But it was effective only to a degree. To cut through the interfirm suspicion and find a compromise among the three competing agendas, Noyce brought in a consultant who had worked closely with Intel. Noyce also implemented a mandatory communications training session based on a model called “Constructive Confrontation” that Andy Grove had developed at Intel.
Noyce leaned heavily on his own personal magnetism. The first time he met with a large group of SEMATECH employees, he started the meeting by removing his tie. When SEMATECH’s spending threatened to outpace its budget and Noyce approved the installation of a Budget Committee that nearly everyone at SEMATECH detested, he and his communications staff mapped out a plan to ease the tension. When Noyce mounted the stage at the next quarterly all-hands meeting, they plotted, the lights would go out. Noyce would then reach under the lectern, pull out a candle, light it, and ask his audience, all sitting in pitch darkness, “Did anyone clear this meeting with the Budget Committee?”42
They thought it would bring down the house. It did. But Noyce nonetheless devoted most of the meeting to fielding irate questions from engineers who wanted their requests approved now. His response in this situation was always a variation on the same theme. “Well, I understand why this would be frustrating,” he would say, “but we’re not going to change the policy. And here is why you should be glad we’re watching expenses.” Noyce’s sympathetic approach usually left the complainers happier than before they had vented their frustrations, but serving as a roadblock slowing the pace of work must have frustrated Noyce as much as it irritated the engineers. Noyce, however, had no choice. If SEMATECH overran its budget in a given quarter, it would look bad enough in Washington to jeopardize future appropriations.43
At the end of this meeting, in response to a question from an employee complaining that his job title did not accurately describe the high level of his responsibilities at SEMATECH, Noyce said that he had never put much stock in titles himself, but if the questioner—or anyone else—wanted to give himself a new job title, that was fine with Noyce. Within days, several employees had filed requests for business cards—which they promptly received. In much the same vein was Noyce’s insistence, throughout his tenure at SEMATECH, that the consortium did not need an organization chart. He thought such documents did little more than encourage undue attention to hierarchy. His COO, who believed otherwise, secretly drew up an organization chart for his own reference.44
But neither Noyce’s charisma nor his attempts at organizational innovation could lead the member companies to “make beautiful music together.” The firms were willing to concede that they had similar problems in their manufacturing processes, but they did not want to go beyond that. This meant that SEMATECH’s mission would not involve sharing black magic (as the smallest companies had hoped) or developing flexible production methods (as the defense interests had desired).
Instead, SEMATECH narrowed its agenda to focus on issues of greatest concern to the consortium’s largest member companies: improving the equipment and materials that they used in their fabs. SEMATECH would identify the areas most critical for next-generation progress and fund contracts for suppliers to develop new tools and materials. The consortium would also work with suppliers to improve existing tools and equipment. Or as Noyce put it, “From our base of shared knowledge, we will define the needs and specifications for the advanced tool set and then … invest in joint R&D projects with U.S. equipment and materials suppliers to develop the advanced tool set.” Once the new equipment was developed, Noyce said, SEMATECH would test, characterize, and demonstrate it to member companies so they could incorporate it into their own fabs.45
Forcing the consortium to pinpoint a mission—even a mission born more of a process of elimination than universal acclamation—was Noyce’s single most important contribution to internal affairs at SEMATECH. Before he arrived, no one at SEMATECH could agree on what the consortium would do. A few months into Noyce’s tenure, the mission was defined.
A focus on improving fab materials and equipment meant that SEMATECH would need to transform itself from a “horizontal” research collaboration among peers to what Noyce called a “‘virtual’ vertical integration” between its members and the suppliers from whom they bought equipment and materials. In many ways this made SEMATECH even more like a Japanese research consortium—roughly 80 percent of which were vertically integrated—than had originally been imagined.46
Once the mission was defined, Noyce began speaking of SEMATECH as an effort to “strengthen the infrastructure” that supported the semiconductor manufacturing business. In this sense, the move to reach out to suppliers can also be understood as a $250 million attempt to reproduce the close relationships that had nurtured the earliest silicon companies in Silicon Valley. SEMATECH’s new mission hearkened back to the days when asking for advice felt like borrowing a cup of sugar and the guy who ran the company that built your furnace used to work for you.
