of mainframes and commercial

minicomputers came under intense competitive pressure from networks of smaller computers, in
client/server, and related configurations. This change is important enough to be treated as a third
industry dynamics.45

44

The Open Architecture Group, which includes Toshiba and Fujitsu, has been attempting to establish the
worldwide PC standard in Japanese.
45

See Bresnahan and Greenstein (1992) for the causes of the new dynamic.

37

6.1 Open platforms connectivities and complementarities

During the 1980s, computing platforms with open interfaces, such as the IBM/Intel/MS-DOS
PC and Unix workstations, became important and took on their own industry dynamics. They allowed
interchange-ability, connectivity, and interoperability in hardware and system software. Local Area
Network (LAN) standards for connecting PCs also began to emerge; while less than fully open, they
were far more open than earlier vendor-specific networking interfaces. All this was based on the vast
scale of the PC market. Parallel movements in Unix/workstation computers and in Internet
networking standards took place in universities and scientific communities.
The existence of open interface standards and of specialized technology firms were mutually
reinforcing. Standard bundles allowed new software firms to compete in the system software market
or in the application software market (which could be packaged or custom) without producing
hardware. Entrants would offer whole new systems by fully developing and producing hardware and
software (hardware manufacturers), by integrating different parts of the system for specific
applications (systems integrators), or by offering specialized software for specific applications.
In a context of standard bundles and application software, the key dimension for successful
innovation became the ability to use complementarities by linking technology, users, and applications.
It must be noted that application solutions, custom software, and software services required a
knowledge base for innovation different from hardware production: an in- depth understanding of
end-uses of information systems in terms of horizontal applications (such as spreadsheet or word
processing), vertical applications (such as software for banks, transport, and so on), and specific users
applications (custom software). Therefore, in order to be successful in applications, firms had to
understand market needs, identify the relevant dimensions of specific market applications, target their
products to those dimensions, and interact with users on a continuous basis.
Growth in the new, client/server architectures changed the linkages between a successful
computer product and technology sellers. Now, computer systems development would involve at
least the choice of computers as clients and other computers as servers plus software and networking
technology to integrate them. Integration services, not unlike traditional systems development
services in mainframe use, would be needed as well. Unbundling of clients from servers meant that
market forces could have a stronger effect. The existence of de facto market standards in the
standalone small computer markets for microcomputers and workstations meant that most new
system developments would take advantage of the considerable technical advances in
Intel/IBM/Microsoft based microcomputers or in Apple computers (a smaller number of systems used
38

Unix-based clients). Similarly, the unbundled client meant that de facto standard software for
microcomputers could be part of large-systems development. The ease of use and familiarity of small
computer systems were developed in the market for individual productivity applications and now
were redeployed in the market for multiuser departmental or company-wide applications.
Unbundled from the client, and the related software, the large computer formerly defined the
computer system became a "server." Servers were rated by technological functionality in performing
a variety of "service" tasks: the "file server", "database server", and "application server". As a
consequence, the traditional marketing strengths of large-system vendors like IBM were devalued.
Close relationships with customers became of little use in selling a standard product.
Computer industry structure underwent, and is still undergoing at this writing, a dramatic
shift. The shift has been large enough for some participants and observers to talk of an old computer
industry in distinction from a new computer industry. The two are different along a wide variety
of dimensions: firm organization, industry structure, technical base and the key technologies for
defining the direction of the industry. Roughly speaking, success has shifted from a model with
vertically integrated firms closely linked to their customers selling integrated products as solutions
and called computer companies because they made computer systems. Success has shifted to a
model of specialized technology-selling firms, with customers (or their systems integrators,
consultants, or outsourcers) putting together their own systems from components, and computer
companies that do not make any hardware. These changes have been accomplished with a huge
turnover in the rankings of successful firms.
Consequently, a process of vertical disintegration has taken place in the industry. In addition
to vertically integrated producers in hardware, system software and application software, such as
Apple and IBM, specialized firms emerged. Given the complexity of the knowledge base required
for innovative activities and the heterogeneity of competencies, cooperation among firms and
networking strategies became quite widespread by providing complementary and specialized
expertises regarding computer hardware or basic operating software, features of specific applications,
characteristics of market niches or user requirements.
As yet, no universally accepted standards for networking computers into client/server
architectures have emerged. Firms with client standards as their competitive advantage (Microsoft),
others with strong server positions (Oracle, IBM), and yet others with networking competencies
(Novell, IBM again) compete to influence the de facto standards setting process. Once again, we see
entry of a wide variety of organizational forms with a wide variety of competencies. In pursuit of the
39

