Chip-Making Partnership Suffering Growing Pains : Policy: Bid to improve U.S. competitiveness highlights the hazards of government-industry cooperation.
On a quiet hilltop overlooking this capital city’s downtown, a state-of-the-art computer chip factory stands partly empty, a monument to the problems that have plagued one of America’s boldest experiments in technology development.
The facility, located here after a spirited bidding war among California, Texas and many other states, was supposed to be the crown jewel of Sematech, the government-industry consortium that spends $200 million a year--half of it from taxpayers--in an effort to help American companies compete with the Japanese in making the brains of modern electronics.
But Sematech has abandoned its original goal of creating a futuristic laboratory for the mass production of computer chips. Now, a new management team is remaking the organization, funneling much of its money directly to producers of chip-making equipment and working to build better relations among the many factions of the U.S. chip industry.
Sematech’s critics consider the change in direction evidence of failure. Yet, many chip industry executives and advocates of an activist government approach to technology policy say the consortium is now on the right track.
Indeed, support for Sematech in Washington appears to be strong. The consortium’s original five-year charter expires next year, but it is likely that both the Administration and Congress will support Sematech’s recently completed proposal to continue its work for another five years.
The new five-year plan, which will be scrutinized by Congress in the coming months, calls for the consortium’s budget to remain about the same and lays out an ambitious agenda for perfecting a new breed of computer-controlled chip-making processes.
Certainly, Sematech has been an interesting learning experience for all of those concerned with improving America’s industrial competitiveness. It highlights many of the hazards inherent in government-industry collaboration, particularly the reluctance of individual companies to commit their best people to cooperative efforts and the potential for government-funded programs to degenerate into political pork barrels.
At the same time, the upside of Sematech’s mixed record suggests that such efforts can be successful if they are kept manageable in size, remain focused on a limited set of clear and feasible goals and worry more about retaining the support of industry than playing politics.
“Sematech has helped the supplier base, it has helped the member companies, it has been cost-effective,” said Craig Fields, former head of the Defense Department agency that helps sponsor Sematech and now chief executive of a neighboring Austin consortium, the Microelectronics & Computer Technology Corp.
“But, it originally had much larger goals in terms of regaining leadership (in chip production technology), and that hasn’t happened. It’s just a very big job.”
The U.S. chip industry has been struggling since the mid-1980s, when Japanese competitors succeeded in taking over the computer memory chip business--the biggest piece of the market. Some of that success was due to “dumping,” or selling products below cost, and in 1986 a special trade agreement was signed to prevent that practice.
But it was clear that the Japanese firms had also developed superior manufacturing capability. Specifically, they had been able to achieve radical improvements in “yield,” or the percentage of working chips in each batch, and thus were able to mass-produce sophisticated chips at lower cost. Their advantage was first apparent in memory chips, but it applied to many other types of chips as well.
So Sematech was founded to bolster the production prowess of American companies. Only by working together--with some government help, the reasoning went--could U.S. chip companies afford to keep pace with their larger Japanese competitors.
William Spencer--the lanky, mustachioed former Xerox Corp. executive who took over as Sematech’s chief executive last October--maintains that most of the specific technical objectives established when Sematech was formed in 1987 have been achieved.
But, he said, it is too early to judge whether the consortium can reach its broader goal: re-establishing America’s ability to compete in high-volume chip manufacturing.
“Sematech is really only about 2 years old,” said Spencer, an articulate and personable engineer who spends a lot of his time explaining--and defending--the very existence of the organization he runs. “We signed our first joint development contracts and made our first (chips) two years ago. We still have a ways to go in managing (the relationships) with the member companies, the equipment companies, the universities . . . . “
As Spencer explains it, Sematech was never supposed to develop the extremely complex technology needed to make computer chips on its own. Rather, the goal was to “leverage” the $4 billion to $5 billion that is already being spent on these technologies in the private sector, in academia and in government laboratories.
Originally, however, much of the work was to take place in Sematech’s own factory--known in industry parlance as a “fab.” California considered the facility such an important piece of high-tech infrastructure that it offered $125 million in incentives to have it built in the state. But ultimately the victor was an Austin site leased from the University of Texas.
The idea was to create an experimental factory in which new production methods could be honed to perfection for the collective benefit of American chip-makers.
But Sematech had problems from the outset.
No one could be found to take the job of chief executive until Robert Noyce--the legendary inventor of the integrated circuit, who died last year--ended up in the position almost by default.
Once it got going, Sematech found that it was difficult to simulate high-volume commercial production without violating its congressional charter by making real commercial products.
The member companies--which include very large, profitable firms such as International Business Machines, American Telephone & Telegraph, Motorola and Intel--were spending far more than Sematech on their own process-development programs. And they would not commit their best engineers to the collective effort.
Meanwhile, another problem was becoming increasingly obvious: The domestic suppliers of the exotic materials and complicated machinery that lie at the heart of chip manufacturing were losing their edge. In 1980, the nine largest chip-equipment firms all were American. But by 1990, four of the top five were Japanese.
