TALK ABOUT YOUR DREAM TEAM
It has to be the alliance of all alliances: On July 13, a deal lasting eight years or more and worth up to $1 billion was announced by IBM, Toshiba, and Siemens. They are, in order, the world's biggest computer company and chipmaker, Japan's second-largest chipmaker, and Europe's No. 3 semiconductor house. Even in the semiconductor business, perhaps more globalized than any other, nothing quite like this linkup has ever taken place. But then, what these powerhouses are setting out to do has never been attempted before.
The three companies are joining forces to develop a semiconductor technology so daunting and expensive that none dared risk it alone. The goal is a 21st century chip on whose tiny silicon surface will be etched what amounts to a street map of the entire world. Those electronic streets--lines just 0.25 micron wide, 400 times thinner than a human hair--will link some 600 million transistors. When the chips become available around 1998, each will store 256 million bits of data, or about two copies of everything that Shakespeare wrote. The same technology will also be useful in creating microprocessors with the power of today's supercomputers.
What it will take to design and produce such dense and elaborate circuits--except for scads and scads of money--nobody really knows. The trio expects that just designing the first 256-megabit memory chip and its fabrication processes will gobble a cool $1 billion, and building factories to produce the chips in volume will run another billion each.
These kinds of astronomical costs are the reason such huge international alliances are sure to become the norm in the future. In fact, on the same day IBM, Toshiba, and Siemens broke their news, Fujitsu Ltd. and Advanced Micro Devices Inc. announced a sweeping plan to collaborate on EPROMs, or electrically programmable read-only memories, and flash chips. In addition to splitting the cost of a new, $700 million plant, the two companies intend to purchase up to 5% of each other's stock as insurance.
All this cross-border partnering muddies the water for U.S. politicians accustomed to championing national competitiveness. In September, a House science subcommittee will hold hearings on Sematech Inc., the Austin (Tex.) consortium set up in 1987 to restore America's competitiveness in chipmaking technology. Sematech has already received $500 million in federal funds, so lawmakers are sure to ask some tough questions, predicts a House staffer. "After all the screaming and yelling about keeping the U.S. competitive when people were trying to get Sematech started," he says, "why, all of a sudden, are we are signing up with our worst enemies?"
Because of antimonopoly laws, says Clyde V. Prestowitz Jr., president of the Economic Strategy Institute and a noted critic of U.S.-Japan trade relations. "You know what would've happened if IBM had slipped into bed with two American companies the size of Toshiba and Siemens." Similarly, the Japan Fair Trade Commission would probably have slapped down any Toshiba plan to nuzzle up to, say, NEC Corp. and Hitachi Ltd.
HOOKUPS. The semiconductor industry started making the argument for international alliances years ago. Pasquale Pistorio, president of SGS-Thomson Microelectronics Inc., Europe's second-largest chipmaker, has been predicting since 1983 that by 2000 only a handful of suppliers will be able to retain global market shares of 5% or more. Others will be niche specialists, with segments of 0.5% or less. William P. "Pat" Weber, head of Texas Instruments Inc.'s chipmaking operations, says that learning to share risks will be a necessary condition of survival.
The worldwide semiconductor industry is already a crisscross of major collaborations. Texas Instruments is teamed with Hitachi, and Motorola Inc. with Toshiba, both since the late 1980s, to develop memory chip technology. Those pacts, and one between American Telephone & Telegraph Co. and NEC Corp., may soon be extended to more advanced chips. Toshiba Executive Vice-President Hideharu Egawa says IBM and Siemens have tacitly agreed Toshiba can share with Motorola technology developed by the three new partners. The reverse will also happen, because IBM is cooperating with Motorola to develop ways to "print" ultra-dense circuits with X rays.
Intel Corp., says Chairman Gordon E. Moore, is "making phenomenal capital investments" in microprocessors for personal computers and other markets. But it signed on with Japan's Sharp Corp. to share the cost of a factory to make flash memory chips, a new type of storage that retains data even when turned off.
With U.S. companies embracing transnational alliances, Sematech may now actually be more important than ever, says Ian M. Ross, president of AT&T Bell Laboratories and former chairman of the National Advisory Committee on Semiconductors, which was established by Congress in 1988. To keep jobs in the U.S., Ross explains, "our companies must be able to negotiate from positions of strength--in manufacturing as well as design." Production has often moved overseas, he adds, because U.S. chipmakers lacked the manufacturing muscle Sematech is helping to develop.
Large U.S. chipmakers, moreover, don't see international alliances tempering their desire for domestic competitiveness. "It won't take any fervor out of my argument that Japan ought to be buying more U.S. semiconductors," declares Intel's Moore. Adds IBM President Jack D. Kuehler: "We've absolutely got to have a healthy semiconductor industry in this country in order to have a healthy electronics industry--and that's also true for Europe and for Japan." That's why IBM is "firmly committed to Sematech," he says, and to Jessi, its European counterpart. Now that Sematech and Jessi are helping to stabilize the U.S. and European chipmaking industries, IBM figured it was time to try a three-way hookup.
To Bell Labs' Ross, the terms of the IBM-Toshiba-Siemens deal prove that American chipmaking knowhow is winning new respect in Japan. The 256-megabit project is centered at IBM's new Advanced Semiconductor Technology Center in East Fishkill, N.Y., where a trilateral team of some 200 engineers will report to a Toshiba manager. It could be the prototype for more ventures, in various industries, uniting all three continents.COSTLY MEMORIES
Year Capacity Line width R&D cost
Megabits Microns Millions
1985 1 1.2 $100
1988 4 0.8 200
1991 16 0.6 350
1995* 64 0.35 600
1999* 256 0.25 1,000
DATA: IBM, DATAQUEST INC., BW
Otis Port in New York, with Richard Brandt in San Francisco, Neil Gross in Tokyo, and Jonathan B. Levine in Paris