The group that would eventually make Santa Clara, CA, "Silicon Valley."
In discussing the development of semiconductors in Silicon Valley, many roads originate with Arnold Beckman, the man who hired William Shockley away from Bell Labs and brought him to the San Francisco Bay area to establish the Shockley Semiconductor Labs of Beckman Instruments (now Beckman Coulter). Semiconductors had been around for several decades - odd materials that had the ability to conduct electricity under certain conditions. They are, Gordon Moore says, "halfway between insulators and metals. The wonderful thing about semiconductors is that you can control the amount of their connectivity through introducing impurities."
Bell Labs, as the research arm for the Bell companies - owners of millions of electromechanical relays across the nation used in its telephone switching networks - was one of the most interested parties in semiconductors and their ability to conduct electrical currents. Bell executives had the foresight to hope that, one day, Bell could replace its troublesome relays with more reliable devices made of semiconductors. Bell had also laid underwater cables that used vacuum tubes for repeaters at regular intervals, making the cables unreliable. So Bell funded a semiconductor laboratory in Murray Hill, N.J., which is where Shockley, Walter Brattain, and John Bardeen produced the work that received the Nobel Prize in 1956 "for their researches on semiconductors and their discovery of the transistor effect."
As the head of one of the country's leading scientific companies, Beckman understood the importance of semiconductors. Hence, he hired the greatest name in the industry to establish his company in that field, setting Shockley up in his own research and development facility. Although Beckman Instruments was based in Los Angeles, Shockley's new labs were set up near Palo Alto, CA, because Shockley's mother lived there.
The Shockley Semiconductor Labs were short-lived, however. With the exception of Robert Noyce, none of the key engineers working there could stomach Shockley for long, despite his unquestionable technical brilliance. In 1957, the labs' senior members selected Gordon Moore to contact Beckman and have Shockley moved aside as leader of the labs. When that didn't work, seven men - and eventually Noyce, making it eight - left the company. Moore still remembers the night he drove to Jean Hoerni's house to make the call to Beckman and also that it did little good.
In the end, Shockley may be most remembered for hiring the talented group and, some say, subsequently driving them away to join Fairchild Semiconductor. Moore, one of the "Traitorous Eight" who left to form Fairchild's semiconductor operation and eventually become one of Intel's founders, remembers the evening he was sitting at home in Maryland, when the phone rang, and the voice on the other end said, "This is Shockley." That's about all I remember about the call, but I took the job he offered. I had been doing pretty much esoteric work at Johns Hopkins University, looking at the spectroscopy of metals. I didn't know the first thing about semiconductors, but Shockley thought he needed a chemist. None of us knew his reputation as a manager at that time, but maybe we should have suspected, as none of his guys from Bell Labs were joining him in California." The son of a local policeman in the small coastal town of Pescadero, CA - directly west of what would become Silicon Valley - Moore wanted to return to California. Shockley's reputation and the incredible wage being offered, $750 a month, were all it took to bring Moore home, where he became one of the formative members of the group that would eventually make Santa Clara, CA, "Silicon Valley."
By the time Moore was on his way to California, commercial semiconductor manufacturing was underway in Boston, Phoenix, Dallas, and New Jersey. The establishment of the Shockley Semiconductor Labs was the first step toward adding the Bay area to that list. Many books have been written about the semiconductor industry's founding, with Charlie Sporck's book excerpted in our August issue as one of the few firsthand accounts (see "The Birth of Fairchild Semiconductor," August 2001, page 60). Trying to capture the semiconductor story for this issue of Upside meant tracking down the people who created the semiconductor industry - a nerve-wracking process. Noyce, eulogized in Upside (July 1990), was one of the first of his generation to pass away. But many of the other larger-than-life people from Fairchild Semiconductor - Eugene Kleiner, Jay Last, Pierre Lamond, Julius Blank, Andrew Grove, and Moore - are still around to provide insight into what it was like when the semiconductor industry was first being created. For this special issue, I particularly wanted to answer the question of whether such a technological discontinuity as the semiconductor revolution will ever appear again. The answer lies in a thorough understanding of how the Valley became "siliconized."
Just as Shockley knew the labs would need chemists, he knew that mechanical engineers would be required, so he hired two: Blank and Kleiner. Blank was a classic engineer and had worked at Babcock & Wilcox, where he designed and built the huge boilers used in power plants and utility companies. As a boy, Blank attended a technical high school in Brooklyn, where he learned the craft of building things. In 1943, the U.S. Army grabbed the young man, sent him to college, had him repair military aircraft, and then sent him to Europe to fight in World War II battles such as the Battle of Hurtgen Forest. By the time Blank returned to the States in 1946 to finish his bachelor's degree in mechanical engineering, he already had a lifetime's worth of practical experience. Then it was off to Babcock & Wilcox, Goodyear Aircraft, and, finally, Western Electric, which set him to work with germanium phototransistors, among other devices, to figure out how to replace its mechanical relays.
In 1956, Dean Knapic, a Western Electric alumnus, offered Blank a job at Shockley Semiconductor Labs. After traveling to Palo Alto to be interviewed by Shockley, Blank, like all the other original employees, underwent lengthy psychological testing - possibly an artifact of Arnold Beckman's experience in one of his firm's plants in Los Angeles, where an employee went berserk and stabbed a co-worker to death, or a result of Shockley's unorthodox views on personnel practices. Despite the days-long grilling, $10,000 a year plus moving expenses looked pretty good.
Blank and Kleiner shared all of the mechanical-engineering work at the company, which was housed in a small stucco building at 391 South Charleston Road in Mountain View, CA (now a chair shop that bears an incongruous brass plaque identifying it as the birthplace of Silicon Valley).
Blank's first assignment was to build a crystal grower. Blank knew nothing about semiconductors. Fortunately, Leo Valdez and Victor Jones, hired by Shockley to actually grow the crystals, shared what they knew about the type of equipment they needed, and Blank went to work. "That's what it was like then. Bobby Noyce would walk in and say, 'I want you to melt some copper on this part.' Really vague instructions were the order of the day. I would do that, and then take it in to him. He'd look at it and tell me how to change it, or make some other suggestions, and we would go back and forth like that, making things up," Blank recalls.
Despite the lack of direction, Blank loved the work. After he joined Shockley Semiconductor Labs, Noyce, Moore, Last, Kleiner, and Hoerni appeared in short order. Blank remembers the group as a bunch of 20-somethings who liked to hang out together and see each other socially. He remembers the entire year and a half at Shockley Semiconductor Labs as an exciting time, ordering power upgrades, phone systems, air conditioners, and the radio-frequency (RF) oscillators needed to melt silicon. An indication of how Blank was regarded by his colleagues is the fact that Blank was nominated to recontact Beckman, after Moore's first attempt, about removing Shockley. "At first, it appeared I was successful," Blank says. Beckman endowed Shockley with a teaching chair at Stanford University that kept the good doctor out of the men's hair. Teaching, in addition to Shockley's speaking and travel demands after winning the Nobel Prize, initially seemed to have solved the issue. After a while, however, with Shockley coming back from trips and ordering entire projects restarted, it was clear that the problems would not be resolved. There was also Shockley's single-minded pursuit of the four-layer diode, perhaps left over from his days at Bell Labs, while many of the others thought silicon transistors were the direction they should be headed in.
Eugene Kleiner - co-founder, many years later, of the venture capital firm Kleiner Perkins - was another of Shockley's early hires. Kleiner was an immigrant from Europe. After leaving Vienna, Austria, he attended secondary school in New York. He then took work as a factory machinist, but, like so many of the remarkable men in this issue, he was drafted into the military. After the war, Kleiner earned a bachelor's degree from the Polytechnic University of New York and a master's in mechanical engineering from New York University. He taught engineering for a short while and then joined Western Electric, where he worked in the morass of the Bell system's relays and switches. Kleiner remembers Shockley as a charming person, a fascinating conversationalist, and, by reputation, one of the stars of Bell Labs.
Reconciling that impression of Shockley with the small, inadequate, and dirty building that Shockley had leased to start the company was the first of many events informing Kleiner that Shockley's technical brilliance was not matched by practical experience. Like Blank, Kleiner's first assignment was to build a crystal grower. It was this experience that brought Shockley down to earth in Kleiner's eyes. "I didn't know anything about growing crystals and [knew] nothing about furnaces. So I asked Shockley, but he didn't know. He just gave me advice, often the wrong advice, so our first device for growing crystals was a monstrosity. It didn't work. It was so large that we had to raise the roof of the building. It never worked. So I went next door to Sears, Roebuck and Co. and bought one of those large standing drill presses, and it had most of the features we needed. We had to change some things and add some things, but it formed the basis of our second and successful, crystal grower."