same rents we predict the shake out with little fear of future contradiction.
6.2 The United States: success from complementarities, knowledge externalities, and
variety in experimentation
Again, country advantages remained in the United States, but for reasons that differed
compared with mainframes, or mini and micro. As we have seen before, in mainframes one world
leader emerged. American advantages coincided with IBM advantages. In mini and microcomputers,
on the contrary, entry conditions and an environment conducive to rapid growth of new firms was
at the base of American advantages. For networked computing, the sources of American advantages
shifted again. Networked computer systems were highly complex and rich in opportunities in all their
various components and dimensions. No single firm could innovate in all parts and subsystems. Open
platforms and standard bundles permitted compatibility and connectivity among various artefacts
introduced by the specialized firms in the various layers and also system integration by some firms
or even by users. The new specialized entrants were of various types: spin-offs from established
computer firms funded by some venture capitalists (technological competence driven), science-based
firms established by University scientists and funded by some venture capitalists, new firms with
market or marketing competencies.
The presence of strong complementarities and local knowledge externalities gave major
international advantages to the United States or more precisely, to Silicon Valley, which had several
firms at the frontier in each market layer. Intense formal and informal communication and high
personnel mobility (together with the high entry and growth rates already present in the mini and
micro period), allowed firms located in the United States (particularly in Silicon Valley) to be exposed
early on to new experiments, knowledge, and technologies. These firms could rapidly take this new
knowledge into account in their new artefacts. Relatedly, they could feed their new developments to
the other producers in the same or in other layers and to system integrators. Therefore, positive
feedbacks and knowledge increasing returns among producers within and across vertical layers were
being created. These mutual positive feedbacks gave American firms major innovative advantages
over competitors located in other areas.
Some American established computer producers (such as IBM) had to reorganize themselves
into application oriented groups. Others had a less successful transition. Honeywell was acquired
by Bull and pulled out of the general purpose mainframe market. CDC experienced major difficulties.
Burroughs and Univac merged to become Unisys. The combined company, however, soon exited
40

the computer business, leaving behind the former marketing organization which became a systems
integration firm.
6.3 Europe: survival in niches
Europe did not have firms active at the frontier in several of the layers of the new computer
industry, as the United States did. As a consequence, local knowledge externalities and positive
feedbacks did not take place. Therefore, entry of European firms has been mainly related to niches
in system integration (Malerba (1992)) and custom software (Malerba-Torrisi (1996)). Cap Gemini
Sogeti has been the most successful firm in this realm. In hardware and system software standard
bundles as well as in software package applications, new as well as established American firms
continued to be market leaders. Among European producers, only SAP and Software AG have been
successful.
In order to survive in the new industry, large established European computer firms tied
themselves up with key microprocessor producers. Bull acquired Honeywell, teamed up with IBM
by using its RISC microprocessors and increased its commitment to UNIX. Siemens followed a
strategy of internal growth and acquisitions (Nixdorf) and joined IBM's Power PC initiative. Finally,
ICL, acquired by Fujitsu, increased its links with Sun (RISC architecture).
In addition, all the main European computer producers moved into vertical markets and
applications: Siemens-Nixdorf in banking, public institutions, universities, hospitals and infrastructure;
ICL in distribution. After an unsuccessful attempt to team up with DEC (Alpha processors), Olivetti
exited from the PC business in 1997 and focused on software and system integration in banking,
office information systems, and distribution. In addition, it entered forcefully in the mobile phones
market.
The internal market program of the EC, through the dismantling of trade barriers, the
harmonization of technical norms and the setting of European standards have attempted to create a
homogeneous market, therefore increasing the incentives for entry and growth of successful European
producers in niches not already occupied by American producers. Also, European R&D cooperative
policies such as Esprit and Eureka have proved somewhat successful in developing European
standards, fostering additional communication and interaction among domestic producers and teaming
up (private) resources for entry in new advanced and expensive technologies. As in the Japanese
case, the evidence on the effects of these European cooperative policies on the successful
development of new products (Malerba (1992)) is controversial.
41