So Sematech began several years ago to move resources to “equipment development” and “equipment improvement” programs, under which funds are awarded to equipment vendors for development work on particular machines. Once Spencer took over, that shift accelerated: Sematech this year will spend more than 60% of its $200-million budget on these programs, rather than on its own production line.
Bill George, a Motorola executive who is now serving as chief operating officer at Sematech, says that, when he first came on board in March, he carried around a notebook in which one page was entitled, “Reasons we should have a fab.” A complete shutdown of the fab was seriously considered, he said, but he ultimately found enough reasons to keep it going.
“Manufacturing products here is very costly and doesn’t result in direct technology transfer (to the member companies),” said George, who previously headed Motorola’s memory chip joint venture with Toshiba. “We will maintain our fab, but not the capability to have volume.”
In Sematech’s revised strategy, a key objective is to foster better relations between the equipment companies--most of which are small--and the chip-makers. In Japan, equipment firms are often part of the same conglomerates as the chip-makers. Even in cases where they are not, there is always a tight working link. That is important, because chip-making machinery is very expensive and takes many months to fine-tune to customer requirements.
Yet there are hazards in Sematech’s new policy of distributing money to equipment suppliers. Although information about Sematech’s spending is kept secret on national security grounds, it appears that the main beneficiaries are the largest chip-equipment firms. And some of the money also flows back to Sematech members.
Applied Materials, for example, is the largest American equipment firm, with sales of $567 million and profits of $34 million last year. The San Jose company has received three Sematech contracts; one of them involves joint development work with Intel, which had $4 billion in sales and $600 million in profits last year.
T. J. Rodgers, chief executive of Cypress Semiconductor and a relentless critic of Sematech, calls pacts such as the one between Applied Materials and Intel “kick-back contracts” because they involve Sematech money flowing back to dues-paying member companies. Rodgers says the deals are an indication that Sematech is little more than a pork barrel project for the big chip firms.
Critics say also that the money flowing to equipment firms is directed at programs that member companies would be investing in anyway, and that there is too much room for favoritism.
Sematech is “in the position of kingmakers: They pick one company for each technology, and it’s usually the biggest one,” said Robert Graham, chief executive of Novellus Systems, a widely respected equipment company that has no Sematech contracts. “There are a lot of companies out there that have people dedicated just to getting money out of Sematech.”
Sematech executives and member companies defend the way that the money is distributed.
Jerry Sanders, the flamboyant chairman of Advanced Micro Devices, resents the focus on Sematech’s government funding, preferring to draw attention to the $100 million that industry contributes.
“My company is putting $10 million into Sematech,” he said, pointing out that dairy farmers, tobacco growers and others who get government money do not provide matching funds. “This is the first instance of industry and government working together to leverage R&D; . . . a noble experiment.”
According to Sematech, the amount of money flowing back to member companies represents about 4% to 5% of the $160 million in equipment contracts awarded over the last several years.
Some of the other priorities that Spencer has established are less sensitive than the equipment grants.
Establishing a standard set of quality and performance criteria for different pieces of chip equipment and coordinating the far-flung chip research efforts in the Department of Energy, Department of Defense and university research laboratories are both considered helpful and non-controversial.
“We can represent a single industry voice. We can go into a (national laboratory) and say, ‘The semiconductor industry needs better plasma processing equipment,’ ” Spencer said.
It is not clear, though, that such a large budget is needed for this kind of activity.
Few, in fact, advocate an increase in Sematech’s funding. Pallab Chatterjee, vice president of Texas Instrument’s semiconductor group, suggests, in fact, that a smaller Sematech might be more effective. The new five-year plan for 1992-1997 calls for funding to remain level.
In many areas, it is difficult to evaluate Sematech’s effectiveness. Much of the technical focus is now on special software and esoteric technologies for further automation and integration of the chip production process. Spencer says the research could lead to a substantial leap forward in the overall design of chip manufacturing processes, but the results will not be measurable for some time.
In any event, the debate over Sematech’s effectiveness tends to be warped by the political climate.
Even industry executives who concede in private that Sematech has not produced a lot of bang for the buck are loath to criticize the consortium in public. It remains, after all, one of the few initiatives that industry has been able to wring from an Administration hostile to anything that smacks of “industrial policy.”
Sematech thrust itself into the political fray in May when it joined with Sen. Lloyd Bentsen (D-Tex.) in accusing Japanese companies of withholding their best chip production technology from American chip-makers.
But the evidence behind those allegations--contained in a General Accounting Office study released in late September--is sketchy.
And such political involvement is risky. It could provide ammunition for those who regard the consortium as a political entity--a kind of high-powered extension of the Semiconductor Industry Assn.’s lobbying efforts--rather than an organization truly dedicated to developing technology.
And if it does its job right, Sematech shouldn’t need a purely political justification. Even critics agree that its goals are legitimate. Over the next few years, it will become clear whether the consortium is the best way to achieve them.
“Sematech started out trying to be all things to all people,” said Craig Barrett, chief operating officer of Intel. “Now it’s narrowed, and it’s starting to accomplish more or less what it set out to do.”