Then it was time to build a furnace, which entailed a similarly unsuccessful set of experiences. Still, "working with that team of men, Moore figured out the dispersal of gaseous materials," Kleiner recalls. (Moore was an expert glassblower. He actually created, by hand, the tube jungles in which gases were distributed, separated, and combined, thus allowing for single-step production of doubly diffused transistors.) "Knapic. Blank. It was a beautiful team. It was exciting," Kleiner continues. Once built, the furnace had to be kept on 24 hours a day. Kleiner and Blank, who lived close to the company, came in every couple of hours during weekends to check on the relatively delicate device. Kleiner remembers, 40 years later, that it was his wife who kept him going during those times (she also sang madrigals with Noyce at parties and social events). More importantly, she wrote a letter to Hayden, Stone & Co. seeking money for several of the "Fairchild Eight" - the letter that found its way to Arthur Rock and convinced Rock to find funding for Fairchild Semiconductor and, later, Teledyne Semiconductor and Intel. It was "one of the great letters of all time," Kleiner muses, thinking of that lovely unsung heroine.
Jay Last, now retired and living in Los Angeles, left Silicon Valley early on to become a vice president at Teledyne, where he helped Henry Singleton build that company from a single division to over 150 divisions. Last, a quiet electronics and optical engineer, worked closely with Noyce, and is responsible for much of the early manufacturing infrastructure of semiconductor technology. Surrounded by his collection of African art and the products of his publishing company, Last is, in many ways, uncomfortable sharing his life with strangers. His years of military research and development have remained a closed book long after his retirement. And the subject of Shockley is one that, despite the passage of 40 years, is not casually remembered.
Still, if you catch his attention, Last will explain that much of what helped build the semiconductor industry came out of Fairchild Semiconductor: improvements in growing silicon ingots and diffusing exotic materials across the substrate of silicon wafers. Last's own contributions from optical science - the creation of photographic masks that were used to expose patterns on the substrates and the development of etching processes through which lines and connections were created on the wafers in miniature - are tossed off casually. "None of that had been done. We were inventing everything," Last says. This includes, most notably, adapting these processes to wafer fabrication and manufacturing. Waving his hands, Last dismisses the greatest revolution in manufacturing since Henry Ford's achievements, but it was Last's special contribution: "the creation of step-and-repeat methods by ganging up these microscopes and devices on mechanically operated stages, so that we could create integrated circuits cheaply, quickly, and reliably." He holds out one of the first commercially integrated circuits, with its five gold leads, showing me the four-transistor device with a sense of wonderment, even after all these years.
"And this," he says, retrieving yet another device from a trove of first technologies. I notice a tiny "open here" legend printed on its side. Last smiles conspiratorially and says, "We knew that the people at TI were desperate to see what we were doing, and we knew that, as soon as we released these, they would get some, so I had these printed up for them." He grins at his one-upmanship with boyish delight.
He won't say it, but those processes, step and repeat, led to "batch manufacturing," which spread out of semiconductor manufacturing and into other technological fields, such as LED manufacturing, biotechnology, and, experimentally, nanotechnology, fostering a worldwide change in manufacturing technologies that will continue throughout the 21st century. Last is sanguine, comfortable with his accomplishments and a hoard of unmentioned patents. I sit pondering this guarded, gentle, and brilliant man who tells me that, despite the passage of four decades, he can walk into a modern-day wafer-fabrication facility and still know what the machines do, which is basically what he and his colleagues first laid down in the 1950s. Last thinks back on Noyce, his closest collaborator at Fairchild Semiconductor, with a fondness that is touching - an almost universal emotion among the giants still walking the Earth who helped to create Silicon Valley.
While the men were in a quandary over their problems with Shockley, there were other companies working with semiconductors. Boston-based Transitron Electronic was a contemporary of the Shockley Semiconductor Labs. Transitron was found by an eccentric pair of brothers named David and Leo Bakalar. One of the pair ran a shoe factory and was cheaper than blue jeans; the other was a Ph.D. who had worked at Bell Labs on transistor research. They recruited engineers to their firm by traveling through Europe and conducting interviews in major cities. They hired the brightest immigrants they could find, brought them to the United States, and worked them silly in the early days of discrete electronics components. Pierre Lamond, Wilfred Corrigan, Robert Swanson, Lester Hogan, and a young Hungarian immigrant named Leslie Vadasz passed through Transitron's doors and built a semiconductor company that vied with Texas Instruments for the number one position in the nation in the late 1950s. At that time, semiconductors like germanium were employed to produce devices such as rectifiers in small aluminum canisters, which electronics manufacturers used by the thousands.
Pierre Lamond, a jeweler's son who fought in the Algerian War and made contacts with American officers during a stint at NATO headquarters, wanted to find a way to work in the United States. It was clear that his field (physics and electron optics) was moving fastest in America. In 1957, a decommissioned Lamond saw an employment ad in the New York Times for Transitron. David Bakalar offered him a job on one of his recruiting trips through Paris. To get Lamond started, the Bakalars put him on the production line. In his third week, he was promoted to the head of production to replace his departing boss, and, in a few more months, he was promoted to device development engineer. It was a whirlwind introduction to semiconductor technology, and, for the one-year duration of his working visa, he absorbed everything that he could before returning to France. Lamond shopped for a company in his native country where he could apply his skills, but he found nothing comparable to his experience in Boston, so, by 1959, he was back at Transitron as the head of development.
Lamond's stint at Transitron was a short one, as most were. The Bakalars' approach was to "second source" parts designed by others (a polite way to describe their energetic copying of other people's products), which they would then sell to customers at cheaper prices. Lamond had heard of Hoerni's work at Fairchild Semiconductor in developing what would become known as the planar process of manufacturing semiconductor products. Lamond approached the Bakalars and told them that, if they were to compete, it was time for Transitron to invest in planar manufacturing, but they would have nothing of it, telling him to wait a few months until all the bugs were worked out of the process. Unfortunately for the Bakalars, the energetic Lamond had befriended Moore while presenting a paper at a technical conference, so the brothers' refusal to move Transitron along the path to original-product development came at a time when Fairchild Semiconductor's human resources manager was already courting Lamond.
In 1961, Lamond joined Fairchild Semiconductor, working for Moore. After one of the periodic diasporas of Fairchild's engineers to form a new company - in this case Signetics - and the loss of Lamond's direct supervisor, he was again promoted to lead device development. He remained at Fairchild from 1961 until 1967, when he became a member of another group leaving Fairchild, under Charlie Sporck, to restart National Semiconductor. Lamond remained with National until 1974.
Many of the Transitron staff passed through Phoenix (Motorola) or Dallas (Texas Instruments) before coming to Silicon Valley. Vadasz spent his three years of apprenticeship at Transitron, for $650 a month, before moving straight to Silicon Valley, where the man who had hired him into Transistron, Lamond, hired him again at Fairchild. As with others at Transitron, Vadasz found himself building whatever he needed. Need vacuum chambers to distribute evaporated dopants? Build them yourself, out of a piece of tube. Need some silicon wafers? Build your own furnace; dump in chunks of silicon lumps; start it spinning; dip in a piece of seed crystal, rotating it in opposite directions; pull it out slowly; and hope for the best. Everything else pretty much required working under a microscope with instruments held in your own shaking hands. Everybody who passed through the Boston sweatshop seems to have come out the better for having been there, and for having left.
For example, Wilfred Corrigan remembers someone from Transitron coming into a room, where he was reading about semiconductors, and telling him that the production line was a mess: "They told me to get out there and get it running right, so I went, and I figured out semiconductors later." Transitron was that kind of place.
Fairchild Semiconductor was, by comparison, a fortress of strong organization. Imagine coming into a company where at least a dozen future semiconductor company presidents and founders are working. At the head of the organization is Sherman Fairchild, founder of Fairchild Camera and Instrument, scion of one of the first investors in IBM, and head of a company with technology interests in multiple industries. A large part of the comparative stability at Fairchild Semiconductor was based upon its hiring of Ed Baldwin as general manager of its new semiconductor division. Baldwin recalls, "I came up from Los Angeles, and they interviewed me - all of the them, Noyce, Moore, [Victor] Grinich, and Last - but it was Moore that made the decision to hire me."
