Radical advances in technology can dislodge established regions or nations from the top ranks of wealth and power. In the late 19th century, Britain famously lost its lead to Germany and the U.S. when the key sectors in the world economy shifted from steam power and textiles to electricity and chemicals. Whatever Britain had done right in the earlier era, after about 1870 it was no longer enough to keep the first industrial nation ahead of its newcomer rivals.

Something similar happened within the U.S. when the microprocessor was invented at Intel in 1971. The outcome of that basic breakthrough would be to strike down the established information technology (I.T.) giants of the American Northeast, in favor of younger companies in such western states as California, Texas, and Washington.

As American computing evolved from the mainframe and minicomputer to the PC and the Internet, the centers of design, strategy, and control that were initially combined at IBM's headquarters at Armonk, New York, scattered far and wide. The sequence of industry stages in FIGURE 11 is from researchers at Morgan Stanley, an investment bank. I have added characteristic home-regions to the mainframe, mini, and PC eras. These are New York State for mainframes, Boston's Route 128 for minicomputers, and the West generally for the PC era. The current stage, Internet-Enabled Systems, began about 1994. Its home-region remains an open question. See Figure 11.  

Our theme is that in the PC era the younger firms in the West revolutionized world computing and in so doing won back a leadership role that was rapidly shifting to Japan. Two quick comparisons help put this idea in perspective:

• The decade after 1987, the period of the unexpected U.S. resurgence vis-à-vis Japan, saw a reversal in the market valuations of America's leading I.T. firms. (TABLE 5.) The West's Intel and Microsoft leapfrogged the Northeast's IBM and DEC (Digital Equipment Corporation). (Norris, 1997.)
• More generally, by a recent ranking 9 of the world's top10 I.T. firms are American, and 8 of the 9 are from the three western states.

This case-study links the American comeback in information technology in the 1990s to the regional realignment that marked the PC era. The module unfolds as follows:

1. The coming Japanese conquest (ca. 1989)

2. The rise of the Wild West companies

3. The break-up of the old computer industry, 1985-1990

4. The U.S. comeback, 1989-1994

5. New companies in the Internet Era (1994-)

6. The location of the top 100 I.T. firms in 1997

7. Europe's potential in the net-centered era
 

As background, we need to recall how different the world looked a decade ago.

1. THE COMING JAPANESE CONQUEST (ca. 1989)

In 1989, Japan gave every indication of pulling away from its technological competitors. The Rising Sun seemed to herald not only a national victory but also an affirmation of the Ministry of Trade and Industry’s (MITI's) strategic intervention and of industrial policy generally. A glance at several specific I.T. sectors shows how comprehensive the victory was expected to be.

(1) Semiconductors. Japan had caught the U.S. in its output of semiconductors by 1986, and by 1988 and 1989 it was supplying over 50 percent of the world market. Despite a partial captive market (e.g., IBM producing its own chips for its own computers), "merchant" memory chips for sale in the open market had been largely taken over by Japan.

That left mainly microprocessors for the U.S.—but even this creative side of semiconductor chips was being bought up by Japanese firms. According to M.I.T.'s (the university, not the Japanese ministry) Made in America, "Without some dramatic realignment of the American merchant industry, its decline is likely not only to continue but to accelerate." (Michael Dertouzos et al., 1989, p. 261.)

(2) Computers. The shift from desktop microcomputers to portables seemed to signal a shift toward Japanese leadership. The flat screens in laptop and palmtop computers had liquid-crystal-diode (L.C.D.) displays, a Japanese strength. (This was also another example of a U.S. discovery—at R.C.A. in 1963—which only the Japanese had seen fit to commercialize, for use on digital watch faces and video games). Hence the evolution of the industry toward laptops was thought to help Japan. Charles H. Ferguson thus wrote, "Some say: 'Japan will make the commodities and the U.S. will profit from design, software, and marketing.' This is fantasy." (1990, p. 66.) His prescription: U.S. government-industry consortia along Japanese lines.

(3) Software. Even in software, the Fifth Generation project (artificial intelligence, or AI) Japan initiated in 1982 was still being touted as a locomotive coming through the tunnel. This was the accepted outlook despite Japan's language and other handicaps in software. If MITI could make it happen in VCR's, the prevailing view then intoned, why not software too?

(4) HDTV. In 1989 lobbyists for a U.S. high-definition television (HDTV) effort to counter Japan's were making major inroads within the Executive Branch of the federal government. They converted Robert Mosbacher, the Secretary of Commerce, and Craig Fields of the Pentagon's Defense Advanced Research Project Agency, DARPA (now ARPA), to the view that the U.S. was hopelessly behind Japan and could only catch up in this "critical" (i.e, to national security) technology with help from the government. While not central to I.T., HDTV was nonetheless feared in the U.S. as an advanced technology that would permanently guarantee Japan's supremacy across consumer electronics and home entertainment generally.

But a funny thing happened on the way to Japan’s inexorable conquest of the world’s I.T. sector. The conquest fell apart on all fronts: chips, boxes, software, television—and , for that matter, telecommunications as well. You name it: if it required creativity and a rapid response, Japan lost it. They lost it, as a rule, to U.S. companies headquartered in the Western states, in an arc from Texas to Seattle.

