WHO'S WINNING THE COMPUTER RACE The U.S. leads in processor design and software, but that may not be good enough. Of six pivotal technologies, Japan predominates in three. And Europe? An also-ran.
(FORTUNE Magazine) – IMAGINE A PROCESSOR more powerful than a mighty supercomputer of just a few years ago. It's your PC, telephone, fax machine, and VCR all rolled into one. You dictate to it, write on it, or type in ideas. It automatically figures out your information and entertainment needs, then tirelessly shops a global network of electronic libraries to serve up whatever might tickle your fancy. Alan Kay, a techno-visionary at Apple Computer, calls these devices ''intimate computers'' and says we'll see them within ten years. Can you imagine not wanting one? Now imagine that these fantastic machines will be made only by Japanese companies. When the millennium arrives, top technologists warn, the U.S. could find itself shut out of this dream market. So might Europe and the rest of the world. Says Frank Carrubba, director of Hewlett-Packard Laboratories and one of nearly a dozen leading technologists who conducted a formal review of American computer prowess last year: ''When we look down the road, we see the possibility of losing ground rapidly and completely.'' It would be an epochal loss. U.S. companies pioneered computing and still dominate it far and away. According to International Data, a Framingham, Massachusetts, research firm, the U.S. share of the $163 billion world market for computer hardware in 1990 was 60%, compared with 17% for Japan and 16% for Europe. In the $144 billion market for software and services, which is far more lucrative (average net margin: 18% vs. 10% for hardware), the U.S. lead is more commanding still: 70%. Yet Japan has gained ground dramatically in fields such as laptops. Its hardware manufacturers have picked up six percentage points of market share since 1986, three at the expense of the U.S. To find how the runners rank today in the global computer race, FORTUNE recently sought out four dozen American, Japanese, and European experts from computer companies, research firms, and university labs. A remarkable snapshot of some key trends emerged: -- Japan will inexorably erode American preeminence in computer hardware. Chalk it up to the familiar story of Japanese manufacturing prowess. The U.S. remains superior at computer theory, but ''the notion that we can save ourselves by doing only leading-edge stuff is wrong,'' says Ralph Gomory, a former research boss at IBM who now heads the Alfred P. Sloan Foundation. Japan is better at refining ideas in metal and silicon and turning out machines in huge quantities. As computers grow smaller, cheaper, and more desirable for everyone, knowledge of production processes and willingness to invest huge sums in miniaturization will determine the winners. For example, while liquid crystal displays (LCDs) are a U.S. invention, Japanese companies used them to make watches and small TVs; now they lead in LCD computer screens. Says Ian Ross, president of AT&T Bell Laboratories: ''The ability to compete in electronics is critically tied to the ability to be a player in high-volume electronics, including consumer electronics. I put this in the same class of challenges for the U.S. as the drug problem, the education problem, and the budget problem.'' -- Computers will interact with one another all the time. They will not only exchange data but also swap tasks, pool their memory, and even cooperate on gargantuan calculations that would defeat any computer working alone. The data communications industry will grow explosively as companies and countries race to build computer networks -- a contest in which the U.S., Japan, and Europe are fairly evenly matched. (In most other fields, Europe seems destined to remain an also-ran. Says David Nagel, vice president of Apple's advanced technology group: ''European computer companies are not terribly behind, but they don't seem to be ahead in anything. That is what we don't want to see in this country. Lost momentum is very hard to regain.'') -- Technology sharing and other kinds of partnerships will flourish. Since 1988, Motorola and Toshiba have been building memory chips jointly at a plant in Sendai, Japan. They split the $420 million tab. Such alliances raise a question on which the experts consulted by FORTUNE differ hotly: In an increasingly global industry, does it matter which nation develops or manufactures a computer device? Some think not. Toshi Doi, director of Sony's computer science research laboratory in Tokyo and a member of Sony's board, believes it is meaningless to assess technology country by country. Says Doi: ''Sony is not a Japanese company at all. We like to have contributions from every country and serve all the people of the world.'' Using that rationale, his company may be about to expand its computer business. The hot rumor in Silicon Valley is that Sony has formed an alliance with Apple aimed at producing so-called notebook machines -- forerunners of that ultimate, intimate computer. John Armstrong, IBM vice president for science and technology, thinks such internationalism has its virtues. Even so, he believes the failure of U.S. manufacturers to keep pace with Japan could upset the ''healthy balance of power'' among computer makers that helps stabilize the global computer market. Says he: ''The U.S. industry may not be investing enough to maintain an appropriate share in this balance.'' On this and the following pages, ten senior technologists from the U.S., Europe, and Japan rate regional prowess in six areas crucial to the industry's future.
