Betting on Chips : Technology: Upstart Silicon Valley firms are finding success by manufacturing abroad. Critics say the strategy helps foreign rivals at the expense of U.S. industry.
In the classic tradition of the Silicon Valley entrepreneur, Gordon Campbell has struck it rich in the computer chip business. The company he founded less than six years ago, Chips & Technologies, had sales of nearly $300 million last year, and Campbell himself has amassed a fortune.
But unlike traditional semiconductor companies, Chips & Technologies does not actually manufacture anything. Rather, like most of the newer, entrepreneurial semiconductor firms, San Jose-based Chips & Technologies designs and markets its products and pays other companies to produce and package them. In practice, that means that most of Chips’ chips are made by Japanese firms.
Such a strategy is clearly an attractive one for semiconductor entrepreneurs, and in many cases it’s the only rational one. A factory to produce silicon chips--the tiny but complex parts that form the internal circuitry of computers--can cost several hundred million dollars, and start-up companies simply can’t afford it.
Yet many in the U.S. chip business, notably big-company executives such as Andrew S. Grove of Intel and W. J. Sanders III of Advanced Micro Devices, believe that design and marketing firms like Chips & Technologies are doing nothing to help the embattled American electronics industry. By working with Japanese, Taiwanese and South Korean manufacturers, these critics say, the firms are bolstering competitors who already lead the United States in key chip-production technologies.
Moreover, since the engineers and executives who form start-up chip companies are almost always refugees from larger firms--Campbell had been a manager at Intel--these newer enterprises may also be sapping the strength of the big American companies carrying the flag for the United States in the international technology wars.
Could it be that the entrepreneur, the icon of Silicon Valley and American capitalism, does more harm then good in a high-tech world where cash-rich international conglomerates rule?
Small companies staunchly defend their contribution to U.S. technological prowess, maintaining that entrepreneurial firms are far more innovative and do a better job of meeting customer needs than large ones.
But others say innovation isn’t enough.
“For years we’ve splintered, and as we’ve done so we’ve hurt the capabilities of the large, powerful companies,” said Alfred J. Stein, president of mid-size chip maker VLSI Technology, based in San Jose. “All these little companies get a niche, and that detracts from the strength one could achieve from massive size.”
Indeed, the debate over the so-called fabless chip companies--they lack the chip fabrication facilities known as “fabs”--is just one aspect of a broader conflict between large and small companies in Silicon Valley. Big companies consider the smaller ones parasitic and short-sighted. Small companies view the established firms as slow moving, bureaucratic and given to blaming their own management failures on the government and the Japanese.
That split has contributed to the controversy over what, if anything, the government should do to enhance the competitiveness of the U.S. chip industry. While the large companies have strongly promoted the Sematech research consortium and the U.S.-Japan semiconductor trade agreement, many small firms consider these policies little more than special favors for certain big companies. Entrepreneurs care far more about lower capital gains taxes and lower interest rates, which would make it much easier for them to raise money and invest.
One of the few things the two camps agree on is that fierce personal rivalries, the growing number of lawsuits and a general atmosphere of contentiousness in Silicon Valley are seriously hampering the development of cross-ownership arrangements, licensing deals, joint ventures and other kinds of alliances that could help American companies compete. If the new fabless chip companies had more ties to U.S. manufacturers, some of the conflicts would disappear.
The semiconductor business is not the only one where America’s love affair with small entrepreneurial firms has worked to the benefit of overseas competitors. The breakup of the greatest techno-giant of all, American Telephone & Telegraph, was a boon to American entrepreneurs, but among the greatest beneficiaries have been European and Asian telecommunications equipment companies.
In chip-making equipment, advanced materials and machine tools, foreign companies have profited from the fragmentation of American industry.
T. J. Rodgers, president of the fast-growing manufacturer Cypress Semiconductor in San Jose and an outspoken critic of the chip establishment, says better links between “the ever-smaller mass of the designers and the ever-larger mass needed to be a manufacturer” are crucial to strengthening the U.S. semiconductor business. He calls the rapid growth of fabless firms that work with Asian manufacturers “a camouflage for the gutting of American industry.”
In that respect, Rodgers is in rare agreement with Grove, who believes the loss of manufacturing capability will help make the United States a “techno-colony” of Japan. Sanders says the United States is on its way to becoming “the Italy of the 21st Century” with strong design capabilities but no manufacturing base, and a lower standard of living as a result.
For their part, many of the fabless entrepreneurs acknowledge the problems posed by Asian manufacturing alliances but say they are the only way to bring innovative products to market. “It’s a matter of doing what we are doing or we wouldn’t exist,” said Bernie Vonderschmitt, founder and president of San Jose-based Xilinx, which designs sophisticated specialty chips.
He said the fabless business model is a symptom of the U.S. industry’s ills, not a cause. “With the tax policies we have, with the cost of capital we have, capital intensive businesses are going to go out of business in the United States,” he said. “Over the next year or two, we’ll be approaching the size where we could have a fab. But we would impair our profitability, and as soon as your profit goes down, Wall Street says good bye.”