Who were these companies? Why did they spring up in the western half of the U.S.? How did they defeat Japan's bid for leadership in I.T., the world's premiere growth sector?

2. THE RISE OF THE WILD WEST COMPANIES

One way to answer these questions is to list a series of examples in which old-style companies in the Northeast (call them "managerial corporations") bungled opportunities to innovate. In the vacuum, younger and more innovative firms (call them "entrepreneurial corporations") took advantage of the figurative wide open spaces of the West to move the industrial system to its next stage of development.

This section is a narrative account of the regional realignment. (Other issues of interpretation are touched upon in Section 3, in connection with cluster theory.)

CASE 1: FAIRCHILD SPAWNS INTEL (1968)

In contrast to mainframes and minicomputers, personal computers are blown up from thumbnail-sized microprocessors. Silicon Valley started with transistors, moved on to memory and logic (or microprocessor) chips, and evolved into a complex producing the whole I.T. spectrum. Its origins as a semiconductor center would ultimately give the Valley a decisive advantage over Route 128.

In this sense, it can be said that Silicon Valley is "a place that was invented one afternoon in 1957 when Bob Noyce and seven other engineers quit en masse from Shockley Semiconductor" to found Fairchild Semiconductor. This was a division of the established Syosset, New York firm, Fairchild Camera and Instrument. (Robert X. Cringely, p. 36.) The path leads from New Jersey's Bell Labs to a moment in 1968 when Noyce and crew would again leave, this time from Fairchild.

Background: The Origins of Silicon Valley. A key technological moment in the Valley's development was William Shockley's arrival in 1955 from Bell Labs. Shockley had been a co-inventor of the transistor in 1947 for Bell Labs, which would later garner him a Nobel Prize. In 1955 Shockley returned from New Jersey to his home state to start a transistor company in Mountain View, near Stanford. (Bell Labs is now Lucent.)

He called it Shockley Semiconductor because the transistor could be switched on or off to register a 0 or 1 in binary code, depending on whether it was in a conductive or non-conductive mode. This "semiconductor" property is present in the minerals germanium and silicon. Years later, in 1971, a newsletter writer named Don C. Hoefler accordingly coined the term, "Silicon Valley." (Rogers and Larsen, 1984, pp. 25-26.)

Shockley moved west to Mountain View in part because it was his home ground and his mother still lived there. But business logic also favored the move. Two key components were already in place to create a seedbed for new enterprises. One was the Stanford Industrial Park launched in 1951 and followed in 1954 by the Stanford Research Park. The impetus was not economic development but the desire to make money from real estate the university owned yet (by the terms of Leland Stanford's gift) could not sell.

The second keystone was Hewlett-Packard, started by the two Stanford students on the eve of World War II to manufacture electronic oscillators, under the guidance of an electrical engineering professor studying negative feedback, Fred Terman. The two components had come together in 1954 when H-P took a lease in the Stanford Research Park and served as the anchor for subsequent tenants. (Rogers and Larsen, chapter 2.)

The Traitorous Eight. Shockley had barely started his semiconductor company when it foundered on a legendary spin-off, which would eventually beget Intel. It has been said that Silicon Valley is "a place that was invented one afternoon in 1957 when Bob Noyce and seven other engineers quit en masse from Shockley Semiconductor" to found Fairchild Semiconductor, as a division of the established Syosset, New York, firm Fairchild Camera and Instrument. (Cringely, 1993, p. 36.)

Fairchild's Traitorous Eight, (as Shockley saw them) share credit with Texas Instruments (TI) for inventing integrated circuits (ICs). Germanium ICs were designed by Jack Kilby at Texas Instruments (TI) in Dallas, but he lacked a method of layering transistors on a flat surface. Jean Hoerni, one of the Fairchild Eight, came up with a "planar" technique to embed rather than stack component layers.

Noyce carried the idea through to create complete circuit maps on a single silicon slice, clearing the way for photolithography (or "burning" the circuits into the slice) and thus for batch production. TI and Fairchild both announced the breakthrough in 1959. ICs came into production within two years, for use by the U.S. government at $100 apiece to miniaturize the future Apollo moon rocket's onboard computer (Palfreman and Swade, 1991, pp. 87-91).

Intel. A decade later, Noyce, Moore, and others jumped ship again to found Intel, a more egalitarian company than Fairchild's eastern owners would permit. As a minister's son from Iowa, Noyce did without dress codes, reserved parking places, closed offices, executive dining rooms, and the other status trappings of more hierarchical and bureaucratic mature U.S. corporations. The remote control thus foundered on the divergent philosophies of Syosset and Silicon Valley:

Noyce's brush with the Northeast’s resistance to change was repeated at Xerox PARC, this time over bringing new products to market. In 1970, the eastern copier firm, Xerox, founded Palo Alto Research Center (PARC) as a flat organization of some 50 creative researchers whose mission was to create "the architecture of information."

As PARC'S web site puts it, they responded "…by inventing personal distributed computing, graphical user interfaces, the first commercial mouse, bit-mapped displays, Ethernet, client/ server architecture, object-oriented programming, laser printing and many of the basic protocols of the Internet." Preoccupied with copiers, however, the New York-based Xerox failed to bring any of these potentially breakthrough technologies to market. That remained for such western firms as Hewlett-Packard, Apple, and Utah's Novell.

