(FORTUNE Magazine) – WITH UNINTENDED irony, Japan's Science and Technology Agency conferred a prestigious award in May on Tomihiro Matsumura. Matsumura, 55, senior vice president for microprocessors at Nippon Electric Corp., was named one of the nation's 32 ''persons of scientific and technological merit'' for boosting microprocessor sales and masterminding NEC's newest series of micros (as the industry often calls these silicon chips that act as the brains of a computer). Yet all but a handful of the 11.5 million computers-on-a-chip that NEC sold in 1984 were made under license from U.S. companies. To get those licenses, the meritorious Matsumura handed U.S. competitors millions of dollars in fees and some of NEC's choicest chip designs and manufacturing technology. And a U.S. company claims that crucial parts of the two new NEC V series micros now available are based on its own chips. For all the triumphs of the Japanese in other segments of the semiconductor business -- they have 70% of the market for computer memory chips -- Japan's chipmakers hardly dominate the $600-million-a-year world market for microprocessors. They accounted for 30% of microprocessor sales last year, compared with 63% for U.S. companies and 7% for the Europeans, according to % Dataquest, a California-based market research firm. But even these numbers overstate the Japanese strength, because 99% of all microprocessors sold were designed by U.S. companies. The announcement of the V series last year served notice that NEC is fed up with American dominance. NEC, whose $1.8 billion in chip sales in 1984 made it Japan's biggest in the field, plans to seize a macro chunk of the market with micros of its own design and thus escape the costs, delays, and humiliations of having to license U.S. chips. Other Japanese companies are eager to do the same, but they're waiting to see whether NEC's revolt succeeds. NEC's chances, say industry observers, are fifty-fifty at best. The first V series microprocessors are just arriving on the market, and the ultimate verdict will come two or three years from now, when statisticians can tally shares of the market for 32-bit micros, the most advanced type of processors. Getting a slice of the microprocessor business is alluring to any big-time semiconductor company. Micros aren't the largest market segment; that distinction belongs to the $7-billion annual trade in computer memory chips. But, Matsumura explains, ''micros are used in so many different products and are sold so often with a bundle of associated chips that you can link 15% to 18% of the $28-billion demand for all semiconductors to the microprocessor.'' Besides, he adds, ''many chips that are separate today -- memories, for example -- will be fabricated tomorrow within the microprocessor itself. If we succeed in micros, we can absorb -- and lead -- much of the semiconductor industry.'' U.S. chipmakers see another compelling motive for Japan's interest in micros. ''The Japanese want to win in semiconductors, but they also want to run the computer industry,'' asserts Dave House, a vice president at Intel, which got about $141 million in revenues last year from micros, more than any other company. The key to the computer, House says, is the microprocessor. Shipments of microprocessor-based computers -- personal computers, engineering work stations, and such -- have grown so much that this year they may exceed shipments of minicomputers and mainframes. Soon powerful 32-bit microprocessors will take over much of the mainframe computer's work (FORTUNE, December 10). When it acquired 20% of Intel last year, IBM was ensuring its own future in the computer industry. NEC, Hitachi, and Fujitsu could follow suit by developing micros of their own design that are good enough to compete in the open market. Whether any newcomer can grab a healthy chunk of the market is unclear. The micro business is a kind of oligopoly, and the barriers to entry are towering. Intel's designs account for 60% of industry sales. Motorola's, which are especially popular for use in computers that process a lot of graphics, have about 30% of the business. (The remaining 10% is split between chips designed by National Semiconductor and by Zilog, an Exxon subsidiary.) Developing and bringing to market a new and different family of processors costs plenty: NEC has already spent an estimated $300 million to get the six members of its V series off the ground. But money alone won't buy market share, nor will manufacturing expertise, the Japanese forte. Breaking into micros calls more for software and marketing skills, areas where the Japanese are weak. ''Potential customers want to know which micro runs the most