6.4 Japan
Japanese computer firms found many of their earlier competencies greatly devalued by
computer systems and networks. Large systems development based on large Japanese-sourced
computers and with one-off software development for individual sites, however well done, was
vulnerable to competitive pressure from much more widely sold hardware and system software
platforms and more flexible development environments. A Japan-only PC standard began to look
much less attractive as worldwide PC standard bundles started to advance rapidly. While some firms
prospered in specific markets (Toshiba in laptop computers, Canon in printer engines, etc.), the
traditional giants had great difficulties. Mitsubishi exited the mainframe business. Fujitsu sought to
invest in multimedia, often through overseas subsidiaries, and became a Sun reseller. NEC saw
declining shares even in the Japanese PC market, and considered conforming itself to the worldwide
standard.
7. Conclusions
The analysis of the long term evolution of the computer industry in the three major advanced
areas (United States, Europe, and Japan) has highlighted several general points.
First, until the 1990's, competitive technical change was not a destroyer of competencies, but
demand opening and competence widening. Second, each industry dynamics had a separate
coevolutionary process and was characterized by a different model of the firm. Third, the United
States, Europe, and Japan exhibited different forms of supply, due to competitive, institutional and
other country-specific features. Fourth, despite separate coevolutionary processes, shifting sources
and locations of comparative advantages, American world leadership persisted. Finally public policy
differed according to the country and the market segment. All these points will be discussed more
in detail in the next pages.
7.1 Competitive technical change was not competence destroying, but demand opening
and competence widening
A look at the role of technological change in the overall evolution of the computer industry
shows that, until the recent competitive convergence of mainframes with networked computing,
competitive technical change has only rarely destroyed the competencies of the main established
leaders. In the early mainframe period, entry came mainly from established electronics and office
equipment producers. The founding of new demand segments, such as mini and micro, did not
42

destroy the existing capabilities of established mainframe producers. Rather, it was demand opening
and competence widening: it opened new demand segments with different types of customers and a
different user-producer relationship.
Later on, however, competitive convergence meant a convergence on the same type of
demand and competence destroying technical change. In fact, the emergence of competitive
networked computers using client/server architectures challenged the large-systems competencies of
established firms (for a more detailed discussion, see Bresnahan and Greenstein (1995)).
Within each segment, technological competencies were routinely destroyed by the
technological and market leader. IBM continually advanced mainframes, and DEC minis, in ways that
devalued not only specific old machines but the technical basis of whole product families. In the PC
market, Intel and Microsoft routinely made own strategic competence destroying investments, as did
Apple.
7.2 Each industry dynamic had a separate coevolutionary process
In each of the three coevolutionary processes, a new technology and demand have brought
entrants into the industry thus affecting industry structure. Entrants have in turn introduced
innovations, modifications, and changes to the original technology. By opening up a new demand
(new types of users), the new technology has created new user-producer relationships and has
affected firmscompetencies, strategies, and organization.
However, in each specific market segment a distinct coevolutionary process took place. In
mainframes, coevolution has been characterized by rapid technological change in favor of processing
power and data flow speed. Large systems required user-producer relationships, the centralized
organization of usersinformation systems and extensive sales and services efforts by large vendors.
Internal finance supported the activities of the large established firms. Market structure was
concentrated and suppliers were vertically integrated. A dominant design (IBM/360) emerged in the
growth phase of the segment and a market leader (IBM) dominated the industry early on, with a
coordinating role over the whole platform and an ability to steer the direction of technical change.
The role of universities was relevant as a seeding role only in the early period, and declined later on.
U.S. government policy played a role in early support for technological exploration and as a major
buyer of early computers. Then, as governments in other countries did, it later opposed the market
leader by antitrust policy or anti-IBM procurement.
In minicomputers and microcomputers, coevolution has been characterized by a type of
43

technological change which developed dedicated application systems in minicomputers or systems