PROCESSOR DESIGN -- The central processing unit in computers -- the part of the machine that does its ''thinking'' -- hasn't changed much since the 1960s, though advances in the past five years have made computers radically faster. A new generation of U.S. designers wants to press the accelerator even harder with processors that do many calculations at once. Machines from MIPS Computer Systems, a Silicon Valley maker of workstations, use a processing technique known as superpipelining. The workstation staggers its computations so that several can be under way simultaneously, at various stages of completion. Companies from IBM to a Campbell, California, startup called Hal Computer -- remember the villain in 2001: A Space Odyssey? -- are working on so-called superscalar machines in which a central processor will divvy up tasks among many subprocessors and coordinate their work. Other U.S. companies have pioneered ''massively parallel'' machines that use thousands of processor chips at once. Although ingenious ideas for new processors surface occasionally in England and other countries, European computer designs have never had much commercial impact, partly because that market has been so fragmented. Virtually no fundamental advances have come from Japan, according to George Lindamood, an analyst at Gartner Group in Stamford, Connecticut, who has studied Japanese computing for two decades. Most Japanese supercomputers and mainframes derive from U.S. designs, mainly IBM's; Japan's microprocessors are typically built under license from U.S. chipmakers. Japan may yet find its way into the business of microprocessors -- computers on a single chip. MIPS and its larger rival Sun Microsystems have allied themselves with Japanese chipmakers to build reduced instruction set computer (RISC) chips. Faster than conventional micros, they could break Intel's and Motorola's tight grip on the desktop market and involve the Japanese intimately in the computer design process.
SOFTWARE -- As computers gain power and the cost of computation drops, users will apply them to more tasks -- from designing products in three dimensions to unraveling the mysteries of DNA. Software that makes possible such work will become ''the key differentiator'' between rags and riches, says Steve Jobs, president of Next Computer.
Score another for the U.S. It's not unusual to see accountants in Japanese companies hunched over a PC using Lotus 1-2-3, the world's best-selling spreadsheet program. And on most computers made in Japan, from supercomputers to PCs, operating system software -- which governs the basic functions of the machine -- is either an American import or closely modeled on one. Japan's growing skill in hardware, though, could threaten U.S. hegemony. Says Armstrong of IBM: ''You don't ask a juggler which ball is his highest priority. Success is to do it all. You can't be a success in computing if you're a whiz at software and can't do hardware.'' Historically, Japanese software efforts have been handicapped by the prevailing view that writing software code is a low-prestige task. To become more competitive, most Japanese computer companies have started major software development programs. In Tokyo, Sony is working on what Doi calls a ''21st- century operating system, for an age when we will find millions of computers connected to a network.''
Sony has led Japan in adopting a revolutionary technique called object- oriented programming (OOP). It enables experts to break a large programming job into smaller modules that can be written piecemeal, with reasonable certainty that the whole thing will work when it is assembled later on. OOP could open the way to a long-sought Japanese goal -- software factories capable of turning out sophisticated programs. Japan's biggest successes so far have been in ''firmware,'' relatively simple programs embedded in consumer electronics goods like VCRs and computer printers to control their functions. These programs can be written in parts and lend themselves to factory-style production. While Europe has held its own in software and a number of important computer languages originated there, few experts expect innovative products. When Borland, a fast-growing maker of PC software, set up a development center outside Paris last year, CEO Philippe Kahn was unable to find a French programmer sufficiently talented to lead it. ''We had to export one of our American stars,'' says Kahn, himself a 1982 emigrant from France to the U.S.