Their choice was for someone as eclectic as themselves. Baldwin's father was an itinerant intellectual: a musician, engineer, and writer. His father was related to a former prime minister of England, where his family hailed from, and the founder of a technical university that still exists in London today. Baldwin earned one of Carnegie Mellon University's first master's degrees in solid-state physics and then a Ph.D. in nuclear physics, also from Carnegie Mellon, where he worked in neutron and proton scattering. Baldwin also built Carnegie Mellon's synchro-cyclotron, one of the progeny devices of the Berkeley cyclotron.
In 1950, the young scientist was recruited by Hughes Semiconductors, of Los Angeles, during that company's heyday. There, Baldwin worked with people such as Simon Ramo and Dean Woolridge, two of the founders of TRW, on the miniature diodes Hughes produced for airborne computers. By 1956, Baldwin had risen to the head of product development in Hughes' computer division, and, when that was folded, he became head of product development for the semiconductor division. While at Hughes, Baldwin wrote his first two patents for the company. The first patent was on semiconductor photodiodes, and the second patent was for a device that would use the diodes to read hundreds of mainframe punch cards per minute. The company promptly gave him a $100 bonus, patted him on the head, and sold his division to RCA for a small fortune. At that point, Baldwin was ready for a career change.
Baldwin remembers Hughes as a great training ground where he learned about operations, plant maintenance, engineering, manufacturing, personnel, and research and development. At Hughes, Baldwin also learned how to work with brilliant technologists and how to act as a professional manager. Nevertheless, an ad in the Wall Street Journal, seeking an executive and a general manager for a startup in Northern California, caught his eye. Baldwin had been looking for new work for a while, and, along the way, he'd befriended a banker who knew a source of capital for Baldwin's possible own company. But Baldwin hadn't heard from the banker in months, and the offer to work with the likes of Hoerni and Moore proved irresistible.
Baldwin had been at Fairchild Semiconductor for several months when the banker turned up, having secured a commitment from a wealthy industrial-manufacturing family by the name of Rheem. The banker was ready to press a check for $5 million into Baldwin's hands. "It was the worst decision of my life. If I hadn't made [that decision] I would probably be at Intel today," he says, looking wistful, and then he smiles and starts telling the story of a lifetime's worth of companies he has started and run. Baldwin reminds me that, at one point, an engineer who left Fairchild Semiconductor to join him at Rheem Semiconductor took one of Fairchild's precious "recipe books" - a source of considerable disgruntlement to this day. Baldwin recalls, "When I found out about it, I gave him the living daylights, and I returned the book." Today, at a spry 82 years of age, Baldwin is running a startup in San Diego, Energy Development Systems, where he will be working in his field "until the day I die," having just earned the most recent of several patents on the high-power capacitors that fuel his dreams for tomorrow.
While Baldwin was departing to found Rheem Semiconductor, Vadasz was in the midst of four years of work putting down deep roots into the development of bipolar, and later MOS, semiconductor products. But Vadasz was also destined to leave Fairchild Semiconductor, and was hired by Noyce and Moore to work in research at Intel, the company Noyce and Moore founded with the intent of designing and building semiconductor memories. There was a growing market for semiconductor components, but Noyce and Moore felt that, if they could build solid-state, or thin-film, memories that would replace expensive and hard-to-manufacture solid-core memories, they would have ready customers among mainframe computer companies. It was a dubious market strategy at best, fraught with tough competition and only occasionally won by Intel. U.S. manufacturers ultimately abandoned the strategy and left memory making to Asian manufacturers.
Despite a well-published error having to do with a numbering scheme introduced much later, Noyce and Moore hired Grove and then Vadasz as their first Intel employees. Both Vadasz and Grove remain at Intel today. Ask Vadasz what he did in the beginning at Intel, and he'll explain that he worked on the technologies that led Intel into MOS manufacturing, the technique that is standard today. He goes on to admit that Intel's first successful memory products were good old-fashioned bipolar ones, like those of Fairchild Semiconductor. Press him for some details on the success he has brought Intel, and he'll tell you that it was Moore and others experimenting in the labs with MOS who finally figured out why the yields in the new technology were so low for Intel and made the process practical.
Out in the dry plains of Texas, Texas Instruments (TI) had been around since 1933, under another name, supporting the scientists working in the oil industry of Texas and beyond. A Texan giant of a man, Jack Kilby, had returned from Illinois to his home state with a bachelor's degree in electrical engineering to work with the resistors, capacitors, and discrete-logic products of his time - basically to build the instruments that TI manufactured. The Iliac was just being built when Kilby graduated from college, but computing was firmly on the horizon, and Kilby increasingly turned his attention to germanium and then silicon, spending a lifetime in the semiconductor field, with a single company. Kilby can sit and preach the semiconductor bible from memory, because, at TI, he did what all of his competitors were doing elsewhere, slowly experimenting with semiconductor materials. He developed the design and manufacturing processes that allowed TI to "get there first," making TI the undisputed leader in transistor, and then semiconductor manufacturing.
There's a hard edge in the 90-year-old's clear blue eyes when you ask him about the industry back then. "We didn't share anything in those days. We were all competitors, in everything. We went to the ISSCC [International Solid State Circuits Conference] and announced our progress each year, but we said as little as possible beyond what we had done, and we certainly didn't tell anyone about our processes." Kilby, recently awarded a Nobel Prize for the invention of the integrated circuit at TI, doesn't talk much about his or TI's primacy in the field; he seems proud just to have been a part of the efforts in building one of the first multitransistor semiconductor microprocessors - a six-transistor part that TI built for use in products such as portable calculators and the Minuteman program. Kilby knows all the names from the Valley - Moore, Sporck, and others - and has only praise to offer them and their work. Still, it seems as though his eyes burn bright, like fanned embers, when he remembers the competition of the time.
Awarding Kilby with the Nobel Prize for the invention of the integrated circuit is a touchy subject. No one has anything critical to say about Kilby, but a few people say that everyone knows that Noyce invented the integrated circuit. Still, the Nobel Prize - as was pointed out to me at least 10 times during the course of researching this article - is awarded only to living people; it's not awarded posthumously. So Kilby, who created an integrated circuit at about the same time as Noyce, ends up the sole winner. It's incongruous and a little bitter for everyone who loved Noyce, yet everyone is happy that the Nobel Prize committee finally recognized the achievement. It's one of those issues that makes people a little grumpy.
No account of the semiconductor industry would be complete without discussing Charlie Sporck, the creator, if not the founder, of National Semiconductor and the legendary head of manufacturing at Fairchild Semiconductor. He retains the same gruff attitude he has always had, but a hint of deep humor resides somewhere under the attitude. Despite getting on in years, Sporck is about half a generation behind many of the people in this story. Sporck was classically trained engineer manager out of the General Electric system when he joined Fairchild Semiconductor as employee number 854. He was a senior executive at Fairchild Semiconductor when the famous Eight were well on their way to creating a new industry, and he left the company just in time to take over a tiny manufacturer of passive electronics components and bend it toward the innovation that he identifies as the mother of all inventions: the planar process, the basic manufacturing technique that defined an industry. In a nutshell, planar processes are the methods by which successive layers of semiconductors and conductors are deposited on a disc-shaped slice of a silicon ingot (and more exotic materials these days) and then successively coated, exposed, and etched with solvents, while keeping the junctions protected by silicon oxide.
Sporck's departure from Fairchild to National may have ratified the notion - tentatively introduced when the Traitorous Eight left Shockley - that it was OK to leave a lucrative career in a good company and move on because, simply put, there are better things to do with one's life and professional career than simply being a satisfied employee.
National Semiconductor latched onto planar techniques just as its arch rival, Raytheon, which had been content to slavishly copy National's products and designs for years, retreated from passive components and moved into what it thought would be greener valleys. National Semiconductor found its métier, pushing Raytheon (which had bought the declining Rheem Semiconductor operations) completely out of the semiconductor industry, and began a decade-long chase to extend the 16000 family of microprocessors. As is the way of the semiconductor, the 16000 family is also in history's dustbin, but rethinking that 20 years later is armchair quarterbacking.
In looking at this 70-year-old, still tree trunk-like and physically imposing, one thinks Sporck should have managed to push National Semiconductor over the top. He clearly had a feeling for the technology and was always at the head of industry wide issues. Sporck looked forward for the country as well as the industry. In retrospect, one wonders if it was the people or the lack of talent that prevented the company from beating Intel. Or perhaps it was just the sheer cussedness of Sporck's determination that somehow jinxed National Semiconductor as a contender. On the other hand, looking at National's new campus and growth, despite the microprocessor recession of today, it's clear that much of the infrastructure laid down under Sporck's reign still pushes the company along.