software,'' observes Edward Sack, president of Zilog, which has been struggling for several years to popularize its own family of microprocessors. ''But it's a chicken-and-egg problem because programmers don't flock to write for your machine unless you have lots of customers. The issue isn't technology, it's scrambling to get the largest users' club.'' Both Japanese and American chipmakers used to finesse the software problem by building ''compatible'' processors, which run the thousands of programs already written for established micros. Many so-called compatible chips, including some from NEC, were direct copies of the most popular microprocessors. But new laws and recent court cases have made compatibility harder to achieve. The U.S. Semiconductor Chip Protection Act of 1984, for instance, clarified the murky copyrights of semiconductor companies by allowing them to protect the layout of a chip's electrical circuits for ten years. Despite the heightened barriers to making compatible chips, NEC has pinned its hopes on micros that can run Intel-type programs. The two V series micros released so far -- the V20 and V30 -- have original circuit layouts, but according to Intel their built-in software for controlling basic chip functions, such as multiplying, is much the same as Intel's. That means the V20 and V30 can handle all the applications software -- word processing and spreadsheet programs, for example -- written for Intel's processors. Intel insists that its built-in control software constitutes a copyrighted program. Now the two companies are fighting it out in court (see box). If NEC loses, its future in microprocessors will be a lot dimmer. SOFTWARE ASIDE, Japanese chipmakers have much of what it takes to succeed in making microprocessors. They already manufacture a wide range of processors for use in their own products. For example, within NEC's Astra 200, a small business computer that sells for $75,000, lurks the company's own 32-bit microprocessor, a chip that can locate and manipulate 32 bits of information at once. Intel's biggest-selling micros, as well as NEC's V20 and V30, handle at most 16 bits at a time, so they can't run programs written for the biggest computers. NEC says it will ship its 32-bit top-of-the-line V70 by 1987. Two U.S. companies -- National Semiconductor and Motorola -- have recently begun shipping 32-bit micros to the commercial market, and Intel will join them later this year. Japan's vertically integrated giants seldom offer to others the micros they make for their own use because the chips are too specialized to have wide appeal. The potential payoffs of going commercial don't justify the huge expenses -- finding backup suppliers of the chips, adapting popular computer languages to run on the processors (as opposed to the specialized languages used internally), and providing development systems, computers tailored to help programmers write software for a particular microprocessor. The Japanese, however, have mastered CMOS technology, for complementary metal oxide semiconductor, a sophisticated design and fabrication process. Chips made this way consume less than one-tenth the power used by chips manufactured by other metal oxide semiconductor processes, so they're ideal for battery-powered consumer electronic products, a Japanese specialty. Until recently CMOS chips ran too slowly for industrial or business use, and they couldn't be packed with tens of thousands of circuits. But those drawbacks have been overcome, and all the world's chipmakers are scrambling to convert their product lines to CMOS. Chips manufactured by the CMOS process throw off relatively little heat, so engineers can pack circuits together densely without fear that the completed chip will melt (FORTUNE, October 17, 1983). Japan's strength in CMOS has forced American microprocessor companies, which lag behind in CMOS, to cough up an unusually large number of licenses in exchange for access to the technology. Oki churns out the CMOS version of Intel's popular eight-bit micro, the 8085; Intel sells some of those parts under its own name. Toshiba does much the same for Zilog, and Hitachi for Motorola. Also, in 1978 Intel licensed NEC to make certain Intel micros -- partly to get an NEC license for a chip that controls the operation of disk memory devices. Motorola granted Hitachi limited rights to its micros and got back rights to a chip that controls displays on computer screens. IBM has made both Japanese-designed chips best sellers by using them in its personal computers. ULTIMATELY, of course, no Japanese chipmaker wants to license U.S. microprocessors. Intel and Motorola let their hottest new micros go only after they have skimmed the cream of the market. Each company has a single most- favored Japanese licensee -- Fujitsu for Intel, Hitachi for Motorola. Other Japanese producers must settle for licenses to older micros. Such smaller companies as Sanyo, Oki, and Toshiba seem willing to go on grumbling at the U.S. hammerlock. At worst, if they use a lot of U.S.