with an increased ease of use and lower price-performance ratio in the case of microcomputers. The
relationship with customers required much less post-sales efforts, maintenance, and service, because
either the user was already technically advanced (minicomputers) or because the system was easy to
use (microcomputers). The structure of the market has been characterized by high entry early on and
then by increasing concentration in platforms both in minicomputers and microcomputers. A major
difference existed, however, between the mini and micro segments. Niche leaders emerged in specific
dedicated applications for minicomputers, while the main microcomputer platforms have been general
purpose with competitive supply of commodity hardware. Concentration emerged at the key
component level-operating system. Venture capital has played a major role in affecting firms' entry
and growth. In this coevolutionary process, the form of government policies (so important in
mainframes) did not play a major role, while more general policies favoring education and skill
development helped microcomputer diffusion.
Finally, in computer networks, a stable market structure has not yet emerged at this writing.
Connectivity and compatibility have led to modular, open, and multi-firms client/server platforms.
Technical change is following a variety of directions with an upsurge in the number of potential
technologies associated with the relevant platforms. In this situation, interdependencies and network
externalities have increased. The structure of the industry has thus been characterized by highly
heterogeneous firms in terms of size and specialization, active in various platform components, and
connected by standard interfaces to firms in other segments. Firms have a wide variety of
mechanisms for commercializing their products. There is widespread speculation that one or a few
of the firms controlling key interfaces for connecting modular products will come to dominate
networked computing, but no single firm has so far been able to govern change and coordinate
platform standards. Public policy again plays a role limited to infrastructure and skills.46
7.3 The appropriate model of the firm was different in each coevolution
In each coevolutionary process, a different appropriate model of the firm emerged, with its own
competencies, organization and strategy (Bresnahan-Greenstein (1995)). Mainframes were produced
and integrated by the same firm and used in a centralized organization by the MIS department. In this
context, a Chandlerian integrated firm became quite successful. This firm was active in the

46

Antitrust actions against potential dominant firm Microsoft have been toothless.

44

development, manufacturing, marketing and distribution of large systems, and produced some of the
components in-house. Market success was related to major and continuous R&D efforts and the
three pronged investments in management, production, and marketing. User-producer relationships
were relevant in establishing competitive advantages, because computer firms supplied systems that
solved usersproblems, had close interactions with the MIS department of large users, and offered
assistance and post-sales services. Large integrated firms controlled and coordinated system
development, even in the presence of modularity, because they could control key interfaces.
Compatibility across products and over subsequent product families allowed the persistence of
existing standards and lock-in of the existing customer base. In minicomputers, firms spent less on
sales, marketing, and support. Systems were characterized by simple programmable processors and
were used for specific tasks, while end users (engineers, scientists, technicians) were technically
sophisticated, in most cases developing their own applications. In addition, there was no need to
develop compatibility across systems for different uses (much less between minicomputers and
mainframes.) In microcomputers, firms specialized in components which were part of the platform
competed with other specialized firms and did not control buyersacceptance of the platform nor of
the standard. In fact, microcomputer platforms involved several disintegrated firms developing parts
of the platform connected by interface standards.

Even the IBM PC platform became the

IBM/Microsoft/Intel platform, in which innovation was decentralized. Control over the direction of
technical progress of the platform by a single firm became very difficult (even by IBM, the sponsoring
firm (Bresnahan-Greenstein (1995)). Distribution took place through retail outlets and other
decentralized distribution channels.
With divided technical leadership, the potential competition among component suppliers
reinforces existing standards. Flexibility in design is limited because unilateral changes by specific
component vendors have to be compatible with the standard. Changing component standards is quite
difficult because it requires the coordination of several firms. These features of microcomputers
continued in the computer network period, where modularity and connectedness have increased
local developments and local feedbacks. This favored the vertically disintegrated firm active in
components and parts of the platform, and contributed to the slow emergence of a dominant platform
for client/server computing in the present.
As a consequence, the decline of the centralized vertically integrated large firm in the 1990s is not
due to a decline in competencies. Rather, it is due to changing market conditions, from a single firm
platform to a multi firm platform, with no single firm able to coordinate efforts and compatibility
45

within the platform (Bresnahan-Greenstein (1995)).

Within the evolution of each market segment, there has been a partial convergence to a single firm
organizational model, punctuated by the introduction of competing organizational models which
slowed convergence. For example, in mainframes after the rise of IBM to world leadership, Amdahl,
Hitachi, and Fujitsu developed a successful imitation strategy. In PCs, after Apple and IBM became
industry leaders, there were both clones and entrepreneurial branded PC firms, such as Compaq.
Please note also that these new strategies were successful because they matched firmscapabilities
with demand, technological and competitive conditions. In some cases, similar strategies were
introduced by other firms but came "too early." In mainframes, RCA attempted earlier the same
strategy as Amdahl but failed. HP and AT&T anticipated Compaq, but did not have great success.
7.4 Countries exhibited different forms of supply
During each industry dynamic in each major geographical area, the form of supply was greatly
affected by competitive, institutional, and other country specific features. Contrary to the United
States, evolution of the Europe on mainframe computer industry has been highly influenced by public
policy protecting national champions, therefore limiting the exit of unsuccessful producers and
perpetuating a concentrated market structure characterized by a national champion and by IBM. In
Europe, mini-micro computer industrial evolution has been characterized by limited entry of new
producers and lack of venture capital, with American entrants dominating the European market and
a few established mainframe producers competing unsuccessfully with them.