CASE 3: IBM AND DEC IGNORE THE COMPUTER-ON-A-CHIP

Noyce and his colleagues thus formed Intel in 1968, as a spin-off (like its competitor National Semiconductor and some 50 other companies) from Fairchild. Intel made its mark on the world in November 1971 when it announced a triple breakthrough: the microprocessor, dynamic random access memory (DRAM), and erasable programmable memory (EPROM) for software. (George Gilder, 1989, p. 101.) Here was the package to make personal computers a reality.

But the big computer companies of the Northeast were not interested: "IBM and DEC...decided there was no market. They could not imagine why anyone would need or want a small computer; if people wanted to use a computer, they could hook into...time-sharing systems." (Palfreman and Swade, 1991, p. 108.) Thus microprocessors languished, scorned by the mainframe and mini- establishments—and not pushed by Intel—for another three years.

What would it take to bring the new firepower into play? The answer came with the now legendary January 1975 issue of Popular Electronics, whose cover showed the MITS Altair kit for a home-made microcomputer based on an Intel 8080 processor chip. Inspired, Steve Wosniak devised the Apple 1 to impress the hobbyists at the Homebrew Computer Club in Palo Alto. When Steve Jobs entered the picture the result was the Apple II, which found a ready market.

Wosniak's hardware breakthrough was matched on the software side by the 19-year-old Seattle-ite, Bill Gates. Using a DEC PDP 10 minicomputer at Harvard to emulate the MITS Altair, Gates and his high-school friend from Seattle, Honeywell programmer Paul Allen, devised a modified version of Dartmouth's mainframe BASIC programming language. Moving to New Mexico to be near the MITS facility, they formed Microsoft to market MITS BASIC, their microcomputer version of the mainframe programming language. Over the next five years, Microsoft would then develop, market, and license other languages for microcomputers, reaching $2.5 million in sales and 25 employees by the end of 1979.

In other words, the four seminal figures in the PC industry after 1975 (when IBM in New York and DEC in Massachusetts saw no future in it) were barely 21 on average and hailed from the San Francisco Bay area and Seattle.

Microsoft—like Compaq in Houston, Dell in Austin, Texas Instruments in Dallas, and WordPerfect and Novell in Utah—is a reminder that the technological transformation of American computing ranged from Texas to Seattle. If Silicon Valley was the West's capital, it sometimes followed the lead of the provinces.

The next episode was played out not in the West at all, but in Florida. Yet the theme remains the same: new territory as a spur to innovation.

CASE 4: THE PC'S ROOTS IN BOCA RATON, SILICON VALLEY, AND SEATTLE

Microsoft's initial takeoff following the New Mexico start-up brought the company to IBM's attention. In the mid-1970s, IBM had actually introduced an expensive PC-like machine that drew little response from its corporate customers and was quickly abandoned. By 1980, as microcomputer sales by Apple, Radio Shack, Atari and Commodore generated over $1 billion, IBM decided to try again.

This time IBM's development team was placed far from Armonk, New York, headquarters in Boca Raton, Florida, with a one-year project deadline. The crash-program deadline, unprecedented at IBM, forced the PC project chief, Bill Lowe, to design a machine built from other people's components—another radical departure for IBM.

Enter Microsoft. Lowe's plan for IBM was initially just to buy Microsoft BASIC, a standard feature of existing microcomputers, and to run it over a CP/M operating system from Gary Kildall's Digital Research of Pacific Grove, California. But when negotiations with Kildall misfired (because he did not show up for the meeting in Pacific Grove), Lowe turned to Microsoft for the operating system as well. Gates replied that IBM should use a 16-bit microprocessor, the new Intel 8088 chip. But since Gates had no operating system for a 16-bit processor, Microsoft now had to come up with one.

Gates' solution was to spend about $50,000 to buy an existing 8088 operating system, QDOS ("Quick and Dirty Operating System") from Tim Paterson's Seattle Computer Products and to rename it MS-DOS. In August 1981 the IBM PC appeared on schedule, featuring MS-DOS (called "PC-DOS" by IBM), and Microsoft BASIC, with available Microsoft versions of FORTRAN, COBOL, and PASCAL.

The package was thus equal parts hardware from Boca Raton and Silicon Valley's Intel and system software from Seattle. The creative points of origin were far removed from Armonk, New York.

Such was the beginning of the IBM-Microsoft collaboration that ended in 1990 with a complete reversal of fortunes, symbolized by IBM's plummeting employment, from 395,000 in 1984 to 243,000 in 1994. Microsoft's standard-setting strategy succeeded to the point where its stock-market value, like Intel's, surpassed IBM's by 1993. (Not the least colorful aspect of the reversal is that IBM unloaded stock in Microsoft and Intel that, if retained, would have been worth $18 billion by 1996.)

In the meantime, it wasn’t just IBM who took a tumble in the 1980s. Something comparable was also happening along Boston's Route 128, where the big four minicomputer companies (Digital, Wang, Data General, and Prime) had entered the 1980s as giant-killers, Davids to IBM’s Goliath.