DISPLAYS AND PRINTERS -- Call it eyeball economics: The more time people spend with their computers, the more they will expect -- and pay for -- sharp, clear displays. The ability to make screens could increasingly determine which manufacturers gain market share. Today, old-fashioned cathode-ray tubes -- CRTs, like the picture tube in your TV -- account for most of the computer-display business. But within ten years the market will shift to flat-panel liquid crystal displays, fancier versions of the display technology in digital watches. They take up less space, consume less power, and last longer than CRTs. RCA developed LCDs in the early 1960s but abandoned them as uneconomic. Now Japanese companies are best at mass-producing them -- a reason Toshiba and others in Japan have won a large share of the laptop computer market. The screens get their name from a layer of fluid sandwiched between plates of glass just four-hundredths of an inch thick. It contains carbon-based crystals whose orientation can be changed by an electric field to allow light to pass. The crystals can be adjusted at thousands of points on the screen called picture elements, or pixels, making it possible to form images and characters. Manufacturing such a device is a diabolically complex task. To control the 307,000 pixels on its latest 10-inch color flat-panel screen, for example, Sharp places three times that number of microscopic transistors on the glass. The process, akin to making computer chips, entails a complicated sequence of photolithography, chemical etching, and vapor coating. If even a few pixels are faulty, inspectors may reject the entire unit. Often, in the early stages of production, over half the displays must be scrapped. Japan is so far ahead in the $2.1-billion-a-year LCD flat-panel business that when IBM decided to enter, it formed a joint venture with Toshiba. The two companies recently opened a $200 million factory outside Tokyo to produce color laptop screens as much as 14 inches across. IBM says it has no intention of going it alone in LCDs. But many see the joint venture as a model for American companies hoping to gain knowledge -- and eventually capture market share -- from their Japanese rivals. Says Michael Borrus of the Berkeley Roundtable on the International Economy, a technology think tank: ''Clearly, IBM is going to end up less dependent on Japan. If push comes to shove, it will have the know-how to produce the displays itself.'' It was another Japanese-American collaboration, this time in computer printers, that spawned desktop publishing, a $1.2-billion-a-year business in North America. When IBM and Xerox introduced laser printers in the mid-1970s, the machines filled half a room. Designed to churn out high-quality pages for an entire office, they cost hundreds of thousands of dollars each. But in 1979, Canon developed laser technology similar to that now used in CD players and designed the innards for everyman's version of the machine -- a simple printer that produced only a few pages per minute but sold for $3,500. Hewlett-Packard and Apple Computer pounced on the devices, created software and circuitry to hook them to PCs, and pushed them to market. All three companies benefited mightily as desktop publishing took off -- Apple perhaps most of all, since that application helped its Macintoshes gain entree to the multibillion-dollar corporate PC market. Still, the specter of Japan haunts at least one widely respected Apple engineer. Says Gary Starkweather, a senior researcher there who invented the first laser printer at Xerox: ''The fundamental leverage in low-end printers is the ability to manufacture. Heaven help us if the Japanese ever learn to integrate them into computer systems themselves. What will be left for us?''
DATA NETWORKS -- As a computer user of the future, you will be neither lonely nor bored. Your friends and co-workers, along with millions of others, will be accessible via your terminal. If you choose, you will be able to see them while you chat, or send and receive written, voice, or video messages. Electronic conferences are likely to be so effective and satisfying that you'll take fewer business trips. If it's information you seek, almost everything that has been printed, recorded, videotaped, or filmed will be available on the network. Says Robert Lucky, who directs communications research at Bell Labs: ''Computers and communications can't be separated. What nature really intended you to have is high image quality on your screen and a wide choice of material delivered to it.'' Tapping such lodes will require communications networks far more sophisticated than telephone systems today. Putting a moving picture in high- definition color on a computer screen, for example, involves the transmission of roughly one billion bits of data, called a gigabit, per second -- more than 50,000 times the capacity of your home phone line. Building what Carrubba of Hewlett-Packard calls ''information highways'' to handle the data will be as crucial to progress in the 21st century as concrete highways have been in this one. Says he: ''Without highways, everybody would have sat around in pastures honking his horn but not knowing where to buy gas or how to go anywhere.'' The U.S. owes its slim lead in this race to sophisticated private networks cobbled together by big companies trying to get a jump on competitors. They use such networks to connect far-flung offices and marry themselves to suppliers and customers. Convinced that data networks are key to America's ability to compete, this year the Bush Administration asked Congress for $92 million to establish an experimental high-speed computer communications network. It will link supercomputers at IBM, MIT, and a score of other labs, enabling researchers to pool their processing power and tackle otherwise impossible tasks. Japan could soon close the gap. It has the largest market share in the lasers and other key components that transform electronic signals into bursts of light for transmission on fiber-optic lines (see Optoelectronics). Now it wants to leapfrog the U.S. in networking. Nippon Telegraph & Telephone (NTT), the national phone company, is negotiating contracts for a huge network that will provide high-capacity service, including video, to every corner of Japan. In Europe, France is most advanced at making data communication available to everyone. Every phone line in France can now carry conversations and data simultaneously. But many experts say this so-called integrated services digital network (ISDN) transmits data too slowly to give future computer users the speed they will need.
