The Man Behind the Microchip

by Leslie Berlin

  In 1984, Noyce decided to buy a Cessna Citation, a light jet that required more sophisticated piloting skills than he had needed for any of his other aircraft. His plane would be delivered to the site of Cessna’s flight school in Wichita, Kansas. After he graduated from a two-week training course with the type rating he needed for the Citation, he would be able to fly his plane home.69

  Noyce paid for Jim Lafferty to attend the flight school with him, and both men found the program grueling. A typical day consisted of eight hours of direct instruction in classrooms and simulators, plus a few evening hours spent reviewing videos. Noyce had never flown such a fast or complex aircraft. He found it frustrating to grope his way around the new instrument panel. In the simulator, he had problems controlling the plane’s speed and thought he flubbed his attempts at landing using instruments. His instructors, who saw students with these troubles all the time, assured Noyce that it would get easier with practice, but he did not enjoy thinking of himself as anything less than a highly competent pilot. A few days into classes, he told Lafferty, “I can’t do this; I’m not ready right now.” He wanted Lafferty, who was having a less difficult time, to finish the program and fly Noyce’s jet back to California. Noyce would hone his skills by flying the plane with Lafferty and would return to flight school only when he was confident he could perform exceptionally well there.

  After a few weeks training with Lafferty (who said Noyce “just wouldn’t give up”), Noyce decided his skills met his own standards and returned to the Cessna school to complete his formal instruction. This time he easily obtained the type rating he needed.

  In 1985, in yet another of his why-not-make-it-better moves, Noyce, along with Jim Lafferty and Mike Markkula, decided that Silicon Valley needed a top-of-the-line facility for private aircraft. The city-owned hangars had ten-year waiting lists. The trio joined forces with a group of 14 investors who bought a 15-acre property west of the San Jose airport and began building on it. The San Jose Jet Center opened in 1986 with 200,000 square feet of hangar and office space. Noyce served on the board of the Jet Center for several years, and the center’s founders can tell a story, very similar to the one at Caere, about his refusing to let the project fail. His message, recalls Lafferty, was always the same: “Go ahead and do it. It’ll be okay. And if it’s not okay, I’ll make it okay.”70

  12

  Public Startup

  In the summer of 1988, Noyce decided to leave his satisfyingly busy retirement and his beloved California to run SEMATECH, a semiconductor manufacturing consortium supported by the SIA and based in Austin, Texas. Noyce’s decision was curious, to say the least. Almost nothing about SEMATECH, which was jointly staffed and funded by 14 semiconductor companies, and that received $100 million annually from the Department of Defense, jibed with Noyce’s strengths or interests. He had spent his career refusing to take on Defense Department-sponsored research projects at his companies, but SEMATECH was its own sort of government research project. He was suspicious of all forms of bureaucracy, but it is hard to imagine a more bureaucratic arrangement than a consortium operated by more than a dozen companies and a hulking government agency. He took great joy in building enduring teams and cultures, but SEMATECH was designed to function with a revolving team of “assignees” who would be paid by individual corporations and work at SEMATECH for a period of only two years. In 1980, Noyce had testified before Congress that he had doubted whether semiconductor companies could ever cooperate on research—and cooperative research was the driving vision behind SEMATECH. “Innovation is not fostered by committee decisions,” he had said. “Pluralism is not an American tradition. … individualism is, and most [American semiconductor] companies have … an entrepreneurial history, and they believe they can do anything better than anybody else.” How did a man with such beliefs, at a stage in his life where he could have done anything, end up running a defense-sponsored behemoth like SEMATECH?1

  SEMATECH was conceived by the Semiconductor Research Corporation, an industry consortium that supports university research. Charlie Sporck championed the project and personally visited dozens of executives at SIA-member companies to argue that no matter how much legislation they pushed through Congress, United States firms would never be able to compete with the Japanese over the long-term unless American manufacturing processes were improved. Sporck originally envisioned SEMATECH as a multi-firm operation, funded to a significant degree by the Department of Defense, that would manufacture memory devices in such volume and with such efficiency as to “destroy the prices,” said Sporck, “really give the Japanese a hard time.”2

  This plan, largely modeled on Japan’s highly successful consortium efforts of the late 1970s, was short-lived. Such an arrangement would almost certainly violate antitrust laws; and even if it did not, SIA-member companies Texas Instruments and IBM continued to sell large volumes of memory devices throughout the 1980s, and they would not appreciate “destroyed prices” for their products.