CASE 5: THE FALL OF THE ROUTE 128 MINICOMPUTER COMPLEX

The reindustrialization of New England from the early 1970s to the mid-1980s was an amazing story (one I sketched in an analysis published by the Federal Reserve Bank of Boston). The linked article, "The Role of Services and Manufacturing in New England's Economic Resurgence," is based on a simple technique known as shift-share analysis .

In the study I contended that (in contrast to New York City's comeback at about the same time), New England's resurgence was powered by manufacturing. The technique allowed a graphical portrayal of manufacturing's role, as initially lagging, then surging ahead on its own, then bringing other sectors along, then collapsing in the mid-1980s. It should be added that the services sector played a nonetheless crucial part in New England's reindustrialization, because two of the three key catalysts to the comeback were venture capital and higher education—service activities.

In any event, after about 1985 the two outwardly similar high-tech clusters, Boston’s Route 128 and California’s Silicon Valley, moved in opposite directions. Along Route 128, the "Massachusetts Miracle" (as touted by defeated presidential candidate Michael Dukakis) collapsed in a heap, wiping out tens of thousands of jobs across New England. But Silicon Valley kept on adding employment, despite California’s high taxes and housing prices.

A little-noted reason for this eclipse was management failure along Route 128. All the key players in the New England complex saw the handwriting on the wall in the early 1980s. The future of computing was the PC, not minicomputers, let alone mainframes. Yet not one of the successful and profitable companies (DEC, Wang, Data General, Prime), had the boldness to cannibalize their profitable minicomputer lines to shift to a PC strategy.

What could account for this collective failure of nerve? Technologically, Route 128's minicomputers were actually mainframe computers shrunk down, not microprocessors blown up, like the PC. For that reason it was much harder for the Route 128 companies to introduce new and uncertain personal computers. Putting it differently, the economies of scopefavoring Silicon Valley’s microprocessor-based complex were missing along Route 128. Facing the technology barrier, managers along Route 128 stayed too long with cash-cow, proprietary (or closed) systems in minicomputers.

The long-term outcome would be a default I.T. role for Route 128 as a software and now Internet specialist, a role MIT's presence more or less guarantees. But the immediate result was for hardware production to move west, to Silicon Valley and then to Texas.

CASE 6: HOW TEXAS BECAME THE PC STATE

Today Texas has the two leading PC producers in the world, Compaq and Dell. How did the Lone Star State become the PC State?

Texas Instruments. Compaq's provenance traces a fairly precise lineage of industrial evolution. In the 1930s engineers with a new instrumentation technology for seismographic oil exploration came from the Northeast to Dallas to found Geophysical Services. In 1951, the original firm gave way to Texas Instruments (TI). As we have seen, the technologies TI employed led naturally to semiconductor research and in 1959 to the co-discovery of the integrated circuit by Jack Kilby, a TI engineer. Military and space contracts from the federal government spurred the company's ascent to one of the top semiconductor manufacturers in the U.S. by the 1970s.

Compaq. In 1982 four TI engineers from the company's Houston facility broke away to form a spin-off. Their leader was Rod Canion, and the company was Compaq. The breakaway team patiently reverse-engineered the then new IBM PC, so that it could legally invent its own BIOS (or interface) chip to emulate the PC for 100-percent software compatibility. Their success created Compaq's breakthrough as the legitimate king of the PC clone-makers. Compaq rose from its inception to Fortune 500 status in only four years—a record Dell would itself later break.

What is the meaning of the TI-Compaq story? The link between resource endowments and innovative capacity. Historically, the development of technological strength in an American region can typically be traced to the region's resource base. (Perloff and Wingo, 1961.) A given resource endowment either generates or fails to spark a related set of resource-processing activities that in turn encourage the development of new skills and technologies. (Norton and Rees, 1979.) The link between iron and coal endowments and metalworking, via the machine tools industry, was how the Manufacturing Belt of the Northeast and Upper Midwest became the nation's seedbed for innovation in the century from 1850 to 1950. The 60-year path from oil exploration to Compaq's world leadership in PC production displays a similar logic.

Dell. In contrast, Dell's meteoric rise in the 1990s has no such precisely traceable lineage. Instead, Michael Dell's strategy has been to devise a new distribution system to "mass-customize" the PC to order and to get the product delivered in a matter of days through the mail. "Because Dell holds very little inventory, it takes advantage of lower component costs and is always selling a fresher product, which can command a higher profit margin." (Fisher, 1998.)

This comment by a journalist in August 1998 accompanies robust earnings announcements that show Dell moving into the position of No. 2 desktop computer seller in the U.S., i.e., moving ahead of IBM and Hewlett-Packard. (Compaq remains in first place.) It would be hard to find a better illustration of the triumph of the Texas PC producers over their rivals in other regions.

3. "THE BREAK-UP OF THE OLD COMPUTER INDUSTRY" (1985-1990)

To recap, Intel’s invention of the microprocessor in 1971 set the stage for the PC—which the Northeast’s computer firms then failed to develop. That task was left to newcomers, adolescent or 20-ish prodigies from California and Washington State. After several failures, IBM finally managed to emulate Apple’s success, but only by moving the PC project’s design far from Big Blue’s headquarters, to Boca Raton in Florida, and only by using components from Intel and Microsoft.