Finally, in the

convergence of mainframes and client/servers, the lack of advanced competitiveness (together with
limited entry) in most of computer components impeded the workings of local network externalities
and interdependencies, thus generating a market structure again dominated by American firms. In
Japan, the evolution of the computer industry has always been characterized by a market structure
in which few large vertically integrated Japanese computer firms have entered and dominated each
of the new segments, and by a public policy that has supported some competition among Japanese
producers together with research cooperative efforts at the technological frontier.
7.5 American world leadership persisted despite shifting sources and locations of
comparative advantages.
The striking feature of the computer industry is that despite of technological discontinuities
and three different industrial dynamic, countriesinternational advantages and disadvantages persisted
46

over long time. Over four decades, the United States has always been at the technological frontier
and the world commercial leader. However, the United States has shifted its competitive advantages
over time, while other countries were not able to match these advantages. What is at the base of this
persistent technological and competitive leadership?
Some factors favoring American competitiveness persisted over time. First, the large size and
rapid growth of the American market has been unmatched elsewhere. Rapid growth is related to
rapid diffusion of new types of computers in the appropriate population of adopters. This is also
related to education in computer technologies and a highly skilled labor force in information
technology. Second, venture capital facilitated the entry of new innovative firms and the funding of
a variety of new initiatives in the United States, in mini, micro as well as in workstations and
computer networks. Finally, U.S. universities have always been a source of entrepreneurship and
have been highly receptive to the launching of new scientific fields and academic curricula.
Other sources of American competitive advantages have been changing over time. In
mainframes, the major sources of American advantages were linked to a single firms advantages:
IBM presented a unique commitment to R&D policies and to the Chandlerian three pronged
investments in management, production, and marketing. No other firm in the world has been able to
match IBMs capabilities and investments. In mini and micro computers, U.S. advantages were
related to the favorable entry and growth conditions for new firms in new market segments and in the
creation of open multi-firm platforms that created local knowledge externalities. In computer
networks, U.S. advantages were related to the presence of local knowledge externalities and strong
complementarities between various components of the open multi-firm standard platform.
Some of these advantages were transmitted from segment to segment. For example, the
success of venture capital in supporting minicomputers (as well as microelectronics ventures) led to
the availability of abundant venture capital in microcomputers and computer networks. Moreover,
some of the entrepreneurs spurring entry in microcomputers and later on in computer networks came
from established firms active in minicomputers first and microcomputers later.
The geographic location of the competencies supporting American success has several times
shifted within that large country. In mainframes, American advantages were related to the areas of
IBM location of R&D and production, centered in New York but widely dispersed.

For

minicomputers, the sources of competitive advantages were mainly centered with the Eastern part
of the United States, with important exceptions like western entrant Hewlett Packard.

In

microcomputing and even more so in computer networks, we have seen a regional shift from areas
47

in the Eastern part of the United States (such as Route 128) to the Central and Western part of the
United States (such as Silicon Valley). This implies the need to carefully consider the unit of analysis
of competitive advantages: the division/department, the firm, the region, or the country (See Saxenian
(1994) for more detail).
7.6 Universities played specific roles during the history of the industry
Universities played two roles in the computer industry. First, very early in the industry, they
were generators of scientific knowledge and prototypes in several advanced countries. At the
beginning of the industry, universities were the locus of the first big university-led projects and
relevant sources of scientific as well as technological knowledge. Second, in the United States they
played a role as sources of scientific knowledge and entrepreneurship in minicomputers and
microcomputers. The role of MIT for innovation and entrepreneurship in minicomputers, and
Stanford and the University of Texas for microcomputers and workstations has been relevant. Only
Cambridge in the UK has played a similar (albeit more reduced) role in Europe.
7.7 Public policy differed according to the country and the market segment
In mainframes, public policy has been of a top-down, mission-oriented type. It has been quite
different in the United States, Europe, and Japan. Early American military policies (and to a lesser
extent the UK) have been successful in the support for early exploration and in opening windows to
different technological alternatives. In addition, nonmilitary procurement fostered competition
through procurement from multiple sources. As we have documented earlier, the American military
and government pursued goals driven by military-government needs, but also helped the technological
and commercial development of the industry. As we have shown in this paper, however, these
policies were not at the base of the success of IBM. In Europe, there has been a major involvement
of the various governments for the support for national champions in an attempt to create strong
competitors to IBM. These policies (research subsidies and public procurement) were not successful,
because they did not foster competition in the domestic market and were protective of a single
(laggard) firm. Moreover at the time of their launch, IBM was already in a dominant position in the
various European countries and the national champions had already accumulated technological and
commercial lags. De facto these policies created a barrier to exit for unsuccessful producers. In
Japan, on the other hand, public policy has been successful in the catching up process with IBM,
because (contrary to the European experience) it nurtured multiple competitors, coordinated imitation
of IBM through coerced licensing, and sponsored collaborative research. As we have seen, the failure
48