  Replacing this mission for SEMATECH proved difficult. A series of SIA-sponsored meetings in Santa Clara in the summer 1987 made it clear that while everyone involved with the SIA wanted a SEMATECH-type organization to do something to improve American semiconductor manufacturing, no one knew precisely what that something should be.3

  In the simplest terms, three different constituencies wanted three different missions for the consortium. Eight of the ten major military contractors were likely SEMATECH members, and they believed SEMATECH should focus on developing “flexible” manufacturing lines that could produce a small batch of specialized chips for one military specification and then be quickly re-tooled via computers to produce a completely different chip to meet another military requirement.4

  At the other end of the spectrum were the large merchant semiconductor firms who eschewed a focus on flexible, specialized batch production in favor of honing the industry’s ability to mass produce large volumes of next-generation chips. To achieve this goal, companies such as Intel, Texas Instruments, and Motorola wanted to improve the equipment and materials that went into their fabs. These companies were each accustomed to spending more than a year and several hundred million dollars evaluating and qualifying the tools they used to build chips. They wanted SEMATECH to do this de-bugging.

  Yet a third group, made up of firms with smaller semiconductor operations (Hewlett-Packard, LSI Logic, and Micron, for example) wanted SEMATECH to focus not on the equipment used in the manufacturing process, but on the process itself. These companies usually lagged behind the big ones by a good year, so by the time they bought the equipment, the larger firms had effectively fine-tuned it for them. What these smaller companies wanted was a forum for sharing “black magic” manufacturing knowhow, a place to learn how other companies had made the process work best for them, and ideally, a “recipe” book they could take back to their own firms telling them exactly how to do it themselves.

  Each group worried that the other two were trying to get more out of SEMATECH than they put into it. Big operations said small ones were trying to avoid doing their own work. Small ones said big ones wanted to focus on advanced equipment improvements that would benefit only the big companies. Firms not centered around military contracts wondered aloud if the flexible manufacturing option was designed to benefit the Department of Defense more than the semiconductor industry. Everyone was concerned that their company’s sizable contribution—membership dues were set at a one-time $1 million entry fee, plus annual dues equal to either 1 percent of a company’s sales or $1 million, whichever was less—not be used to further a competitor’s agenda.

  The three competing visions for SEMATECH were never reconciled. Instead of a single unifying plan for the consortium, the brainstorming group produced a four-inch-thick “black book” that one person described as “a grandiose scheme to do everything that the member companies couldn’t do—but with a total investment of only $100 million per year.” The annual R&D budget at Intel, Motorola, or Texas Instrume
nts was roughly $400 million—and yet none of these companies had accomplished even a portion of the ambitious agenda outlined in the black book.5

  The black book also established a three-phased technical plan for SEMATECH. Each phase was defined by the width of the electrical paths on the semiconductor circuits. The number of circuits that can be traced on the surface of a semiconductor chip depends, in part, on the width of the circuit paths, which are measured in microns, one-millionths of a meter. (A human hair is about 75 microns in diameter.) The narrower the path, the greater the number of traceable circuits. The Phase I objective for the American chip industry was to build chips with 0.80-micron paths within three years, an easily attainable goal. The Phase II objective was to reduce this size to .50-micron lines by the second quarter of 1992. In Phase III, to be complete by the end of 1993, the geometries would be .35-micron—an ambitious target, since most projections held that companies working on their own would not be able to produce .35 micron chips (such as 64Mb DRAMs) before 1996. SEMATECH claimed that the three stages roughly corresponded to recovering global competitiveness, maintaining parity, and regaining global leadership for the United States semiconductor industry—but how SEMATECH would aid in the achievement of this neatly phased agenda was left unspecified.6