When Sporck left Fairchild Semiconductor, he took quite a group with him: Lamond; Fred Bialek; Floyd Kvamme; Don Valentine, who joined later; and Roger Smullen, who had joined Fairchild, as employee number 853, the same day as Sporck. Smullen started out at Fairchild as an entry-level engineer, fresh out of the University of Minnesota with a degree in mechanical engineering. Hired on at the princely salary of $540 per month, Smullen started in quality assurance, becoming a foreman in the wafer fab within eight months, under Lamond. Smullen was one of the first group of semiconductor engineers who attended the "Fairchild University" classes taught by Hoerni, Grove, Moore and Noyce. By studying under the men who had made up the techniques, Smullen learned how to do his job better, and he quickly moved up to manufacturing foreman and then to process engineer, under Sporck.
Smullen looks back on the Fairchild Semiconductor era and explains his and others' departures simply: "I got one share of stock as my reward for eight years of work, at an option price of $200 for the share. It was clear that these [Fairchild management] guys weren't going to share the wealth of what we were building." So, when Lamond asked Smullen to come to National Semiconductor and run the digital circuit group, he left, spending four years at National, where there was a distribution of shares.
By 1971, Hoerni, in his third startup since leaving Fairchild, invited an increasingly well-known Smullen to run the digital memories division of the newly united Intersil and AMS companies, building RAM cards for IBM mainframes. Smullen worked with Jack Gifford, who ran the analog division of Intersil, before leaving Intersil in 1979 to take some time off. Smullen then joined Franklin "Pitch" Johnson in helping to revive Applied Micro Circuits Corporation (AMCC), a San Diego-based bipolar manufacturer that has since helped breath life into, and also benefited from, the return of bipolar technology for digital telephony. Clearly, Sporck had chosen well in hiring Smullen, and, just as clearly, Sporck had been right in abandoning Fairchild Semiconductor.
During the time between Sporck's departure from Fairchild to take over at National and Noyce and Moore leaving to found Intel, Fairchild found itself largely leaderless. It's too simple of a telling, but essentially true, that the manager of arch rival Motorola's semiconductor division, Lester Hogan, was chosen to run Fairchild; Darth Vader was invited into the company and shown all of its secrets. Rather than saving the company, many felt Hogan betrayed it. That was the downfall of Fairchild Semiconductor as the industry leader, and, although Fairchild is successful today, it is a largely reborn company, after two tortuous decades. Included in those two decades is yet another saga, National Semiconductor's acquisition and sale of Fairchild, but moving on to that story would force us to ignore the story of others at Fairchild, such as Wilfred Corrigan.
For a brief time, Corrigan, the eventual founder of LSI Logic, Jerry Sanders, the eventual founder of AMD, and others worked together under Hogan at Fairchild. Corrigan was a classically trained chemist, and chemists were greatly needed in the early days of semiconductors. He had worked in Boston at Transitron - the second-largest manufacturer of semiconductors, after Texas Instruments, at that time - before he and his new Norwegian bride jumped on a plane to Phoenix. Corrigan worked for Motorola for eight years, becoming a senior manager of Motorola's semiconductor operation, and then Hogan selected Corrigan to go to California, where he and the other "heroes," as they were called, would attempt to save Fairchild Semiconductor. Corrigan is one of the unsung technologists, and a much-recognized CEO, who helped a nascent industry develop the basic techniques, materials, and processes for semiconductor-wafer manufacturing. His area of specialization was in the use of RF microwaves, and similar technologies, to turn solid materials into gases and deposit them on ultra thin layers once again as solids, over the surface of photo masks that covered the wafers to produce the "integration of circuits."
Every semiconductor manufacturer in the late 1960s was fully integrated: grew its own silicon or geranium ingots, designed and manufactured its own circuits, performed its own testing, and integrated circuits into packaging suitable for their end use. Corrigan recalls that he couldn't even buy the exotic liquid mixtures that he needed for the processes he was attempting to develop and make precise. Instead, he had to gather the raw materials and mix them together, somehow managing to live through all of the explosions that occurred in his development labs, although his arms are permanently scarred from his work. After years of work at Motorola, the end result was the creation of repeatable techniques that reliably produced products of remarkably high quality. Hence, Corrigan's invitation to join Fairchild and his eventual uplifting to president of the semiconductor division, until that division was sold to Schlumberger.
It seems like hyperbole, but you look at Corrigan and think that he is a cross between Thomas Edison (Corrigan has several patents in his name) and John D. Rockefeller (Corrigan helped make an industry and now has the wealth to demonstrate his record of accomplishments). From humble chemist to founder of a company that, 20 years later, has nearly $3 billion in revenue, he's still here - just like his colleague Jerry Sanders. If a movie were to be made about Silicon Valley, the Hollywood studios simply wouldn't know what to do with a character like Corrigan: a scientist, entrepreneur, businessman, and marketer, with his great breakthrough at LSI Logic. He was the first to attempt, and succeed at, a commercial business based on building full-custom integrated circuits for others at a time when the industry had different models. Corrigan had the knowledge and the gonads to attempt to change the basic rules of operating in an industry created and populated by giants. Compare him to Edison, and he blushes and says, "Nonsense." Still, there's a Bentley in the parking lot, and, today, thousands of companies can afford to design and build application-specific integrated circuits largely as a result of one modest man who thought he could change the rules.
Another Transitron alumnus and transplant to Silicon Valley's semiconductor industry, Robert Swanson, founded a company the same year as Corrigan. But Swanson's interim learning ground was at National Semiconductor with Sporck. Today, Corrigan's company, LSI Logic, sits across the street from Swanson's Linear Technologies. Trained as an industrial engineer, Swanson found his way into Transitron because of his knowledge of statistics. Transitron was making linear-analog products as well as transistors, but it really had no way to track what was happening with different batches of materials that were cooked at different temperatures and produced varying quantities of usable products. Swanson's job was to figure all that out and improve yields. The statistics he knew cold; the rest this still-garrulous, twinkling-eyed Irish son of Boston figured out on the job. In the process, he became the manager of entire manufacturing lines, then multiple product lines, and then entire factories. By the time he went to work for Sporck at National Semiconductor, Swanson was starting up factories in Scotland and Germany, and he later ran the linear product line at headquarters.
"Charlie sued me for seven years after I left. I guess he figured that National was the only company that was supposed to be in the linear business. It was the biggest piece of business, and I did take some of the brightest National guys out with me," Swanson recalls. However, Sporck had done much the same to Fairchild Semiconductor before Moore and Noyce left. Swanson watched as National Semiconductor entered the watch business, minicomputers, and every new business on the block, while his own division, and the old-fashioned analog devices like power amplifiers or rectifiers that his division produced, got little of the limelight. He left and, 20 years later, runs a $2 billion business he founded producing the same kinds of parts that he once built for National Semiconductor. No bones about it: Swanson, the most youthful 50-year-old that you can imagine, built a global company in the face of disbelief, but with the support of VCs like Don Valentine, Thomas Perkins, and others who put up the $4.5 million Swanson and his partners wanted for 30 percent of their nascent company. In speaking with Swanson, you know that he's a guy you'd like to go drinking with; in fact, he looks like he might be pretty good in a bar fight, too.
Today, Linear Technologies builds the products in your personal computer that take the power delivered by the battery and separate it out into the individual power voltages required by the keyboard, the hard drive, the display, the input/output channels, and everything else. This is pretty complicated stuff that is still done best by the analog products that Linear Technologies sells around the world.
The final stellar graduate of the original semiconductor companies is the now more mature-looking T.J. Rodgers. In 1981, on the day after Linear Technologies' IPO, Rodgers' Cypress Semiconductor went public, with Morgan Stanley as the lead banker. Talk about a lack of coincidence; talk about a self-made man. Speaking of which, this man can talk. When he starts talking, after a while - when you get past the in-your-face presentation and the "I have an opinion on almost everything" personality - you hear the story of a young football player for Dartmouth College who was too smart to remain a jock and who came to Stanford University to earn a master's degree and a Ph.D. in electrical engineering. Rodgers also remembers William Shockley, who was Rodgers' professor at Stanford - the kind of professor who would invite his graduate students up to the blackboard to outline an answer to his question, let them write away for half a class, and then tell them how stupid they were. Tell Rodgers that he's a little like Shockley, and he won't disagree. He probably even likes the comparison in some ways. It's Shockley's brusqueness, forthrightness, intelligence, and impatience that he shares, and not Shockley's inability to lead or manage.