But those harmonious relationships had long since disappeared. The supplier and manufacturing businesses had grown in very different directions. The supplier industry was only one-quarter the size of its manufacturing counterpart. Revenues for 1989, for example, were $5 billion and $20 billion, respectively. And in contrast to the manufacturing side, which was dominated by a few large companies such as Intel and Texas Instruments, 88 percent of the more than 800 American supplier companies were small businesses with annual sales of less than $25 million.47
By the time the manufacturing firms that belonged to SEMATECH decided to focus the consortium’s attention on shoring up relations with suppliers, it seemed that most communications between the two industries were being conducted by lawyers. Suppliers believed that manufacturers said they wanted quality but really made decisions based on short-term costs. The suppliers also felt handicapped by the manufacturers’ acute concern for intellectual property, which made it impossible to develop industry-wide equipment standards. Moreover, because most supplier companies were too small to build state-of-the-art fabs to test their equipment in a realistic mass-production environment, they had to de-bug on a customer’s line—and then suffer castigation for “poor quality products” from manufacturers who thought the equipment should have been perfected earlier and who at the same time refused to share data on how the equipment performed.48
Manufacturers, of course, felt differently. As one long-time IBM employee explained, “Much of the very expensive equipment purchased by the chipmakers didn’t work according to specifications when it got to their factories. This was particularly true with the latest models that were purchased for a new factory. The chip maker could have upwards of a billion dollars tied up in a factory that didn’t work. It frequently took eighteen to twenty months to get up and running, [during which] time the chipmaker was losing his proverbial rear end and was also losing market share. Naturally all of the manufacturing middle management was shifting all the blame to the equipment makers [and] top management accepted this input.”49
One person who attended the early talks between SEMATECH and the suppliers’ trade organization SEMI said that he had attended friendlier lawsuits. Another participant recalls a representative from one organization referring to the other group’s members as “thieves”—and then receiving an ovation for his candor.50
THE MOVE TO FOCUS on suppliers irritated the Department of Defense. Improved basic manufacturing—SEMATECH’s original stated mission—offered a direct benefit to Defense efforts, because it increased the likelihood that American weapons could be built with American electronics. But bolstering suppliers seemed to many in Washington to be more clearly beneficial to industry than to the national defense. In late 1990, the head of DARPA came to SEMATECH, and during a meeting with the CEO, slowly and deliberately wrote D-A-R-P-A on the white board. Then he announced, “This stands for ‘Defense Advanced Research Projects Agency.’ SEMATECH is none of these things. It’s not defense-related, it isn’t advanced research, and it’s not being run
as a project. You don’t fit in our program.”51
By early 1989, Noyce was already feeling the Defense Department’s disapproval of the new mission. He was also finding it increasingly difficult to work with Paul Castrucci, his COO. Castrucci had done an unassailable job overseeing the construction of SEMATECH’s fab, which at $75 million and 32 weeks’ building time, had set records in an industry that spent, on average, $200 million and 18 to 24 months to build a state-of-the-art fab. Shortly after the fab’s completion, SEMATECH provided detailed engineering, business, and manufacturing design plans to member companies and the government—the first “technology transfer” at the consortium and an auspicious beginning in the eyes of several member companies.52
But as time progressed, Castrucci’s top-down managerial style proved incompatible with Noyce’s approach and SEMATECH’s new mission. In March 1989, after several weeks of agonizing indecision, Noyce asked for Castrucci’s resignation. He named Turner Hasty, who had served as interim operations chief prior to Noyce’s arrival, as COO. Noyce and Castrucci agreed to keep their counsel and say nothing more about the resignation after the initial announcement, but a few months thereafter Noyce, in a move that won him a plush muzzled eagle doll (representing himself) to keep on his desk, added a few details. The problems arose not over policy, but over what Noyce saw as “drifting” within the organization. He came within a breath of calling internal operations at SEMATECH the organization’s “weak link.” Noyce was spending nearly 40 hours per week on these internal operations and finding them so frustrating that he told several people that he suspected the consortium might run more smoothly with fewer member companies and interested government parties.53
Meanwhile, Noyce continued to serve on the Intel board—he tried not to miss a meeting—and to deliver occasional speeches as a representative of the company. He also had to fulfill the ongoing requirements of the full-time “Mr. Outside” work for which he had been hired. He spoke to at least one group nearly every week; in his first ten months at SEMATECH, he gave a speech or testified 58 times.54
Noyce was himself somewhat to blame for his packed speaking schedule. The communications team would decline an interview request, recalled Miller Bonner, but “whoever we declined would go in the back door and get Noyce on the phone. And he’d say, ‘Well, okay. We can make that happen’—because he didn’t want to disappoint people.” The problem abated somewhat after Bonner took Noyce aside and said, “Bob, we’ve really got to focus this. I mean, you just can’t say yes to everybody.”55
Noyce was not simply being nice when he agreed to speak. He believed, as did Bonner, that “we [SEMATECH] need publicity; we need support of many, many masters. You’ve got the local, state, and federal governments. You’ve got the supplier industry. You’ve got the downstream [users of semiconductors]. You’ve got the members [of SEMATECH]. And you’ve got the non-members.”56
Noyce added, “We’ve got to have someone very obviously, very soundly behind anything we say. I think that we need to try to express the objectives and the philosophy of why we are doing this. We need to try to articulate that pretty clearly so people can subscribe to it.” Noyce, of course, was the “someone” who needed to ensure the support of SEMATECH’s “many masters,” particularly in the nation’s capital. Between September 1989 and March of 1990, he flew to Washington, D.C. almost every month for two- or three-day visits that usually included congressional testimony, visits to policymakers and staffers, strategy sessions with the Dewey, Ballantine attorneys who had also worked with the SIA, interviews with the press, and working breakfasts, lunches, and dinners. It was not unusual for him to have a dozen different meetings on his agenda each day he was in Washington. In the regular returns to the East Coast and the need to garner support from outsiders mostly unfamiliar with semiconductors, Noyce was re-living some of his least favorite aspects of the general manager’s job at Fairchild.