  The black book was crowned with a mission statement so vaguely worded that no one could object to it. “To provide the U.S. semiconductor industry the capability for world leadership in manufacturing.”7

  Meanwhile, the SIA’s platoon of lobbying executives began to push for federal government funding for SEMATECH, emphasizing that the ten companies already committed as members of the would-be consortium together accounted for some 80 percent of the American industry’s manufacturing base. SEMATECH would be a one-shot way to help an entire industry that would also be helping itself by providing half the money for the consortium. SEMATECH’s proponents, in fact, enjoyed describing the effort as “a high-tech barn raising.” As one of them explained, “You know, the farmer’s out there, and lightning hits the barn, and the neighbors from miles around come out and help him build a new one. And yet, then they turn right around and compete very heavily with each other as to who’s going to raise the best cotton crop or whatnot.” So it would be with SEMATECH, they promised.8

  Noyce did his part to promote the consortium, speaking to both the House and the Senate on the need for a vehicle through which “the industry will develop and test advanced manufacturing technology under realistic production conditions.” He pointed out that while the semiconductor industry already shared scientific and engineering knowledge—through published papers and technical conferences—no mechanism existed to share manufacturing knowledge. SEMATECH could fill that void.9

  A report issued by a Defense Science Board Task Force on Semiconductor Dependency in February 1987 boosted the SIA’s lobbying for SEMATECH. “The implications of the loss of semiconductor technology and manufacturing expertise, for our country in general and our national security in particular, are awesome indeed,” proclaimed the report. Task force members recommended that “cooperative government, industry, and university action” be taken immediately to ensure such a loss never occurred.10

  The contemporaneous and complementary efforts of the SIA and the Defense task force were no coincidence. Noyce served on the Defense task force advisory board, along with several senior executives from Texas Instruments and Motorola. Among the eight nongovernment members of the task force were the top two men from the SIA’s Semiconductor Research Corporation, as well as Jack Kilby. To be sure, these men possessed the expertise one would hope to see involved with a high-level government task force. Many of them also knew the SIA’s thoughts on the best solution for the industry’s ills, and these thoughts undoubtedly influenced the task force’s recommendations. On May 12, the Defense Department and representatives of the industry signed a memorandum of understanding that cleared the way for the funding process for SEMATECH to begin.11

  The Defense Department imprimatur meant a good deal in Reaganera Washington, but the SEMATECH proposal nonetheless faced strong opposition. A principal researcher for an interagency survey on the chip industry led by the National Science Foundation said that the semiconductor industry had “wildly overstated” its problems. The White House Science Council, chaired by the president’s science adviser, William Graham, received the SEMATECH proposal skeptically and ultimately recommended funding by only a one-vote margin. The Department of Commerce, in a generally favorable review of the “Benefits and Risks of Funding for SEMATECH,” raised questions about how the consortium’s results would be distributed and whether the operation ran the risk of collusion or of detrimentally centralizing the nation’s previously diverse research in semiconductors.12

  One angry member of the House Science and Technology Committee, who had recently seen California’s last major steel mill close in his district, all but accused the semiconductor industry of hypocrisy in his objections to SEMATECH. “Most of them are right-wing Republicans … basically against any government intrusion,” he said. “But when they get into trouble because they’ve been damn fools, then they come to us for bailout.” Certainly Noyce’s testimony before Congress to urge support for the consortium raised a few eyebrows. “His latest mission hardly evokes the industry’s history of rugged independence,” said a skeptical Washington Post of Noyce. “He’s in Washington pleading for a handout.”13

  Momentum was on the side of the SIA and SEMATECH, however. In its lobbying for the 1986 trade agreement with Japan, the SIA had made semiconductors a cause celebre in Washington. “The semiconductor industry has garnered enormous congressional support in the past few years,” explained Republican senator John McCain from Arizona in 1987. “There is a perception that they [sic] represent the high technology edge of America’s industrial competitiveness.” The SIA’s efforts were also aided by the organization’s refusal to commit to a specific location for the consortium before the funding vote in Congress.14