OPTOELECTRONICS -- At IBM's research center in Hawthorne, New York, four computers communicate purely by light. They are linked by a single optical fiber in an experimental network called Rainbow I. Ordinary fiber-optic networks rely on electronic switches to make sure messages reach the proper destination. But Rainbow sorts them by color, using filters that split each light pulse into four hues, each alloted to a single machine. Light can be splintered into so many wavelengths that similar networks may one day link tens of thousands of terminals. Optical fibers move information more efficiently than electrical wires. But since computers are electronic, their data must be converted into light pulses before they can be dispatched on an optical strand, and translated back again once they reach their destination. That work is done by tiny lasers and control chips from AT&T and Japanese companies like NEC and Hitachi. Not far off are double-duty chips that will combine electronic control circuits with lasers and microscopic channels for light. They will make it simpler to link computers and fiber-optic lines. The chips will also let engineers replace heavily trafficked circuits within computers with higher- capacity optical fibers. While the U.S. has multimillion-dollar optoelectronics projects at IBM and AT&T, the overall Japanese effort is bigger. It includes Fujitsu, Hitachi, NEC, and other companies. Worried that America might slip further behind, Congress recently appropriated $50 million for ''advanced precompetitive research'' including optoelectronics work. The designation allows companies to collaborate without fear of violating antitrust laws, much as Japanese companies have long cooperated under government guidance. Says Paul Green, head of IBM's Rainbow project: ''Money isn't the issue so much as the officially blessed collegiality. This way I can call up Bell Labs and share stuff.''
CHIPMAKING EQUIPMENT -- What a difference a decade makes. In 1980 all the major producers of materials and equipment used to manufacture semiconductors were American. By 1990 four of the top five companies in this critical $22 billion subindustry were Japanese. In the interim, Japan surged past the U.S. to become the world's No. 1 maker of computer chips and its No. 1 buyer of chipmaking equipment. Manufacturers of everything from toasters to elevators used the 1980s to beef up the performance of their products with microelectronics. In Japan, semiconductor companies built more chip plants to supply this burgeoning market and turned increasingly to Japanese companies to equip them. Says Dan Hutcheson, whose San Jose firm VLSI Research compiles data on chip equipment: ''The problem for the U.S. isn't so much that our technology is inferior as that there isn't the customer base to support a business.'' Losing the dominant position in any thriving industry is bad enough. But some experts think the U.S. decline in chipmaking equipment could presage the disappearance of the entire U.S. computer business. In this era of microminiaturization, as entire machines shrink onto chips, whoever has the finest chipmaking gear may eventually make the best computers. The experts fear that Japan will exploit U.S. chipmakers' dependence. In May the Department of Commerce accused Japanese chip equipment companies of discriminating against U.S. customers by selling their newest machines exclusively at home. The sole remaining U.S. company in the top five, Applied Materials of Santa Clara, California, specializes in chip etching and vapor deposition equipment. It stayed successful by shifting much of its R&D and marketing to Japan, where it operates a $110 million research center near Tokyo. Says CEO Jim Morgan: ''We've been able to maintain our technological edge partly through our close relations with our Japanese customers. They're pushing semiconductor advances faster than anybody else.'' Applied Materials does not compete in photolithography, which analyst Hutcheson calls ''the mother of all chipmaking technologies.'' Nikon and Canon dominate the $1.2 billion market for the machines, which use light to define chips' microscopic circuit patterns. Until the mid-1980s, U.S. companies renowned for superb lens designs led the field. But -- sound familiar? -- it was the ability to manufacture lenses better that enabled Japan to leap ahead. Nikon and Canon perfected that skill by turning out hundreds of thousands of top-quality cameras for consumers. U.S. equipment makers may reassert themselves. Sematech, an industry- government consortium in Austin, Texas, now pumps $200 million a year into chip fabrication research. Among other things, the money helps struggling U.S. manufacturers such as Hampshire Instruments of Rochester, New York. It is racing giants like IBM and AT&T to develop lithography devices that use X-rays rather than visible light to define unprecedentedly small circuits. The market share of American chipmakers -- and the future of the U.S. computing business -- could depend on such advances. The margin of victory may be a hundred times thinner than a human hair.
CHART: NOT AVAILABLE CREDIT: FORTUNE CHART CAPTION: FORTUNE SCORECARD: PROCESSOR DESIGN FORTUNE SCORECARD: SOFTWARE FORTUNE SCORECARD: DISPLAYS AND PRINTERS FORTUNE SCORECARD: DATA NETWORKS FORTUNE SCORECARD: OPTOELECTRONICS FORTUNE SCORECARD: CHIPMAKING EQUIPMENT