Rodgers was interested in a new form of MOS technology, VMOS, which promised to provide some significant advantages to semiconductor products. American Microsystems spotted Rodgers' dissertation, bought the patent on his VMOS idea, and hired him to boot. Rodgers spent the next five years discovering why VMOS was a dead end - lessons that he calls his MBA. The Rodgers went to work for Sanders at AMD. Thinking about these two men at one company presents some amusing images. Still, you can see that Rodgers would have contributed mightily to Sanders's success with AMD. But, like Corrigan and Swanson, Rodgers was destined to run his own show. Rodgers builds, as he proudly points out, static RAMs, the product business that Intel largely abandoned to the competition around 1984. If you could ever get this guy to shut up, you'd maybe tell him you think he's one of the smartest sons of a bitch you've ever met - and, in the context of this issue, that's saying something - but he's not about to give you the chance. Instead, Rodgers tells you he hates ass-kissers and is a reluctant republican, and then he shoves a copy of his latest ass-chewing memo to the company's engineers, regarding a recent mistake, under your nose.
We started this story about semiconductors with Moore, Noyce and Last, wondering about the odd, boyish enthusiasm that is still visible 30 years later. The story of Marcian Edward "Ted" Hoff and the making of the first microprocessor, the Intel 4004, is similar.
"Him, oh, he's some marketing guy," one person said to me, making me immediately think of William Davidow and preparing me to discover a young superachiever with a bachelor's, a master's, and a doctorate in electrical engineering, with an emphasis in what we have come to call artificial intelligence. Hoff was employee 12 at Intel, where he remained for what would be a wide and rich career for some, but was just the beginning for a self-admitted 60-year-old inventor who keeps lasers and other technical wonders in his basement.
If Noyce was the grand old man, Moore the technical resource, and Vadasz the semiconductor wizard, then Hoff was the local postdoc, hired straight out of the labs at Stanford University, who was going to settle onto a workbench and make stuff happen. Hoff figured out how to push beyond the bipolar technology and products mandated by Noyce and into CMOS. And, by burying himself in the circuit lines and logic equivalencies of the semiconductors, he figured out that there were many different parts that Intel and other companies were designing and manufacturing that could be combined to make a device that served many general-purpose functions. This device became known as the microprocessor and, more specifically, the Intel 4004 and then the Intel 8008 microprocessors. What kind of man is capable of pushing a company full of already-acclaimed geniuses into further achievements?
As you wander through a rambling conversation with Hoff, he is more interested in pointing out all of the individuals who did this or that work: how Federico Faggin's appearance at Intel made the microprocessor that he imagined possible and how the guy at the next workbench, Dov Frohman, invented the erasable programmable read-only memory, and introduced the concept of upgrading semiconductors once they left the factory. Bemoan the loss of the old Silicon Valley and the wonderful old places where you could push through dusty bins of spare parts and components - the kind of places that fired up a later generation of young inventors, like Steve Wozniak - and Hoff puts you straight, rattling off a list of such places and their locations. He then tells you which place has what kind of parts before going into a diatribe about how the big argon laser that he bought at one of these places eats up a bunch of power and he can get one with what he is really interested in. At 65 years of age, Hoff is as much of a "Hardy Boy" today as he was at 20, and he probably will be until they put him in a box. Then we'll only have a few recollections of this kind of man, yet another of the special people who make Silicon Valley a land of dreams.
Hoff was managing a group at Intel when he hired a young Federico Faggin, straight from Italy with a doctorate in solid-state physics. According to Faggin, on his first day at Intel, an engineer from Japan Business Systems showed up and asked him about the product that he was building. When Faggin tried to explain that he was new to Intel, the Japanese representative flew into a tirade, telling him, "You bad," among other choice "jinglish" criticisms. After three days of tirades, Faggin convinced Masatoshi Shima, the Japanese representative, that he would design the custom circuit that the company had contracted with Intel to build, but he had to have a fair chance. So began a long-standing collaboration between Shima and Faggin, who would later hire Shima to work with him when he founded Zilog. Shima helped Faggin design the Z-80, the world's most produced microprocessor, which is still under license and being built today, but that's a story for another time.
When these two stood facing each other at Intel, Shima had the logic design for a chip that was to be included in a new calculator that Intel was supporting by replacing a handful of discrete-logic parts with a single chip. Faggin recalls that Noyce, Grove, and Vadasz were "all out worrying about memories. Intel was a memory company at that time, and the competition was thick. This full-custom thing was a side project for them; they were worrying about how to [recover] from a huge and recent failure in memories. This single-chip computer project wasn't even on their radar. They weren't concerned with it in the least. So I had to design and build the thing. Shima knew what functions he wanted in the calculator, and he had logic diagrams from the company. I had to figure out how to put that in silicon."
It's a long and convoluted story, but Faggin succeeded in building what was to become the first mass-production microprocessor at Intel. However, it was not until Japan Business Systems went into financial difficulties and released Intel from exclusive production limitations on the product that Faggin began to try to drag the Intel executives' attention toward what he had just created: the first microprocessor. "They just didn't get it," Faggin says. Well, they got it enough to allow Faggin to push ahead and design and develop the Intel 8008, a precursor to the current generation of microprocessors that is Intel's claim to fame, but it wasn't enough to keep a frustrated Faggin motivated. He left Intel, taking Shima and a young Ralph Ungermann (who later found the first Ethernet company) with him to Zilog, where they built the first microprocessor used widely in personal computers.
The semiconductor industry is rife with stories like this, and this is only a very truncated version of the tale. Still, if the Nobel Prize committee, in its wisdom, should decide to award a prize for a little piece of revolutionary silicon, Hoff - who, Faggin says, is the only person at Intel who supported his ideas and work, the only manager who "got it" - and Faggin would probably be named.
But, before we leave the semiconductor industry in the early '80s, let's spend a few moments with one other leader - a young man who grew up with 10 brothers and sisters in the cornfields of Indiana; attended the local technical university; graduated with a degree in electrical engineering; entered the U.S. Navy before the war had ended, but too late to fight; and started his first career at RCA. Bernard Vonderschmidtt was one of the early employees at RCA's solid-state division, during the company's heyday, and was elevated to run the division when it was making $500 million per year. Given his life and career thus far, building a hugely successful semiconductor division with a multinational company on the East Coast, Vonderschmidtt may not, at first glance, seem to belong in this pantheon of West Coast heroes. But allow me to explain.
Vonderschmidtt came to California in 1981 to run Zilog for Exxon, which had provided its founding capitalization. He stayed for three years before leaving, in disgust, as Exxon imploded in the computer field. One of the company's engineers, Ross Freeman, left with Vonderschmidt, and, together, they raised $4.25 million from John Doerr and several of Doerr's contacts. With that money, Vonderschmidtt and Freeman founded a new kind of semiconductor company called Xilinx, based upon the premise that the design of the microprocessor - or, more specifically, the logic employed by a designer in a chip's configuration - could be customized, as done with a full-custom chip (the freedom first given to designers by Corrigan and LSI Logic), but through the use of a common substrate of semiconductor product, more accurately referred to as a field-programmable gate array. In short, the designer takes logic-design software from Xilinx, applies a design to a base chip (the gate array), and programs the design onto a chip in a matter or days.
Similarly, the test function that often proves a full-custom design process has erred in some way is reduced to a matter of hours. In effect, the semiconductor revolution evolved into microprocessor, and, at Xilinx, the microprocessor has seemed to evolve to a stage where the customer, not the semiconductor company, determines to what use the silicon will be put. Today, Vonderschmidtt's company, which started with 600 gates (logic operations) on a chip, is about to produce gate arrays with 10 million transistors on a single chip. The company's four buildings, housing 2,500 employees, seem impervious to the current downturn faced by the general-purpose microprocessor-based companies, because the design of the microprocessor, according to Vonderschmidtt, is finally moving into the hands of those who will drive its future: the customers. It's an odd way to end this discussion of the microprocessor industry, thinking that the industry is about to change dramatically. But, after talking with the people in this story, you can well imagine that this is an industry likely to be reborn from within.
Bell Labs, as the research arm for the Bell companies - owners of millions of electromechanical relays across the nation used in its telephone switching networks - was one of the most interested parties in semiconductors and their ability to conduct electrical currents. Bell executives had the foresight to hope that, one day, Bell could replace its troublesome relays with more reliable devices made of semiconductors. Bell had also laid underwater cables that used vacuum tubes for repeaters at regular intervals, making the cables unreliable. So Bell funded a semiconductor laboratory in Murray Hill, N.J., which is where Shockley, Walter Brattain, and John Bardeen produced the work that received the Nobel Prize in 1956 "for their researches on semiconductors and their discovery of the transistor effect."