-designed parts in their products, they can usually get a license to manufacture some of the components for their own use. Fujitsu won't discuss plans to become self-sufficient in micros. It's obsessed with beating IBM, and as long as IBM bases many products on Intel's micros, Fujitsu treasures its Intel licenses. The company recently got rights to make Intel's 80286 microprocessor, which drives IBM's superfast personal computer, the PC AT. Reportedly Fujitsu is about to announce a clone of the PC AT. Hitachi is much stronger in peripheral circuits and microcontrollers, which are vital to the company's TV sets and VCRs, than in microprocessors. ''Micros matter mainly because they help sell our bread and butter -- peripherals,'' says Sarv Thakur, president of Hitachi America. Hitachi has announced a laboratory prototype of a 32-bit micro that is compatible with Motorola's processors. The company says it hopes to market the device by 1987, but it's more likely to use the prototype as a bargaining tool to get a license for Motorola's 32-bit chip, the 68020. If negotiations break down, Hitachi probably won't push hard to sell its compatible micro. The company still smarts from the ''Japscam'' incident of 1983, when IBM sued it for stealing secrets and exacted a painful settlement. Hitachi wants to avoid new lawsuits at any cost. Says Thakur: ''We would rather skip a whole generation of products than make a compatible part without a license.'' | BY CONTRAST, the six micros NEC has announced in the V series blanket Intel's line of micros and for the most part are compatible with them. NEC is going for broke: it has lined up Zilog in the U.S. and Sony in Japan as backup suppliers, introduced 11 peripheral chips to go with the V series, teamed up with Hewlett-Packard's Japanese subsidiary to make a development system, and signed deals with a gaggle of U.S. software houses. NEC suffers from coming late to market. Intel launched the first of its 16- bit micros five years ago; some 10,000 different electronic products, from laboratory instruments to personal computers, have been designed around the chips since then. NEC started running seminars last October to entice designers to use the 16-bit members of the V series, but its chips have been designed into fewer than two dozen products so far. The cycle of creating products based on 16-bit chips has passed its peak, and customers who already use Intel parts are loath to switch. As a Japanese company, of course, NEC should have an edge in supplying Japanese customers, especially if the buyer is another arm of NEC. The company's product divisions, however, aren't enslaved by the semiconductor division: NEC's latest personal computer, launched in June, is based on an Intel 16-bit microprocessor. NEC could do much better with the 32-bit members of the V series, which it hopes to launch within a year or two. Though the company will follow Motorola, National Semiconductor, and Intel into the market, NEC should still be in time to hit the peak of design activity for products driven by 32-bit micros. NEC points to several good omens. Since Intel is trying to make its 32-bit micro compatible with its older products, it probably will have to compromise the new chip's design. NEC designed its 16-bit micros with a 32-bit chip in mind, so the Japanese company is freer to wring maximum performance from its 32-bit micro. Because all 32-bit chips will be easier to program than previous generations, and since it should be relatively easy to switch software from one type of 32-bit micro to another, designers won't be as locked into the particular chip they select. They should therefore be more willing to take a flier with a newcomer like NEC. The company could get a windfall because Nippon Telegraph & Telephone is trying to standardize future equipment around a commercial 32-bit micro. NEC has been a favored supplier since 1913.

Whether or not NEC's bold bid to become a power in micros succeeds, Japan's other chipmakers will be watching closely. NEC is the country's most powerful semiconductor house. Should it stumble, that would stifle any hope the others have of breaking the U.S. stranglehold. But every big Japanese chipmaker has the technical ability to produce advanced microprocessors -- so if NEC gets even 10% of the market, the others are sure to unlimber their own designs and join the battle.