By the mid-1980s, as Japan moved into the I.T. passing lane, IBM summoned its PC management back to its Armonk headquarters, where the PC was smothered—partly by jealous competition from IBM’s mainframe managers! Meantime, the initial outsourcing to Intel and Microsoft meant that clones using the same components were now taking away larger and larger shares of the PC market. IBM was about to fall, and Japan was ready.

Moreover, Japan had by the mid-1980s seemingly wrested the semiconductor lead from Intel. Intel had lost money in 1983 and 1984 in the face of heightened Japanese competition in DRAM memory chips. Andrew Grove, Intel's Hungarian-refugee CEO has since said, "There is at least one point in the history of any company when you have to change dramatically to rise to the next performance level. Miss the moment and you start to decline." (Andrew Grove, 1993, p. 58.) At Intel the moment came in 1985. (FIGURE 12.) The company surrendered memory chips to Japan and turned solely to microprocessors (at the time, 286s).

What happened between Intel's company-saving decision and 1990 Grove describes as "The breakup of the old computer industry... [which] gave Intel its chance and made the mass-produced computer possible." The change can be described in terms of Grove's sketch of vertically integrated companies vs. new horizontal tiers differentiated by component. (FIGURE 13.) The old system had self-contained, relatively closed and proprietary systems a la Route 128 and IBM. "These vertically integrated companies would compete against [each other]...and buyers had to commit to the whole package of one manufacturer or another." (Grove, p. 57.)

OPEN SYSTEMS

By contrast, the new model of competition is based on open (i.e., published) technical standards and full compatibility between every component-maker's products and every other's. In FIGURE 13, for example, each horizontal line represents a product axis along which companies in a particular segment of the market (systems software, monitors, printers, software applications, etc.) compete. The products from each segment must be fully compatible with those on every other horizontal tier--or customers will not buy them. This new system Grove terms "industrial democracy," in the sense that "It resists central guidance. Nobody can tell anyone else what to do." (P. 57.) In contrast to the old regime, choices abound and competition drives prices down.

Consumers also benefit from an accelerated pace of technological change. In a demonstration of the "Arrow effect," IBM had notoriously restricted the pace of technological change with a view to maximizing its profits over time. Its mainframe installations were known for "golden screwdriver" techniques, in which a demand for more performance (at higher rental rates) would prompt a visit from an IBM technician who would insert a few lines of code into existing software and unlock new power in the machine. (James Carroll, 1993, p. 217.) Similar restrictions of hardware potential marked IBM's missteps with its PC-AT in the mid-1980s.

The new rules force the pace of technological change and translate lab potential into product. Grove’s analogy is skis. "Any ski boot works with any binding. Any binding fits any ski. That permits innovation to take place independently in boots, bindings, and skis." (Grove, p. 57.)

STANDARD-SETTING

But who makes the profits required for high levels of sustained RD? Charles Morris and Charles Ferguson contend that the key to profitability is to control the standards, protocols, and formats by which the different parts of an information system are linked. Put the other way around, we find a (perhaps belated) recognition that Japan is only human.

As Grove observes, "A leading-edge product requires leading-edge manufacturing capability, and you can't buy it." (Grove, pp. 57-58.) It requires massive investment, which requires massive profits, which come from competition via standard-setting.

That is the puzzle the successful Wild West firms solved in the 1990s. In turn, their ability to handle the pace of innovation given by Moore's Law while still maintaining continuity of standards created shock waves worldwide. It gave the U.S. a second wind as the race with Japan carried into the 1990s.

4. THE U.S. COMEBACK, 1989-1994

In every one of the four I.T. sectors sketched in the introduction to this case, what actually happened was more or less opposite what most people had expected in 1989.

(1) Semiconductors. Timelines show Japan taking the lead in 1985, pulling far ahead by 1989, then being overtaken by 1993. Not only has U.S. pressure from the high-markup microprocessor end of the chip spectrum hurt Japan. Korea has attacked from the commoditized memory-chip end, in a bid reminiscent of that country's success vis-a-vis Japan in steel and shipbuilding.

(2) Computers. Computer "boxes" have also displayed a surprising U.S. resilience since 1989. One indicator is the failure of Japanese microcomputers to make much of an inroad into the U.S. market. Following a jump from 9 to 13 percent between 1989 and 1990, Japan's U.S. share fell back to 6 percent in 1991. Commenting on this reversal, Steve Jobs observed in 1992 that "The United States computer manufacturers have re-invented themselves and are holding on to the most desirable market in the world." (Quoted in Markoff, 1992.) The result finds Japanese firms supplying U.S. computer makers with flat screens and memory chips, but struggling to sell the U.S. markets the actual computers.

(3) Software.As to software, point one is the demise of Japan's Fifth Generation project. After 10 years, MITI gave up the ghost in mid-1992. "The problem for Japan is that the computer industry shifted so rapidly that the technological path the Fifth Generation took—which seemed a wise choice in 1982—turned out to be at odds with the computer industry's direction by 1992." (Andrew Pollack, 1992.) The lack of interest in the software that resulted led MITI to give it away free, though few took them up on the offer.