Indeed, Wall Street analysts, impressed with the dramatically higher profit margins that fabless companies show compared to the rest of the industry, are among their greatest supporters. Advocates of the fabless model also maintain that most of the value in advanced chip products lies in design, and therefore it makes little difference who actually does the manufacturing. Even some large chip manufacturers, including Intel, now have some of their products manufactured by others.
Critics of the fabless companies acknowledge that many have done a superior job in identifying new markets, developing innovative products and moving quickly to stay abreast of the latest trends--all the things that entrepreneurial companies are supposed to do.
But they say companies that are dependent on Asian manufacturing, however profitable they may be now, will either be swallowed up by their erstwhile production partners or find that their partners have become competitors. And the net result for American industry will be the continued transfer, via start-up companies, of U.S.-developed technology to Asian competitors.
Hector Ruiz, senior vice president at Motorola’s semiconductor group, says that because companies that manufacture for others usually demand licensing or other rights to the product, fabless chip companies have allowed Asian competitors “to gain access to technology at very low cost.”
Few believe that the answer to this quandary is to encourage great size for its own sake. Big U.S. electronics firms such as General Electric, RCA and Honeywell were conspicuous failures in the chip business. It was tiny upstarts, notably Fairchild Semiconductor and later Intel, that built the industry.
“If we were part of Intel, we wouldn’t exist,” said Rodney Smith, president of San Jose-based Altera Corp. He finds more than a little irony in hearing one-time entrepreneurs such as Grove and Sanders beating the drum for large companies. “It’s just sour grapes on the part of the big guys.”
John C. East, president of Sunnyvale-based Actel, another fabless firm, laments the fact that “culture, cost of capital, and lack of government support” have driven much of the chip production out of the United States. But he says the United States must at least try to take advantage of its continued ability to innovate, and if the only way to do that is to work with foreign companies, then so be it.
East also points out that although his company is very happy with its original manufacturing partner, Matsushita, it specifically sought out an American company when it was time to find additional production capacity. Most of the fabless firms, in fact, have some of their manufacturing done by American companies.
But fabless firms generally have far stronger links with Asian companies. Pierre Lamond, a venture capitalist who has been involved in many start-ups, said it’s generally much easier for small companies to find Japanese partners than American ones.
Indeed, while Japanese and Taiwanese firms build new factories with the explicit goal of acting as a production partner for U.S. design firms, most American chip manufacturers have expressed little interest in making chips for others.
“The larger companies have to change their attitude,” said Charles Ferguson, a Massachusetts Institute of Technology researcher who advocates the creation of groups of interlocking companies that cooperate with each other yet remain independent. He and others are sharply critical of the larger firms’ unwillingness to work with their smaller brethren.
But the big companies say manufacturing for others is not profitable enough, and they worry that the fabless company’s products will be competing with their own. Intel Chairman Gordon Moore calls pure contract manufacturing “a nuisance business” that the company will not actively pursue.
Sanders of AMD says U.S. firms can’t compete for manufacturing business because Asian companies, with lots of cash to play for the long term, “are willing to make deals that in the short term are terrible for them.” American companies, he says, simply cannot match the kind of pricing and support that Asian manufacturers are willing to provide.
Some U.S. companies are pursuing new kinds of arrangements that encourage entrepreneurial creativity while keeping the technology from migrating overseas. Cypress Semiconductor has acted as a venture capital financier for several start-up design firms, and is also a production and distribution partner for them.
Cypress has also sold a minority interest in one of its factories to Altera Semiconductor, which thereby obtained some secure manufacturing capacity at a fraction of the cost of building a factory. More of these arrangements may be on the way.
“We’re now seeing a lot of business plans (for start-up chip companies) which include a corporate player who brings manufacturing or distribution,” said Andrew Rappaport, president of Technology Research Group, a Boston consulting firm.
At Chips & Technologies, though, no one seems too worried about dependence on the Japanese. Chips has manufacturing deals with four big Japanese chip firms--Toshiba, Fujitsu, NEC and Yamaha--as well as Taiwan Semiconductor Manufacturing Co. and three American companies, LSI Logic, National Semiconductor and NCR. But Campbell, who serves as chairman, president and chief executive, said most of the work is done with the Japanese firms.
Some analysts believe that Chips is entering a difficult period as Intel, AMD and a host of start-up companies zero in on its market. But if Campbell needed further vindication of his fabless business model, he got it in September when Seeq Technologies--a company Campbell helped found but left after a dispute with his partners--announced that is was shutting its factory. San Jose-based Seeq thus became the first semiconductor manufacturer to go fabless.
HOW “FABLESS” CHIP FIRMS OPERATE
1. The “fabless” chip company, often in consultation with customers, designs a chip or set of chips using sophisticated computer workstations and highly specialized electronic design software. 2. The information describing the design is shipped in the form of computer tapes to the company that will do the manufacturing. Through a complex photo-chemical process, the manufacturer etches the chip designs onto a raw silicon wafer. Initially, the process requires close attention to improve efficiency, and engineers from the fabless design company sometimes assist in fine-tuning the process. 3. The complete silicon chips are shipped to another company which packages them in ceramic and plastic casings. 4. The packaged chips are shipped back to the original chip company for testing. After testing, they are delivered to customers or distributors. Chips eventually endup on circuit boards.
