Sematech
Sematech was a 1987 consortium of fourteen US chip companies and DARPA that helped rebuild America's semiconductor manufacturing industry.
Sematech was a consortium of fourteen American semiconductor companies, formed in 1987 with matching funds from the Defense Advanced Research Projects Agency (DARPA). Each side contributed roughly $100 million a year. The fourteen companies represented about 85% of the country's chip manufacturing capacity. The goal was to stop the United States from losing an industry Americans had created. Japanese manufacturers had captured the memory chip market through better manufacturing yields and cheaper access to capital. American memory chip makers were leaving the business year after year, and defense officials worried that American weapons would soon depend on foreign-made chips. Sematech is best known for strengthening the American equipment suppliers who built the machines used to make chips.
How Robert Noyce redirected Sematech
Robert Noyce co-invented the integrated circuit and co-founded both Fairchild Semiconductor and Intel. In 1988, at age 60, he left Intel to run Sematech. Noyce had grown up in Grinnell, Iowa, the son of a Congregational minister. In college he was a champion diver and nearly got expelled for stealing a pig from the town mayor's farm for a barbecue. People in the industry called him the Mayor of Silicon Valley.
Sematech's first plan was to build a shared demonstration factory in Austin, Texas, where member companies could learn better manufacturing techniques. It did not work. The fourteen members were fierce competitors, and none of them wanted to share their own methods with the others.
Noyce changed direction. Instead of asking rival companies to share secrets, he pointed Sematech at a problem no single company owned: the manufacturing equipment they all bought. American chipmakers depended on American equipment suppliers, smaller companies that built the tools used to make chips. These suppliers were falling behind their Japanese counterparts, which were larger and better funded. If the equipment suppliers failed, every American chipmaker would depend on Japanese tools.
Sematech shifted about half its funding to these equipment suppliers. It sent engineers from member companies to work on suppliers' tools, funded the development of new machines, and jointly qualified new equipment so that no single company had to bear the full cost of proving a machine worked. Sematech also standardized how equipment was tested and how factories measured their own performance. It published an industry roadmap that told equipment makers what the chipmakers would need and when, so suppliers could plan years ahead.
Noyce died of a heart attack on June 3, 1990, at age 62, after a morning swim in Austin. He was still running Sematech.
Did Sematech save the American chip industry?
American firms pulled back ahead in chip manufacturing during the 1990s, and Sematech often gets the credit. Economists are more cautious about how much credit Sematech deserves.
Several other things happened at the same time. Intel shifted from memory chips to microprocessors. The personal computer boom created enormous demand for those microprocessors. Japan's asset-price bubble collapsed in the early 1990s, cutting off the cheap capital that had funded Japanese chip factories. And a 1986 trade agreement between the United States and Japan committed Japan to stop selling chips below cost and to give foreign firms a larger share of the Japanese market.
What can be said with confidence is narrower. Sematech clearly strengthened American equipment makers. It cut duplicated research across the fourteen member companies. And it gave competitors a forum to agree on standards and a shared technology roadmap, things no single company would have produced on its own. Sematech worked best at fixing shared problems that no individual firm had the incentive to solve.
Sematech as a model for chip policy
In 1994, Sematech's member companies decided they could fund the consortium themselves, and DARPA's matching payments ended in 1996. In 1998, Sematech opened its membership to foreign companies.
The consortium continued doing technical work. In 1997, a Sematech task force had ranked extreme ultraviolet (EUV) lithography last among four candidate technologies for making smaller chips. After a separate Intel-led EUV consortium ended in 2003, Sematech picked up the task of making defect-free mask blanks for EUV, based at a facility in Albany, New York. Sematech relocated its headquarters to Albany in 2007. In 2015, the SUNY Polytechnic Institute absorbed what remained of the consortium.
Today Sematech is cited more often as evidence for industrial policy in chipmaking than it is examined on its own terms. The CHIPS and Science Act of 2022 funded a National Semiconductor Technology Center modeled in part on Sematech.
The historical record supports a more modest conclusion. A consortium can solve shared problems that no single company has reason to pay for alone. Sematech did not invent a new product or create a new market. It fixed the supply chain that its members depended on, and that was the part of the problem it was built to handle.