of Japanese policy has been related to a market shift, not to the policy as such.
Please note that antitrust policy played two different roles. First, it played an anti-IBM role
both in the United States and in Europe. In the United States, IBM had to unbundle mainframe
hardware from software and was forced to handle Amdahl and the PCMs more gently. In Europe,
antitrust policies have been very attentive to IBM as well. Second, antitrust policy had a competition
increasing role only in the United States. In fact, in Europe it was highly tolerant of the domestic
national champions
On the contrary, in microcomputers and computers networks, public policy has been focused
mainly on infrastructure, education, and standard. Direct or indirect support for the creation of
favorable conditions, such as an advanced infrastructure, the creation of skills, and so on, has proved
quite successful in enlarging the size and fostering the growth of the market, increasing
communication and interaction and assisting entrepreneurship. (Malerba (1992))
Two final comments on policy refer to the dynamic setting in which these policies are
launched and take place. First, the discussion of the computer industry clearly shows us that public
policy has to be "adapted" and tuned to the specific stage and market segment that has been selected
for intervention. For example, apart from their relative success or failures in various countries
government policies for mainframes could not be used for minis, microcomputers, or computer
networks. Second, imitation policies aim at a moving target. Think of catching up policies. In
situations of rapid technical change, catching up policies should not focus only on the established
world leader and the winning technology. The policy target itself may be displaced by a new world
leader or by a shift in the relevant technology or market. This is the case of the successful catching
up by Japan in mainframes during the 1980s, which, however, was confronted by a major shift in
technology and market at the moment of the catching up. In this sense, public policies should be
flexible and sensitive, and keep open windows on a wide range of technologies and market
developments.

49

Bibliography
Anchordoguy, Marie (1989), Computers Inc.: Japans Challenge to IBM, Harvard University:
Harvard University Press, Cambridge (Mass) and London.
Bresnahan, Timothy and Shane Greenstein (1995), Technological Competition and the Structure
of the Computer Industry, CEPR.
Bresnahan, Timothy and Shane Greenstein (1995), The Competitive Crash in Large-Scale
Commercial Computing, in Growth & Development: The Economics of the 21st Century, (ed) Ralph
Landau, Nathan Rosenberg and Timothy Taylor, Stanford University Press.
Brock, Gerald W. (1975), The U.S. Computer Industry: A Study in Market Power, Cambridge
Mass., Ballinger Publishing Co.
Callon, Scott (1994), Divided Sun: MITI and the Breakdown of Japanese High-Tech Industry
Policy ,Stanford University Press.
Campbell-Kelly, Martin (1989), ICL: A Business and Technical History, New York.
Chandler, Alfred (1977), The Visible Hand: The Managerial Revolution in American Business,
Cambridge, Mass: Belknap Press.
Cortada, James (1993), Before the Computer: IBM NCR, Burroughs, and Remington Rand and
the Industry They Created, 1965-1956, Princeton, NJ, Princeton University Press.
Cottrell, Tom (1994), Fragmented Standards and the Development of Japans Microcomputer
Software Industry, Research Policy, v23, n2, pp 143-174.
Davidow, William H (1986), Marketing High Technology : An Insiders View, William Davidow,
New York, Free Press, London, Collier Macmillian.
Ferguson, Charles (1990), Computers and the Coming of the U.S. Keiretsu, Harvard Business
Review, V68, n4, pp 55-70.
Ferguson, Charles and Charles Morris (1993), Computer Wars: How the West Can Win in a PostIBM World, New York: Times Books, Random House.
Fisher, Franklin, James McKie, and Richard Mancke (1983), IBM and the U.S. Data Processing
Industry: An Economic History, New York, Praeger.
Flamm, Kenneth (1987), Targeting the Computer: Government Support and International
50

Competition, Washington D.C., The Brookings Institution.