Equally important is the triumph of Microsoft's Windows platform, an exercise in cumulative standard-setting that has given an edge to U.S. computer companies relative to, say, NEC or Toshiba, which were late to commit to the standard.

(4) HDTV. As with the Fifth Generation project's commitment to the wrong technological trajectory, so too with HDTV. U.S. companies have developed digital approaches that appear to have leapfrogged Japan's analog approach. Thus "...enlightened federal regulation, rapidly advancing digital technology and cooperation between competing organizations have combined to vault the late-starting United States into a clear lead in the race to develop practical high-definition television." (William J. Cook, 1992, p. 14.)

In 1994 the director general of broadcasting in Japan's Ministry of Posts and Telecommunications conceded as much. He created a furor by revealing that his ministry was contemplating withdrawing support for Japan's HDTV program. The announcement was taken to signal "The triumph of American-style HDTV, something almost unimaginable five years ago...." (Andrew Pollack, 1994.)

CAN INDUSTRIAL POLICY HURT COMPETITIVENESS?

These four distinct sectors suggest that industrial policy can retard change in a dynamic technological environment. Pollack comments that the HDTV episode is especially telling: "...Japan's plan for HDTV showed the drawbacks in this country's system of Government-backed cooperative industrial development. The system allows for great staying power and steady progress down a particular path, but does not adjust well when the technological road turns." (Emphasis added.)

By the mid-1990s, such second thoughts about MITI and the role between Japan’s bureaucrats and its giant firms had become widespread. The failure of the Fifth Generation software project and of the HDTV campaign, combined with Japan’s distant lag in its over-regulated telecommunications sector—all these stand in contrast to the diversity and dynamism of the U.S. technological landscape. What had seemed to work so well for Japan in the automotive and consumer-electronics sectors in the 1970s and 1980s looked strangely dated today.

LESSONS

We have considered six examples of tensions between the old computer industry of the U.S. Northeast and more entrepreneurial actors in the South and West. We also compared the dismal 1989 prospects and startling U.S. comebacks during the 1990s in computers, software, semiconductors (more specifically, microprocessors), and HDTV. The counterparts were surprising setbacks for Japanese efforts in each of the four components of Information Technology, as well as in the related sector of telecommunications.

The logical link between these two sets of events is what Andrew Grove termed "the breakup of the old computer industry" between about 1985 and 1990. The relentless drumbeat of 18-month product cycles for chips given by Moore’s (and Joy’s) Laws required quick responses by players throughout the I.T. sector. The technology’s momentum in effect required entrepreneurial agility. Agility’s nemesis is bureaucracy—which in the mainframe culture of IBM would slow decision-making to a standstill as Microsoft heated up the system-software design wars of the 1980s. In a parallel quest, Intel’s radical bet-the-company reinvention after 1985 wrested standard-setting leadership for microprocessors away not only from IBM but from Japan as well.

The new business model that took hold after 1985 spawned competition via open (published) systems, compatible components, and uniform technical standards across vendors. In addition, the characteristic PC firm was specialized in a particular slice of the sector:

The competitors that succeeded under the new rules were not only American, but from the West. "From a global perspective, this change in vendor business models led to an even more dominant U.S. competitive position. Most of the companies that mastered the horizontal model turned out to be American, usually from the western half of the country." (Moschella, p. xi, emphasis added.)

Without the regional realignment, the history of the U.S. computer sector would have remained the preserve of IBM and Route 128 (the aging upstarts). Japan would likely have taken outright leadership in the I.T. sector from the U.S. Its great electronics companies, notably Fujitsu and Hitachi, but also Toshiba and NEC, gave every indication in the 1970s of knowing how to catch and overtake Big Blue. Instead, that would fall to such standard-setters as Intel and Microsoft.

To be sure, some observers still viewed the U.S. resurgence as only temporary. Eamonn Fingleton, for example, wrote a 1995 book with the uncompromising title, Blindside: Why Japan Is Still on Track to Overtake the US by the Year 2000. But as 2000 approaches, the forecast seems a bit strained. Perhaps a more plausible comment on the state of the world’s I.T. sector today came from a Czech computer expert commenting on software in March 1996. "Americans are showing an unbelievable burst of creativity. By relying on sophisticated tools, Americans have shifted the competitive arena from sweat labor to imaginative design."

With the arrival of the Internet in 1994, the creativity factor would play an even larger role.

5. NEW COMPANIES IN THE INTERNET ERA (1994-)

 
"Put simply, the story of computer industry competition has been one of new waves of technology, led by new waves of vendors, rapidly overpowering much of the existing order. …[T]he network-centric era will result in market and supplier restructuring every bit as great of those of the PC revolution." (Moschella, 1997, pp. vi-vii.)

In a useful simplification, the Internet or network-centric era can be dated from 1994, the year the barriers finally came down to the creation of a "network of networks." The Defense Department's ARPANET had been around since 1969. By the mid-80s the National Science Foundation had helped it evolve into a university research network based on the Pentagon's software standard, Transmission Control Protocol/Internet Protocol (TCP/IP). In 1989 Tim Berners-Lee, a British scientist working at the physics research lab CERN in Switzerland, had devised the hyperlink system of document linkage and access—an example of which you are now reading. The problem remained, however, how to hook up and standardize the numerous proprietary networks (e.g., ATT and MCI) competing for corporate and consumer business.