As the head of one of the country's leading scientific companies, Beckman understood the importance of semiconductors. Hence, he hired the greatest name in the industry to establish his company in that field, setting Shockley up in his own research and development facility. Although Beckman Instruments was based in Los Angeles, Shockley's new labs were set up near Palo Alto, CA, because Shockley's mother lived there.
The Shockley Semiconductor Labs were short-lived, however. With the exception of Robert Noyce, none of the key engineers working there could stomach Shockley for long, despite his unquestionable technical brilliance. In 1957, the labs' senior members selected Gordon Moore to contact Beckman and have Shockley moved aside as leader of the labs. When that didn't work, seven men - and eventually Noyce, making it eight - left the company. Moore still remembers the night he drove to Jean Hoerni's house to make the call to Beckman and also that it did little good.
In the end, Shockley may be most remembered for hiring the talented group and, some say, subsequently driving them away to join Fairchild Semiconductor. Moore, one of the "Traitorous Eight" who left to form Fairchild's semiconductor operation and eventually become one of Intel's founders, remembers the evening he was sitting at home in Maryland, when the phone rang, and the voice on the other end said, "This is Shockley." That's about all I remember about the call, but I took the job he offered. I had been doing pretty much esoteric work at Johns Hopkins University, looking at the spectroscopy of metals. I didn't know the first thing about semiconductors, but Shockley thought he needed a chemist. None of us knew his reputation as a manager at that time, but maybe we should have suspected, as none of his guys from Bell Labs were joining him in California." The son of a local policeman in the small coastal town of Pescadero, CA - directly west of what would become Silicon Valley - Moore wanted to return to California. Shockley's reputation and the incredible wage being offered, $750 a month, were all it took to bring Moore home, where he became one of the formative members of the group that would eventually make Santa Clara, CA, "Silicon Valley."
By the time Moore was on his way to California, commercial semiconductor manufacturing was underway in Boston, Phoenix, Dallas, and New Jersey. The establishment of the Shockley Semiconductor Labs was the first step toward adding the Bay area to that list. Many books have been written about the semiconductor industry's founding, with Charlie Sporck's book excerpted in our August issue as one of the few firsthand accounts (see "The Birth of Fairchild Semiconductor," August 2001, page 60). Trying to capture the semiconductor story for this issue of Upside meant tracking down the people who created the semiconductor industry - a nerve-wracking process. Noyce, eulogized in Upside (July 1990), was one of the first of his generation to pass away. But many of the other larger-than-life people from Fairchild Semiconductor - Eugene Kleiner, Jay Last, Pierre Lamond, Julius Blank, Andrew Grove, and Moore - are still around to provide insight into what it was like when the semiconductor industry was first being created. For this special issue, I particularly wanted to answer the question of whether such a technological discontinuity as the semiconductor revolution will ever appear again. The answer lies in a thorough understanding of how the Valley became "siliconized."
Just as Shockley knew the labs would need chemists, he knew that mechanical engineers would be required, so he hired two: Blank and Kleiner. Blank was a classic engineer and had worked at Babcock & Wilcox, where he designed and built the huge boilers used in power plants and utility companies. As a boy, Blank attended a technical high school in Brooklyn, where he learned the craft of building things. In 1943, the U.S. Army grabbed the young man, sent him to college, had him repair military aircraft, and then sent him to Europe to fight in World War II battles such as the Battle of Hurtgen Forest. By the time Blank returned to the States in 1946 to finish his bachelor's degree in mechanical engineering, he already had a lifetime's worth of practical experience. Then it was off to Babcock & Wilcox, Goodyear Aircraft, and, finally, Western Electric, which set him to work with germanium phototransistors, among other devices, to figure out how to replace its mechanical relays.
In 1956, Dean Knapic, a Western Electric alumnus, offered Blank a job at Shockley Semiconductor Labs. After traveling to Palo Alto to be interviewed by Shockley, Blank, like all the other original employees, underwent lengthy psychological testing - possibly an artifact of Arnold Beckman's experience in one of his firm's plants in Los Angeles, where an employee went berserk and stabbed a co-worker to death, or a result of Shockley's unorthodox views on personnel practices. Despite the days-long grilling, $10,000 a year plus moving expenses looked pretty good.
Blank and Kleiner shared all of the mechanical-engineering work at the company, which was housed in a small stucco building at 391 South Charleston Road in Mountain View, CA (now a chair shop that bears an incongruous brass plaque identifying it as the birthplace of Silicon Valley).
Blank's first assignment was to build a crystal grower. Blank knew nothing about semiconductors. Fortunately, Leo Valdez and Victor Jones, hired by Shockley to actually grow the crystals, shared what they knew about the type of equipment they needed, and Blank went to work. "That's what it was like then. Bobby Noyce would walk in and say, 'I want you to melt some copper on this part.' Really vague instructions were the order of the day. I would do that, and then take it in to him. He'd look at it and tell me how to change it, or make some other suggestions, and we would go back and forth like that, making things up," Blank recalls.
Despite the lack of direction, Blank loved the work. After he joined Shockley Semiconductor Labs, Noyce, Moore, Last, Kleiner, and Hoerni appeared in short order. Blank remembers the group as a bunch of 20-somethings who liked to hang out together and see each other socially. He remembers the entire year and a half at Shockley Semiconductor Labs as an exciting time, ordering power upgrades, phone systems, air conditioners, and the radio-frequency (RF) oscillators needed to melt silicon. An indication of how Blank was regarded by his colleagues is the fact that Blank was nominated to recontact Beckman, after Moore's first attempt, about removing Shockley. "At first, it appeared I was successful," Blank says. Beckman endowed Shockley with a teaching chair at Stanford University that kept the good doctor out of the men's hair. Teaching, in addition to Shockley's speaking and travel demands after winning the Nobel Prize, initially seemed to have solved the issue. After a while, however, with Shockley coming back from trips and ordering entire projects restarted, it was clear that the problems would not be resolved. There was also Shockley's single-minded pursuit of the four-layer diode, perhaps left over from his days at Bell Labs, while many of the others thought silicon transistors were the direction they should be headed in.
Eugene Kleiner - co-founder, many years later, of the venture capital firm Kleiner Perkins - was another of Shockley's early hires. Kleiner was an immigrant from Europe. After leaving Vienna, Austria, he attended secondary school in New York. He then took work as a factory machinist, but, like so many of the remarkable men in this issue, he was drafted into the military. After the war, Kleiner earned a bachelor's degree from the Polytechnic University of New York and a master's in mechanical engineering from New York University. He taught engineering for a short while and then joined Western Electric, where he worked in the morass of the Bell system's relays and switches. Kleiner remembers Shockley as a charming person, a fascinating conversationalist, and, by reputation, one of the stars of Bell Labs.
Reconciling that impression of Shockley with the small, inadequate, and dirty building that Shockley had leased to start the company was the first of many events informing Kleiner that Shockley's technical brilliance was not matched by practical experience. Like Blank, Kleiner's first assignment was to build a crystal grower. It was this experience that brought Shockley down to earth in Kleiner's eyes. "I didn't know anything about growing crystals and [knew] nothing about furnaces. So I asked Shockley, but he didn't know. He just gave me advice, often the wrong advice, so our first device for growing crystals was a monstrosity. It didn't work. It was so large that we had to raise the roof of the building. It never worked. So I went next door to Sears, Roebuck and Co. and bought one of those large standing drill presses, and it had most of the features we needed. We had to change some things and add some things, but it formed the basis of our second and successful, crystal grower."
Then it was time to build a furnace, which entailed a similarly unsuccessful set of experiences. Still, "working with that team of men, Moore figured out the dispersal of gaseous materials," Kleiner recalls. (Moore was an expert glassblower. He actually created, by hand, the tube jungles in which gases were distributed, separated, and combined, thus allowing for single-step production of doubly diffused transistors.) "Knapic. Blank. It was a beautiful team. It was exciting," Kleiner continues. Once built, the furnace had to be kept on 24 hours a day. Kleiner and Blank, who lived close to the company, came in every couple of hours during weekends to check on the relatively delicate device. Kleiner remembers, 40 years later, that it was his wife who kept him going during those times (she also sang madrigals with Noyce at parties and social events). More importantly, she wrote a letter to Hayden, Stone & Co. seeking money for several of the "Fairchild Eight" - the letter that found its way to Arthur Rock and convinced Rock to find funding for Fairchild Semiconductor and, later, Teledyne Semiconductor and Intel. It was "one of the great letters of all time," Kleiner muses, thinking of that lovely unsung heroine.
