Sematech could not possibly have accomplished that mission, given its funding level–and it didn’t. The U.S. semiconductor industry did return to world leadership in the manufacture of both semiconductors and semiconductor processing equipment, but Sematech was not crucial to either accomplishment, although it certainly did play an important supporting role.
Why did Sematech take on a mission that it could not accomplish? Why did it initially swagger about like an Arnold Schwarzenegger-style action hero, when it was more like Danny DeVito in size? Because it really thought it was Schwarzenegger.
This brings us to the first lesson that Sematech can teach us about taking the consortium approach to increased competitiveness: Initial planning can be pivotal. Inadequacies here may haunt you for years to come.
“When Sematech was first created, a planning group back in California dreamed up what a consortium like Sematech could do,” says M. Franklin Squires, Sematech’s senior vp and chief administrative officer.
“According to their model, Sematech would address all of the technology issues associated with semiconductor manufacturing. The original thought was that we would have programs in the front end |of the semiconductor wafer production process~ and in the back end of testing, packaging, and assembly, and everything in between, including materials research.”
As the consortium got going, however, “people very quickly realized that at $200 million we would not be able to do all those things,” continues Squires. “So the focus of Sematech narrowed to include just the heart of the wafer fab operation. There was to be little materials work, no packaging, no testing, and very little on manufacturing methods.”
Faulty early planning also led to the departure of two member companies. LSI Logic and Micron Technologies decided to leave the consortium as soon as their four-year commitment ended in 1992, primarily because manufacturing methods research had been greatly curtailed.
The next shock came in early 1989, when a Sematech study found that its members expected to buy less than 40% of the equipment they needed for advanced wafer production from U.S. companies. European semiconductor makers reported that they expected to buy only 30% of their advanced equipment from U.S. suppliers.
Sematech should have considered the U.S. equipment suppliers’ plight early on, but didn’t. Now it had to shift its priorities again.
It would be wrong to think that bumbling characterized Sematech’s early years. Mistakes were made, yes, but much of the groundwork for what Sematech is today was laid at that time.
One of the smart things the early planners did was see to it that Sematech remained an industry-led consortium, despite the presence of the federal government. The government was accorded about the same status as a member company, when it came to policy-making. That meant that if the government ever tried to impose its own agenda, it could be outvoted time and time again.
This was especially important because the fledgling consortium was placed under the wing of the Defense Advanced Research Projects Agency (DARPA, now ARPA), an agency that has had some trouble understanding what commercial R&D is all about.
“DARPA/ARPA has long prided itself on funding high-risk scientific breakthrough-type R&D activities, while Sematech is at the other extreme,” says Squires. “We are very near-term focused. We want most of our projects to have commercializable results in 18 to 24 months. And that’s about as far apart as you can get.”
In 1992, DARPA proposed phasing out funding for Sematech. The plan called for Sematech to get only $80 million in fiscal 1993 and nothing thereafter. The plan was not adopted, but it should serve as a warning to any consortium wishing independence and government funding at the same time.
Another smart thing Sematech did early on, according to a September 1992 U.S. General Accounting Office (GAO) report, was to involve member company senior executives in establishing Sematech’s R&D priorities and in overseeing its technological progress. These executives “emphasized the importance of tracking the progress of ongoing R&D projects and proposed corrective actions when priority projects fell behind schedule,” all of which helped Sematech “tighten its management controls.”
Finally, right from the start, Sematech provided funding for university research through its Centers for Excellence program. Sematech now works closely with 31 universities and research centers in 14 states on long-term research projects that complement its own near-term R&D. The $20 million per year Sematech now provides is managed by the Semiconductor Research Corp.
When Sematech recognized how bad off the U.S. equipment suppliers were, it moved quickly to help. One complaint it heard repeatedly was that suppliers were being hurt by a lack of communication between themselves and their customers, the semiconductor makers.
So in June 1990, Sematech unveiled its Partnering for Total Quality program, whose aim is to foster closer long-term working relationships between semiconductor manufacturers and their key suppliers. The program calls on manufacturers to (1) share strategic goals and plans with suppliers; (2) give them greater access to information about the long-term performance of their equipment; (3) provide them with competitive analysis information; and (4) support them in product development.
By 1991, about 48% of Sematech’s budget was devoted to helping U.S. equipment suppliers improve existing equipment and develop next-generation manufacturing technology.
Sematech’s ongoing equipment improvement program helps U.S. equipment suppliers in several ways, says Chris Mack, vp of R&D at Finle Technologies, Austin, TX. Mack served a two-year stint at Sematech on assignment from the Dept. of Defense.
“First of all, Sematech helps fund research; it provides actual dollars,” Mack says. “Also, Sematech has developed various standards–for tools, for how tools interface, for software, for how tool reliability and cost of ownership should be measured–things that all the member companies have agreed on.
“Therefore, when suppliers meet these standards, they can get qualified for all of the member companies simultaneously, whereas in the past, they would have to jump through a different set of hoops for each company, which became very time-consuming and expensive.”
Sematech’s equipment improvement program has become a shining example of how generic, precompetitive R&D can help a consortium’s member companies and their suppliers at the same time.
The program helps Sematech’s remaining 11 members (AT&T has purchased NCR) by taking much of the guesswork out of making equipment purchases. Members gain an intimate knowledge about the capabilities of the tools under development, which allows them to position their manufacturing operations to insert the tools the minute they’re commercially available, Squires says.
This program allowed Sematech to dramatize that U.S. suppliers have now achieved parity with the Japanese, by demonstrating last January that American-made equipment could manufacture leading-edge integrated circuits having device widths as small as 0.35 ||micro~meter~.
More importantly, the program has begun changing the buying habits of Sematech’s member companies. For example, when Motorola was planning its state-of-the-art MOS 11 manufacturing facility several years ago, it estimated that it would have to purchase about 80% of its equipment from foreign sources. The company ended up purchasing 80% from U.S. companies, and credits Sematech for much of the turnaround.
Likewise, Intel says it purchased $150 million more in U.S. equipment last year than it had anticipated, because of Sematech.
When the GAO asked member companies to list Sematech’s most important initiatives, Partnering for Total Quality and the equipment improvement program came up repeatedly. But these were by no means the only things mentioned.
“Several member companies have stated that one of Sematech’s most important contributions is as a forum for communications, enabling members to discuss manufacturing problems and solutions,” says the September 1992 report. “In many cases, member companies have found that they were protecting similar trade secrets and experiencing similar problems. Through discussions, members have better identified the nature of manufacturing problems and the success or failure of approaches taken to fix them.”
As it looks to the future, Sematech says it will seek to broaden its horizons to include those technological areas that were not funded sufficiently during its first six years. These include research into materials, design, test, packaging, and computer-integrated manufacturing.
Some measure of Sematech’s present standing in the semiconductor industry can be seen from the fact that even T.J. Rodgers–president and CEO of Cypress Semiconductor Corp. and Sematech’s most clamorous critic–is now willing to offer it praise (if not without a touch of sarcasm):
“There are a lot of good things you can say about Sematech. A lot of its programs and developments are valuable. I think that as far as government agencies |sic~ go, it’s pretty well run. My objection isn’t to what Sematech does; my objection is that I don’t think taxpayers ought to be footing the bill.”