The problem was effectively solved in 1993 by programmers at the University of Illinois (the source also of the widely used free e-mail program, Eudora). Headed by Marc Andreeson, they came up with a good graphical-user-interface (GUI) browser, MOSAIC. Then Andreeson decamped for Silicon Valley and helped launched Netscape Navigator for profit in 1994.

At this juncture Metcalfe's Law kicked in. To repeat: the costs of adding users to a network increase linearly, while the benefits expand quadratically. If a network's users increase in number from 99 to 100, for example, the costs to the network go up by the incremental cost per node, the same as if the number increased from two to three users. But the number of additional two-way connections go up by 99, vs. only 3 more when a third subscriber is added. The larger the network, the greater the value to existing users of new members.

The smaller, isolated proprietary networks of the 1980s had failed to break through to the threshold that was now accessible via the Internet and a Mosaic-class browser. After 1994, any such constraints would be rent asunder. The shockwaves are with us still, as the new communications links redefine every sector of the economy.

NEW FUNCTIONS, NEW COMPANIES

The Internet permitted a blending of computing, communications, and entertainment in the mid-90s that, like the PC before it, changed the rules of the I.T. game. As to the pattern of regional advantage, one indicator of the new regime is the appearance of new companies. Another is the re-making of existing ones.

To get a sense of these tendencies, we can turn to a recent list of the world's top I.T. firms, then look at specific firms and their locations.

The July 1997 PC Magazine list of the world's 100 "most influential" I.T. firms appears as TABLE 6. The criteria for the list are subjective, but plausible. Perhaps the main caveats are (1) the list is American, and biased to that extent, and (2) these are the top firms from the perspective of a PC magazine, not from the standpoint of mainframes, telecommunications, or biotechnology. The list may well be open to debate as to exact ranks of companies, and its makeup and rankings will change from one year to the next. For our purposes, however, it appears sufficiently reliable to serve as a roadmap for the new geography of I.T.

The impact of the Internet can be gauged by the fact that 15 of the most influential 100 I.T. firms in 1997 had not existed in 1989. (TABLE 7.) In addition to Netscape, these included such firms as PointCast , U.S. Robotics, DeLorme Mapping (Maine), Progressive Networks (Ohio), Yahoo! and Firefly Network (Massachusetts). Eight of the 15 new firms from the 1990s were located in California, three in the Northeast, two in the Midwest, one in Texas, and one in Kentucky.

In addition, a number of other companies on the list are labeled as telecommunications- or Internet-related. They include ATT, idealab!, MCI, CompuServe, SAP AG (Germany), Hayes, 3Com, Santa Cruz, Number Nine Visual Technologies, Quarterdeck, Creative Technology (Singapore), Cisco Systems, Macromedia, America Online, Bay Networks, Madge Networks (the Netherlands), Cabletron Systems, and Ascend Communications (recently acquired by Lucent).

It is important to recognize that every company on the list of 100 (like most companies regardless of industry) experiences the Internet as a revolutionary technology. Tables 6 and 7 are more specific. They include companies that either sprang into existence to take advantage of the Internet or that qualify as I.T. companies because they have expertise in communications or media.

Beyond these two sets of firms, of course, the firm that leads the list, Microsoft, did a drastic change of course after 1995 to try to catch up with and overtake Netscape in the browser market. Without going into the antitrust case now being heard, we should nevertheless touch upon one aspect of the Microsoft vs. Netscape-AOL-Sun Microsystems conflict that is now taking shape.

FROM ILLINOIS TO SILICON VALLEY TO VIRGINIA

Has the Internet had much impact on the pattern of regional specialization in I.T.? The events of late 1998 offer a new angle on this question, in that they reveal the inability of Silicon Valley companies to set the agenda for the Internet era.

Not only is PC production centered in Texas. Not only has Microsoft set the software standards for the world to follow. Now it turns out that the struggle for commercial leadership on the Internet will take place between a Seattle-area firm and one based in Virginia: America Online. That is the implication of AOL's $4 billion takeover of Netscape, as bolstered by the Valley's Sun Microsystems.

As a columnist for the San Jose Mercury News observes,

Now we are on the eve of legal (antitrust) and technological (Open-Systems software, exemplified by Linux) challenges that seem likely to destroy Microsoft's position as a standard-setting natural monopoly. Would it be too nostalgic to recognize the possibility that the PC and the Internet explosion benefited from the Windows standard that Microsoft created—and from the Wintel duopoly Microsoft and Intel shared?

6. THE LOCATION OF THE TOP 100 I.T. FIRMS IN 1997

One way to sum up the impact of the regional realignment of information technology is to say that for the moment, Seattle, Silicon Valley, Texas, and now Virginia make the rules, and the rest of the world adapts to them. That statement used to hold for IBM. Then, in the late 1980s, it looked to everyone as if Japan's great electronics companies would replicate earlier triumphs in home electronics and automobiles. But that did not happen. Once again the U.S. holds a clear lead in I.T. The difference is that the sector's dynamism comes not from a company with a dress code (IBM), but from a variegated spectrum of younger enterprises in the West.