Jay Last, now retired and living in Los Angeles, left Silicon Valley early on to become a vice president at Teledyne, where he helped Henry Singleton build that company from a single division to over 150 divisions. Last, a quiet electronics and optical engineer, worked closely with Noyce, and is responsible for much of the early manufacturing infrastructure of semiconductor technology. Surrounded by his collection of African art and the products of his publishing company, Last is, in many ways, uncomfortable sharing his life with strangers. His years of military research and development have remained a closed book long after his retirement. And the subject of Shockley is one that, despite the passage of 40 years, is not casually remembered.
Still, if you catch his attention, Last will explain that much of what helped build the semiconductor industry came out of Fairchild Semiconductor: improvements in growing silicon ingots and diffusing exotic materials across the substrate of silicon wafers. Last's own contributions from optical science - the creation of photographic masks that were used to expose patterns on the substrates and the development of etching processes through which lines and connections were created on the wafers in miniature - are tossed off casually. "None of that had been done. We were inventing everything," Last says. This includes, most notably, adapting these processes to wafer fabrication and manufacturing. Waving his hands, Last dismisses the greatest revolution in manufacturing since Henry Ford's achievements, but it was Last's special contribution: "the creation of step-and-repeat methods by ganging up these microscopes and devices on mechanically operated stages, so that we could create integrated circuits cheaply, quickly, and reliably." He holds out one of the first commercially integrated circuits, with its five gold leads, showing me the four-transistor device with a sense of wonderment, even after all these years.
"And this," he says, retrieving yet another device from a trove of first technologies. I notice a tiny "open here" legend printed on its side. Last smiles conspiratorially and says, "We knew that the people at TI were desperate to see what we were doing, and we knew that, as soon as we released these, they would get some, so I had these printed up for them." He grins at his one-upmanship with boyish delight.
He won't say it, but those processes, step and repeat, led to "batch manufacturing," which spread out of semiconductor manufacturing and into other technological fields, such as LED manufacturing, biotechnology, and, experimentally, nanotechnology, fostering a worldwide change in manufacturing technologies that will continue throughout the 21st century. Last is sanguine, comfortable with his accomplishments and a hoard of unmentioned patents. I sit pondering this guarded, gentle, and brilliant man who tells me that, despite the passage of four decades, he can walk into a modern-day wafer-fabrication facility and still know what the machines do, which is basically what he and his colleagues first laid down in the 1950s. Last thinks back on Noyce, his closest collaborator at Fairchild Semiconductor, with a fondness that is touching - an almost universal emotion among the giants still walking the Earth who helped to create Silicon Valley.
While the men were in a quandary over their problems with Shockley, there were other companies working with semiconductors. Boston-based Transitron Electronic was a contemporary of the Shockley Semiconductor Labs. Transitron was found by an eccentric pair of brothers named David and Leo Bakalar. One of the pair ran a shoe factory and was cheaper than blue jeans; the other was a Ph.D. who had worked at Bell Labs on transistor research. They recruited engineers to their firm by traveling through Europe and conducting interviews in major cities. They hired the brightest immigrants they could find, brought them to the United States, and worked them silly in the early days of discrete electronics components. Pierre Lamond, Wilfred Corrigan, Robert Swanson, Lester Hogan, and a young Hungarian immigrant named Leslie Vadasz passed through Transitron's doors and built a semiconductor company that vied with Texas Instruments for the number one position in the nation in the late 1950s. At that time, semiconductors like germanium were employed to produce devices such as rectifiers in small aluminum canisters, which electronics manufacturers used by the thousands.
Pierre Lamond, a jeweler's son who fought in the Algerian War and made contacts with American officers during a stint at NATO headquarters, wanted to find a way to work in the United States. It was clear that his field (physics and electron optics) was moving fastest in America. In 1957, a decommissioned Lamond saw an employment ad in the New York Times for Transitron. David Bakalar offered him a job on one of his recruiting trips through Paris. To get Lamond started, the Bakalars put him on the production line. In his third week, he was promoted to the head of production to replace his departing boss, and, in a few more months, he was promoted to device development engineer. It was a whirlwind introduction to semiconductor technology, and, for the one-year duration of his working visa, he absorbed everything that he could before returning to France. Lamond shopped for a company in his native country where he could apply his skills, but he found nothing comparable to his experience in Boston, so, by 1959, he was back at Transitron as the head of development.
Lamond's stint at Transitron was a short one, as most were. The Bakalars' approach was to "second source" parts designed by others (a polite way to describe their energetic copying of other people's products), which they would then sell to customers at cheaper prices. Lamond had heard of Hoerni's work at Fairchild Semiconductor in developing what would become known as the planar process of manufacturing semiconductor products. Lamond approached the Bakalars and told them that, if they were to compete, it was time for Transitron to invest in planar manufacturing, but they would have nothing of it, telling him to wait a few months until all the bugs were worked out of the process. Unfortunately for the Bakalars, the energetic Lamond had befriended Moore while presenting a paper at a technical conference, so the brothers' refusal to move Transitron along the path to original-product development came at a time when Fairchild Semiconductor's human resources manager was already courting Lamond.
In 1961, Lamond joined Fairchild Semiconductor, working for Moore. After one of the periodic diasporas of Fairchild's engineers to form a new company - in this case Signetics - and the loss of Lamond's direct supervisor, he was again promoted to lead device development. He remained at Fairchild from 1961 until 1967, when he became a member of another group leaving Fairchild, under Charlie Sporck, to restart National Semiconductor. Lamond remained with National until 1974.
Many of the Transitron staff passed through Phoenix (Motorola) or Dallas (Texas Instruments) before coming to Silicon Valley. Vadasz spent his three years of apprenticeship at Transitron, for $650 a month, before moving straight to Silicon Valley, where the man who had hired him into Transistron, Lamond, hired him again at Fairchild. As with others at Transitron, Vadasz found himself building whatever he needed. Need vacuum chambers to distribute evaporated dopants? Build them yourself, out of a piece of tube. Need some silicon wafers? Build your own furnace; dump in chunks of silicon lumps; start it spinning; dip in a piece of seed crystal, rotating it in opposite directions; pull it out slowly; and hope for the best. Everything else pretty much required working under a microscope with instruments held in your own shaking hands. Everybody who passed through the Boston sweatshop seems to have come out the better for having been there, and for having left.
For example, Wilfred Corrigan remembers someone from Transitron coming into a room, where he was reading about semiconductors, and telling him that the production line was a mess: "They told me to get out there and get it running right, so I went, and I figured out semiconductors later." Transitron was that kind of place.
Fairchild Semiconductor was, by comparison, a fortress of strong organization. Imagine coming into a company where at least a dozen future semiconductor company presidents and founders are working. At the head of the organization is Sherman Fairchild, founder of Fairchild Camera and Instrument, scion of one of the first investors in IBM, and head of a company with technology interests in multiple industries. A large part of the comparative stability at Fairchild Semiconductor was based upon its hiring of Ed Baldwin as general manager of its new semiconductor division. Baldwin recalls, "I came up from Los Angeles, and they interviewed me - all of the them, Noyce, Moore, [Victor] Grinich, and Last - but it was Moore that made the decision to hire me."
Their choice was for someone as eclectic as themselves. Baldwin's father was an itinerant intellectual: a musician, engineer, and writer. His father was related to a former prime minister of England, where his family hailed from, and the founder of a technical university that still exists in London today. Baldwin earned one of Carnegie Mellon University's first master's degrees in solid-state physics and then a Ph.D. in nuclear physics, also from Carnegie Mellon, where he worked in neutron and proton scattering. Baldwin also built Carnegie Mellon's synchro-cyclotron, one of the progeny devices of the Berkeley cyclotron.
In 1950, the young scientist was recruited by Hughes Semiconductors, of Los Angeles, during that company's heyday. There, Baldwin worked with people such as Simon Ramo and Dean Woolridge, two of the founders of TRW, on the miniature diodes Hughes produced for airborne computers. By 1956, Baldwin had risen to the head of product development in Hughes' computer division, and, when that was folded, he became head of product development for the semiconductor division. While at Hughes, Baldwin wrote his first two patents for the company. The first patent was on semiconductor photodiodes, and the second patent was for a device that would use the diodes to read hundreds of mainframe punch cards per minute. The company promptly gave him a $100 bonus, patted him on the head, and sold his division to RCA for a small fortune. At that point, Baldwin was ready for a career change.
