The largest manufacturer of semiconductors in the world has never sold a single chip.
But after almost 30 years of making chips exclusively for its computers and other products and -- more recently -- serving as patron and de facto R&D; center for the entire U.S. chip industry, International Business Machines Corp. is looking for a payback.
The Armonk, N.Y., computer giant is hoping its $6.4 billion semiconductor operation has developed technology -- and products -- it can finally make some money on. And its first real customer may be Apple Computer Inc., its old desktop #i computing rival.
This is just a piece of what its executives jokingly refer to as "the new IBM," a company that now is open to all kinds of interactions with competing companies. Over the past five years, IBM has forged partnerships, invested in technology ventures, participated in industrywide research programs, started selling its disk drives and has even begun joint manufacturing and sales of chips.
But no matter what chips IBM ends up selling, the driving force behind its far-flung and low-profile semiconductor business will continue to be supplying the company's bread-and-butter line of mainframe-to-laptop computers.
"If we want to remain a leader in the computer business, we have to stay a leader in the semiconductor area," declares Jack Kuehler, IBM's president. "And there is much more pressure to do that now than in the past."
At a time when fierce competition is depressing profits in the computer business, IBM is spending more than $1 billion a year -- roughly equivalent to one year's revenues at Advanced Micro Devices Inc. -- on semiconductor research and equipment. To make sure IBM can afford to develop advanced products in the future, it is forging research partnerships with U.S. and European firms.
And it cannot back off from helping the troubled American semiconductor industry, which increasingly looks to IBM for money, technology and guidance about the future. In the past few years, the company has invested hundreds of millions in a variety of U.S. chip ventures, from equipment companies to research consortia.
"IBM is the godfather of the American semiconductor industry," says Dan Hutcheson, president of VLSI Research Inc. "What they've done throughout the last half of the '80s and right through until now is try to coddle the [industry] along in the right direction."
This is a different IBM than it was a few years ago, says Paul Low, vice president and general manager of technology products. "We never used to do these kinds of arrangements," Mr. Low says. "We never admitted we needed to share. We did change, and we are still changing."
One of the biggest changes in the IBM mind-set is the plan to sell its reduced instruction-set computing, or RISC, microprocessor, the R6000.
IBM's deal with Apple gives it the opportunity to sell microprocessors "on a selected basis," says Michael Attardo, an IBM vice president and general manager of the General Technology Division, which houses IBM's semiconductor operation. In fact, Mr. Attardo and other executives think IBM might actually sell its version of the processor to Cupertino-based Apple before manufacturing partner Motorola Inc. can get it up and running.
In the future, IBM will consider selling other products as well, Mr. Attardo says. "This is the new IBM. We are open to anything that makes sense," he says.
What hasn't changed at IBM is the central role semiconductors play in its family of computers. Ten years ago, Mr. Attardo says, chips accounted for just 10 percent of the value of a computer. Today, they make up 30 percent to 40 percent of what the computer is worth, and by 1995, semiconductors will be represent upward of 60 percent of the value of the whole system.
Because a growing number of functions are being packed into a computer, Mr. Attardo says his worldwide chip operation is under increasing pressure to develop faster and more complex parts. Although IBM makes a wide range of semiconductors, from custom chips to microprocessors, the core of the entire operation is dynamic, random-access memories, or DRAMs.
IBM is now the largest U.S. maker of DRAMs, in part because most U.S. chip companies were driven out of the DRAM business in the mid-1980s when Japanese companies were illegally "dumping" memory chips in the United States at prices below their cost.
IBM uses its DRAM manufacturing process as the basis for all of its products. "It's given us a leading-edge RISC microprocessor, as well as other things," Mr. Attardo says. IBM currently licenses its 4-megabit DRAM technology to several semiconductor operations, including Sematech, the semiconductor manufacturing research consortium in Austin, Texas.
But IBM is having to work harder and harder to stay ahead of everyone else in memory technology, Mr. Attardo says. Although the company was the first to manufacture the last three generations of DRAMs, the gap between IBM and its Japanese competitors is narrowing. The company was barely months ahead of the Japanese in making the latest 16-megabit DRAM, and now it is working hard to ensure it can be producing a 64-megabit chip by 1995.
One way IBM is trying to improve its odds in the memory race is by asking its engineers to work on three successive generations of products at the same time. In other words, IBM is making the 16-megabit chip right now, developing the 64-megabit and doing early research on the 256-megabit memory, Mr. Kuehler, the IBM president, says. In fact, Mr. Kuehler says IBM wants to develop semiconductor technology, new computers and new manufacturing facilities all at the same time.
And by spending huge amounts on manufacturing and equipment research, IBM hopes to have the tools it needs to stay ahead of the Japanese, Mr. Kuehler says. The company has spent about $750 million in the past few years on its Advanced Semiconductor Technology Center in East Fishkill, N.Y., which houses much of its experimental manufacturing program.
Mr. Kuehler describes the facility as a "consortium inside of IBM."
IBM is not working on the future in a vacuum. In the memory area it has teamed up with Siemens AG of Munich to develop a 64-megabit DRAM, and it may work with Siemens on future DRAM research, Mr. Kuehler says. It has also agreed to jointly make and sell DRAMs with Siemens at IBM's chip-making plant in France.
IBM's investments internally are not enough to ensure that it will remain a technology leader, however. The company is also one of the world's largest buyers of semiconductors, and it does not want to buy a large percentage of its chips from Japanese companies it may compete against in other areas.
"What we would like to see is a more balanced world in terms of semiconductors," Mr. Attardo says. "We really have to strengthen our semiconductor infrastructure."
To accomplish that, IBM has been very active over the past five years in industry trade groups and research consortia. More important, it has helped fund several struggling equipment companies to make sure their technology developing did not end up owned by Japanese operations.
IBM took a lead role in helping to fund Silicon Valley Group's purchase of Perkin-Elmer Corp's lithography operation, and it also helped set up Etec, a company in Fremont, Calif.
But IBM executives don't want to be seen as the lone source of help for the industry.
"We don't view ourselves as the single savior in technology, and we don't want to be seen in a role as the provider of technology to the United States," Mr. Attardo says. "We just want to be one of the companies that help out."