Although Miller Bonner, like everyone else who knew Noyce well and watched him speak, thought Noyce seemed a little nervous at the podium, by this point Noyce possessed an extremely savvy public presence. A rare full-text transcription of a lengthy television interview that Noyce knew would be edited shows him interrupting himself midsentence—“that wasn’t very good,” he says—and then rephrasing the comment in a more pithy way. Noyce similarly edited himself early in his tenure at SEMATECH when he told the Intel employee interviewing him that the CEO’s job was more enjoyable than he had imagined it would be. Almost immediately, he added, “I suppose I shouldn’t be quoted that way.”
Miller Bonner served as Noyce’s speechwriter for much of his SEMATECH tenure. They used the same modular speech-building structure that had worked well for Noyce at Intel. Often the pair would assemble Noyce’s presentations in the air en route from Austin to the location where he would speak, Noyce sitting in the cockpit with his pilot, shouting comments and ideas back to Bonner, who would key them into his portable computer.
Noyce always arrived at the podium with every word of his speech written out, but the text served more as a starting point than a script. Indeed, Noyce extemporized so much that Bonner, who regularly received requests from journalists for copies of Noyce’s talks, had the following statement printed across the cover page of his speeches: “Dr. Noyce may deviate from the attached text, but he stands by the text as written.”
NOYCE DID MANAGE to work in a bit of fun. He came to a SEMATECH Halloween party dressed as Bruce Springsteen, complete with a jet-black wig, a red bandana, and a t-shirt that read THE BOSS. He installed a “remote car starter” in the Mustang convertible he had bought when he moved to Texas. Whenever he spied from his office window someone near his car in the parking lot, he loved to push a button on his key-chain transmitter and watch the reaction when his empty car seemed to start itself. He also took a break from SEMATECH to return to Grinnell for his fortieth college reunion.57
Another enjoyable diversion came in September 1989, when Noyce was surprised to learn that the afternoon-long meeting on his calendar was really a trip to nearby Bergstrom Air Force Base, where members of the 67th Tactical Reconnaissance Wing helped him to suit up and then took him to fly an RF-4C jet. A photo of Noyce in the rear cockpit of the plane shows him giving an enthusiastic thumbs-up before take off. “He was like a kid in a candy store,” recalled one person who accompanied him to the base.58
Jim Lafferty, now running the San Jose Jet Center, provided Noyce his favorite distraction when he put him in touch with an entrepreneur named Wayne Higashi who, along with inventor Paul Pires, had developed a mechanical transmission that they believed could reduce fuel consumption in many cars by as much as 20 percent. Noyce invited Higashi to Austin to demonstrate a prototype of the transmission, which was about the size of a shoe box and powered by a half-horsepower motor.
After about two hours with Higashi, Noyce said, “I don’t think it has a uniform output.” Without a uniform output, the speed of whatever device the transmission turned—automobile wheels, for example—could not be constant. They would rotate faster at some times and slower at others.
Higashi explained that he and Pires had developed a mathematical model of the transmission that proved its output was uniform. Any variation Noyce thought he saw must be the result of non-uniform input from the motor.
Noyce remained unconvinced. “I don’t think it works,” he said. He then made the move that had endeared him to so many people who worked with him. “It doesn’t work, but it looks like a great idea,” he said. “Leave your drawings and business plan with me, and I’ll think about it.” He then proceeded to spend every available minute of the next several weeks in his workshop, musing on the transmission—often until well past midnight. “With the two opposing shafts geared together, the system is equivalent to a simple crank arm system,” he wrote, sketching as we went along. “This still gives basically a sinusoidal output, slightly modified by arm length, etc. The basic case is d/a>>1, l>>a. … But try as you may, this won’t give a
linear output (i.e. constant angular velocity)!” He then proceeded to fill several pages with equations, graphs, and line drawings, jotting notes all the while—“must have clutches [sic] efficiency!”—and clearly having what he told Ann Bowers was “the most fun he’d had since he had been in college working on physics.”59
In short order Noyce proved mathematically that the transmission’s output could not be uniform. When he showed this proof to Higashi on his next trip to Silicon Valley (where Highashi worked), the entrepreneur was skeptical. Noyce suggested, very pleasantly, that Higashi start the transmission. Noyce then reached into his pocket, extracted a credit card from his wallet, and held the card against the output drive gear on the transmission. The frequency oscillation of the clickety-clack of the card hitting the gear tooth was clearly not uniform. Instead of hearing a regular k-k-k-k-k-k-k-k-k-k sound, the men heard something like k-k-kkk-k-k-kkk. It was not random, but it was variable. “It amazed me,” Higashi recalls. “Noyce developed his own complex mathematical model—and proved us wrong—just by looking at our drawings and observing our model.”60
Noyce showed Higashi a second set of calculations from his basement research. He had mathematically modeled the non-uniformity of the transmission and had some ideas about how the entrepreneurs could use it to their advantage. It might work in a diesel vehicle, which has an engine with an output even more variable than that of the transmission. Or how about installing the transmission on a bicycle to make it easier to pedal more efficiently?
The Man Behind the Microchip Page 45