Flamm, Kenneth (1988), Creating the Computer: Government, Industry and High Technology,
Washington D.C., The Brookings Institution.
Fransman, Martin (1990), The Market and Beyond: Cooperation and Competition in Information
Technology Development in the Japanese System, Cambridge (England); New York, Cambridge
University Press.
Fransman, Martin (1995), Japans Computer and Communication Industry, Oxford, Oxford
University Press.
Freiberger, Paul (1984), Fire in the Valley: The Making of the Personal Computer, Berkeley,
California, Osborne/McGraw-Hill.
Freeman, Christopher (1965), Research and Development in Electronic Capital Goods, National
Institute, Economic Review, 34, 40-97.
Gordon, Robert J. (1989), The Postwar Evolution of Computer Prices, in Technology and
Capital Formation, edited by Dale W. Jorgenson and Ralph Landau, MIT Press.
Hendry, John, (1990), Innovating for Failure: Government Policy and the Early British
ComputerIndustry, Cambridge, MA.
Katz, Barbara and Almarin Phillips (1982), The Computer Industry, in Richard R. Nelson, ed.,
Government and Technical Progress: A Cross-Industry Analysis (Pergamon Press), pp 162-232.
Jublin, J. and J. Quatrepoint, (1976) French Ordinateurs: de laffaire Bull *a lassassinat du plan
Calcul, Paris, Alain Moreau.
Kelly, T. (1987), The British Computer Industry. Crisis and Development, Croom Helm, London.
Langlois, Richard (1990), Creating External Capabilities: Innovation and Vertical Disintegration
in the Microcomputer Industry, Business and Economic History, volume 19, pp 93-102.
Langlois, Richard (1995), Cognition and Capabilities: Opportunities Seized and Missed in the
History of the Computer Industry in R. Garud et. al (ed.) Technological Entrepreneurship:
Oversights and Foresights , New York, Cambridge University Press.
Langlois, Richard and Edward Steinmueller, The Evolution of Competitive Advantage in the
Worldwide Semiconductor Industry, 1947-1996", this book.
51

Malerba, Franco (1985), The Semiconductor Business, University of Wisconsin Press, Madison.
Malerba, Franco (1992), The Organization of the Innovative Process, in Rosenberg, N.,
Landau, R., and Mowery, D. (ed) Technology and the Wealth of Nations, Stanford University Press,
Stanford.
Malerba, Franco and Torrisi Salvatore (1995), The Dynamics of Market Structure and
Innovation in the Western European Software Industry, in D. Mowery (ed) The International
Computer Software Industry: a Comparative Study of Industry Evolution and Structure, Oxford
University Press, Oxford.
Nelson, Richard (1992), U.S. Technological Leadership: Where Did It Come From and Where
Did It Go?, University of Michigan Press, Ann Arbor.
Nelson, Richard (1993), National Innovation Systems: A Comparative Analysis, New York,
Oxford University Press.
Norberg, Arthur (1993), New Engineering Companies and the Evolution of the United States
Computer Industry, Business and Economic History, v22, n1, pp 181-193.
OECD, (1966), Gap in Technology-Electronic Computers, Paris.
Pugh, Emerson (1995), Building IBM: Shaping an Industry and its Technology, Cambridge, Mass,
MIT Press.
Rosenberg, Nathan (1994), Uncertainty and Technological Change, Stanford University.
Saxenian, AnnaLee (1994), Regional Advantage: Culture and Competition in Silicon Valley and
Route 128, Cambridge, Mass. Harvard University Press.
Sobel, Robert (1986), IBM vs Japan: the Struggle for the Future, New York, Stein and Day.
Thomas, D. (1990), Alan Sugar. The Amstrad Story, Century London.
Torrisi, Salvatore (1995), The UK Computer Industry, Bocconi University, Milan, mimeo.
Torrisi, Salvatore (1996), Discontinuit e credibilit delle strategie di ingresso nellInformatica
in Bussolati C., Malerba F. e Torrisi S. (ed), Levoluzione delle industrie ad alta tecnologia in Italia,
Il Mulino, Bologna.
Usselman, Steven (1993), IBM and its Imitators: Organizational Capabilities and the Emergence
of the International Computer Industry, University of North Carolina, Charlotte.
52