As we said, Silicon Valley dominates the list numerically—but not strategically. TABLE 8 shows the distribution of the 81 American firms on the list among U.S. regions. Within the U.S., 54 of the 81 are in the West. Numerically, 44 of the 53 western entries are from California. In terms of ranks, Washington (whose only firm on the list, Microsoft, leads it) and Texas (with 5 entries, but two in the top 11), are also prominent. (The absence of Amazon.com, another Seattle-area firm, must be an artifact of the timing of publication of the list, in mid-1997.)

In keeping with the theme of the new firm—the entrepreneurial vs. the managerial corporation—the ages of the 100 firms become younger as we move west. The firms founded before 1960 are more likely to have a location in the Northeast or outside the U.S. In the Far East (as it were), among Japan's 10 entries, 8 were founded before World War II, and the average founding date is 1927. (The remaining elder statesman on the list is Philips Electronics of the Netherlands, founded in 1891.)

While they made the list, few among these mature firms could be said to thrive in the new game. The only two stars from among the19 are IBM (which has risen from the grave in a new incarnation) and Hewlett-Packard—which is also the sole California firm among those on the list founded before 1960. Many of the other entries on the vintage list are struggling. In particular, three of the four great Japanese electronics combines are losing money in 1998, an unprecedented sign of weakened competitive positions.

It is surprising how extensive the U.S. comeback in I.T. has been. Only 19 of the top 100 firms are from outside the U.S., and Toshiba, at number 10, is the highest ranking of them. Indeed, 43 of the top 50 are American. The role of U.S. firms is thus even more dominant than the 81 percent share suggests, since most of the 19 non-U.S. firms ranked below number 50. (TABLE 6.)

The nationalities of the 19 firms are mainly Asian, with Canada and Europe hosting three each. Japan accounts for 10 listings: Toshiba (number 10), Softbank (21), Canon (27), Sony (43), Seiko (51), Matsushita (57), Sharp (60), Fujitsu (61), Hitachi (62), and NEC (63). Canada has three: Corel (17), Matrox Graphics (66), and ATI Technologies (77). Europe has three (the Netherlands' Philips Electronics, the U.K.'s Madge Networks, and Germany's SAP), but none in the top 50. In Asia, Taiwan's Acer is ranked at 34, Singapore's Creative Technology at 40, and Korea's Samsung/AST Research at 55. Of all the companies mentioned, perhaps the only one that today strikes fear and envy in the U.S. is Germany's SAP ("systems analysis and program" development). (Deborah Claymon, 1998.)

7. EUROPE'S POTENTIAL IN THE NET-CENTERED ERA

How does this episode in the history of technology relate to earlier crises of national competitiveness? One way to interpret the issue is to view the U.S. as a nation of country-sized regions at different stages of economic development. In that light, the I.T. sector experienced an internal, regionally focused maturity crisis in the Northeast a la 19th-century Britain. (R. D. Norton, 1986.) The difference was that the newcomer companies, created by entrepreneurs in younger regions, were still American.

By the same token, one reason for the eclipse of Europe's I.T. sector seems to be the smaller role played by entrepreneurs, relative to mature firms. The result, as Lester Thurow (1998) observes, is that Europe dropped behind Japan and the U.S. in the world's growth industries:

When breakthrough technologies occur, it is very difficult for old large firms to lead. They have to cannibalize themselves to save themselves, and that is simply very difficult to do. If one looks at the 25 biggest firms (based upon stock market capitalization) in the United States in 1960 and again in 1997, six of America's twenty-five biggest firms either did not exist in 1960 or were very small. In contrast, in Europe all of the twenty- five biggest firms in 1997 were big in 1960. In the past four decades Europe has been able to grow no new big firms that could lead the world technologically.

One characteristic of the transition is the shift in what he terms supplier structure away from the current horizontal value chain toward a communications chain. Apace with this he sees a corresponding shift in supplier leadership from U.S. made components to national telecommunications carriers. In other words, he assumes that national governments will retain control over major telecommunications suppliers, preventing complete globalization in this sector. The upshot is a localization of the present unified global market in which competitive advantage is gained through sheer design or cost efficiency.

In an ingenious application of Michael Porter's diamond model of national competitiveness, Moschella assigns number grades (in the form of stars) to the U.S., Japan, and Europe in a variety of categories he deems important for the next few years. The detailed evaluations are listed in TABLE 9 (below).

The bottom line is a better outlook for Europe than for Japan. Summing over Porter's categories of (1) factor conditions, (2) related industries, (3) demand sophistication, and (3) domestic rivalry, Moschella computes aggregate ratings. The scorecard finds the U.S. with 4 1/2 stars (out of a maximum of five), Europe 3 1/2, and Japan 2 1/2. By this reading, however preliminary, we are about to turn the page to a new chapter in which Europe plays a larger part.

So much for the regional origins of the digital economy. Our next step is to interpret the geography of innovation and entrepreneurship from the standpoint of metropolitan areas, or clusters.