  In June of 1987, the House approved $500 million for SEMATECH in an omnibus trade bill. Four months later, the Senate unanimously approved an amendment to the 1988 Defense authorization bill in order to provide $100 million to SEMATECH in fiscal 1988 and 1989. With this authorization, funding for SEMATECH emerged very much along the lines that the SIA had imagined: a $250 million annual budget, with $100 million from the federal government, an equal amount provided by member companies, and the remaining $50 million from the local and state governments of the community in which the facility would be located. Incredible as it may seem, these sums were approved for an organization that still lacked a cohesive vision. Noyce nonetheless proclaimed SEMATECH a “bargain” for the American people—taken together, he pointed out, the consortium’s total annual federal funding would buy only half a B-1 bomber.15

  SEMATECH established temporary headquarters in Santa Clara and soon a site-selection committee issued a request for proposals for a permanent site. They expected perhaps a dozen responses but received proposals from some 135 locations in 34 states, all of them lured by SEMATECH’s plan to employ nearly 800 high-tech workers who would collaborate on research and work in the brand new, multimillion-dollar, state-of-the-art fab that SEMATECH planned to build in order to simulate the “realistic production conditions” to which Noyce referred in his Senate testimony.16

  The prospect of such a high-profile, high-tech anchor offered a singular appeal to the regions vying for the SEMATECH site. Almost from the birth of the transistor, cities and suburbs had tried to attract the glamorous electronics industry. In the 1960s, Frederick Terman, the Stanford provost who had offered to help William Shockley find suitable home for his company, consulted with interested regional developers in Texas and New Jersey eager to build electronics zones of their own. Charles Degaulle and members of the Japanese Diet had visited the Stanford Industrial Park. In the United States, the recession of the 1970s brought renewed interest in regional revitalization through high technolog
y, and by 1981, the National Governors Association had formed a “Task Force on Technological Innovation” to seek “improvements in the national tax code which would encourage industrial innovation” and to help states “fulfill their federalist role as supporters of economic development in their states.” One year later, the Congressional Joint Economic Committee issued a study that concluded, “High technology companies offer a brighter future for America but they [also] offer salvation for those regions of America that have borne the brunt of our economic decline.”17

  In the mid-1980s, regions across the country and around the world began trumpeting their high quality of life and low cost of living to traffic-and mortgage-weary technologists in Silicon Valley. “Remember the Silicon Valley as it was 20 Years Ago? That’s Albuquerque Today!” promised one representative advertisement in the San Jose Business Journal. By 1989, the United States boasted regions calling themselves Silicon Forest (Portland, Oregon), Silicon Gulch (Phoenix, Arizona), Bionic Valley (Salt Lake City), Silicon Valley East (Troy-Albany, New York), Silicon Prairie (Dallas, Austin), and Silicon Mountain (Colorado Springs). Several European and Asian countries also possessed technology regions named in homage to Silicon Valley.18

  Arguably the most successful of the American “Silicon Elsewheres,” as they were derisively called on the San Francisco Peninsula, was Austin, Texas. The high-technology economy in the Lone Star State’s capital emerged from the ashes of the domestic oil industry. Between 1982 and 1985, Austin added 10,000 manufacturing jobs to its economy—two-thirds of them in high-tech businesses. In this sun-drenched city, reported a somewhat concerned San Jose Mercury News in 1985, high-tech talk was becoming “as common as a tall Lone Star beer on a hot day.” Within a few years, several prominent Silicon Valley companies—Intel, ROLM, Tandem Computers, Advanced Micro Devices, National Semiconductor—had sites in or near Austin. Motorola and Data General had facilities in the area, too, as did both IBM (which owned a 3.8-million-square-foot plant in Austin) and Microelectronics and Computer Technology Corp. (MCC), one of the earliest electronics research consortiums. Austin could thus offer SEMATECH an established high-tech infrastructure as well as the attractions of the University of Texas and Texas A&M, low housing and labor costs, a vibrant cultural life, and no corporate or personal income taxes.19