Baldwin remembers Hughes as a great training ground where he learned about operations, plant maintenance, engineering, manufacturing, personnel, and research and development. At Hughes, Baldwin also learned how to work with brilliant technologists and how to act as a professional manager. Nevertheless, an ad in the Wall Street Journal, seeking an executive and a general manager for a startup in Northern California, caught his eye. Baldwin had been looking for new work for a while, and, along the way, he'd befriended a banker who knew a source of capital for Baldwin's possible own company. But Baldwin hadn't heard from the banker in months, and the offer to work with the likes of Hoerni and Moore proved irresistible.
Baldwin had been at Fairchild Semiconductor for several months when the banker turned up, having secured a commitment from a wealthy industrial-manufacturing family by the name of Rheem. The banker was ready to press a check for $5 million into Baldwin's hands. "It was the worst decision of my life. If I hadn't made [that decision] I would probably be at Intel today," he says, looking wistful, and then he smiles and starts telling the story of a lifetime's worth of companies he has started and run. Baldwin reminds me that, at one point, an engineer who left Fairchild Semiconductor to join him at Rheem Semiconductor took one of Fairchild's precious "recipe books" - a source of considerable disgruntlement to this day. Baldwin recalls, "When I found out about it, I gave him the living daylights, and I returned the book." Today, at a spry 82 years of age, Baldwin is running a startup in San Diego, Energy Development Systems, where he will be working in his field "until the day I die," having just earned the most recent of several patents on the high-power capacitors that fuel his dreams for tomorrow.
While Baldwin was departing to found Rheem Semiconductor, Vadasz was in the midst of four years of work putting down deep roots into the development of bipolar, and later MOS, semiconductor products. But Vadasz was also destined to leave Fairchild Semiconductor, and was hired by Noyce and Moore to work in research at Intel, the company Noyce and Moore founded with the intent of designing and building semiconductor memories. There was a growing market for semiconductor components, but Noyce and Moore felt that, if they could build solid-state, or thin-film, memories that would replace expensive and hard-to-manufacture solid-core memories, they would have ready customers among mainframe computer companies. It was a dubious market strategy at best, fraught with tough competition and only occasionally won by Intel. U.S. manufacturers ultimately abandoned the strategy and left memory making to Asian manufacturers.
Despite a well-published error having to do with a numbering scheme introduced much later, Noyce and Moore hired Grove and then Vadasz as their first Intel employees. Both Vadasz and Grove remain at Intel today. Ask Vadasz what he did in the beginning at Intel, and he'll explain that he worked on the technologies that led Intel into MOS manufacturing, the technique that is standard today. He goes on to admit that Intel's first successful memory products were good old-fashioned bipolar ones, like those of Fairchild Semiconductor. Press him for some details on the success he has brought Intel, and he'll tell you that it was Moore and others experimenting in the labs with MOS who finally figured out why the yields in the new technology were so low for Intel and made the process practical.
Out in the dry plains of Texas, Texas Instruments (TI) had been around since 1933, under another name, supporting the scientists working in the oil industry of Texas and beyond. A Texan giant of a man, Jack Kilby, had returned from Illinois to his home state with a bachelor's degree in electrical engineering to work with the resistors, capacitors, and discrete-logic products of his time - basically to build the instruments that TI manufactured. The Iliac was just being built when Kilby graduated from college, but computing was firmly on the horizon, and Kilby increasingly turned his attention to germanium and then silicon, spending a lifetime in the semiconductor field, with a single company. Kilby can sit and preach the semiconductor bible from memory, because, at TI, he did what all of his competitors were doing elsewhere, slowly experimenting with semiconductor materials. He developed the design and manufacturing processes that allowed TI to "get there first," making TI the undisputed leader in transistor, and then semiconductor manufacturing.
There's a hard edge in the 90-year-old's clear blue eyes when you ask him about the industry back then. "We didn't share anything in those days. We were all competitors, in everything. We went to the ISSCC [International Solid State Circuits Conference] and announced our progress each year, but we said as little as possible beyond what we had done, and we certainly didn't tell anyone about our processes." Kilby, recently awarded a Nobel Prize for the invention of the integrated circuit at TI, doesn't talk much about his or TI's primacy in the field; he seems proud just to have been a part of the efforts in building one of the first multitransistor semiconductor microprocessors - a six-transistor part that TI built for use in products such as portable calculators and the Minuteman program. Kilby knows all the names from the Valley - Moore, Sporck, and others - and has only praise to offer them and their work. Still, it seems as though his eyes burn bright, like fanned embers, when he remembers the competition of the time.
Awarding Kilby with the Nobel Prize for the invention of the integrated circuit is a touchy subject. No one has anything critical to say about Kilby, but a few people say that everyone knows that Noyce invented the integrated circuit. Still, the Nobel Prize - as was pointed out to me at least 10 times during the course of researching this article - is awarded only to living people; it's not awarded posthumously. So Kilby, who created an integrated circuit at about the same time as Noyce, ends up the sole winner. It's incongruous and a little bitter for everyone who loved Noyce, yet everyone is happy that the Nobel Prize committee finally recognized the achievement. It's one of those issues that makes people a little grumpy.
No account of the semiconductor industry would be complete without discussing Charlie Sporck, the creator, if not the founder, of National Semiconductor and the legendary head of manufacturing at Fairchild Semiconductor. He retains the same gruff attitude he has always had, but a hint of deep humor resides somewhere under the attitude. Despite getting on in years, Sporck is about half a generation behind many of the people in this story. Sporck was classically trained engineer manager out of the General Electric system when he joined Fairchild Semiconductor as employee number 854. He was a senior executive at Fairchild Semiconductor when the famous Eight were well on their way to creating a new industry, and he left the company just in time to take over a tiny manufacturer of passive electronics components and bend it toward the innovation that he identifies as the mother of all inventions: the planar process, the basic manufacturing technique that defined an industry. In a nutshell, planar processes are the methods by which successive layers of semiconductors and conductors are deposited on a disc-shaped slice of a silicon ingot (and more exotic materials these days) and then successively coated, exposed, and etched with solvents, while keeping the junctions protected by silicon oxide.
Sporck's departure from Fairchild to National may have ratified the notion - tentatively introduced when the Traitorous Eight left Shockley - that it was OK to leave a lucrative career in a good company and move on because, simply put, there are better things to do with one's life and professional career than simply being a satisfied employee.
National Semiconductor latched onto planar techniques just as its arch rival, Raytheon, which had been content to slavishly copy National's products and designs for years, retreated from passive components and moved into what it thought would be greener valleys. National Semiconductor found its métier, pushing Raytheon (which had bought the declining Rheem Semiconductor operations) completely out of the semiconductor industry, and began a decade-long chase to extend the 16000 family of microprocessors. As is the way of the semiconductor, the 16000 family is also in history's dustbin, but rethinking that 20 years later is armchair quarterbacking.
In looking at this 70-year-old, still tree trunk-like and physically imposing, one thinks Sporck should have managed to push National Semiconductor over the top. He clearly had a feeling for the technology and was always at the head of industry wide issues. Sporck looked forward for the country as well as the industry. In retrospect, one wonders if it was the people or the lack of talent that prevented the company from beating Intel. Or perhaps it was just the sheer cussedness of Sporck's determination that somehow jinxed National Semiconductor as a contender. On the other hand, looking at National's new campus and growth, despite the microprocessor recession of today, it's clear that much of the infrastructure laid down under Sporck's reign still pushes the company along.
When Sporck left Fairchild Semiconductor, he took quite a group with him: Lamond; Fred Bialek; Floyd Kvamme; Don Valentine, who joined later; and Roger Smullen, who had joined Fairchild, as employee number 853, the same day as Sporck. Smullen started out at Fairchild as an entry-level engineer, fresh out of the University of Minnesota with a degree in mechanical engineering. Hired on at the princely salary of $540 per month, Smullen started in quality assurance, becoming a foreman in the wafer fab within eight months, under Lamond. Smullen was one of the first group of semiconductor engineers who attended the "Fairchild University" classes taught by Hoerni, Grove, Moore and Noyce. By studying under the men who had made up the techniques, Smullen learned how to do his job better, and he quickly moved up to manufacturing foreman and then to process engineer, under Sporck.
Smullen looks back on the Fairchild Semiconductor era and explains his and others' departures simply: "I got one share of stock as my reward for eight years of work, at an option price of $200 for the share. It was clear that these [Fairchild management] guys weren't going to share the wealth of what we were building." So, when Lamond asked Smullen to come to National Semiconductor and run the digital circuit group, he left, spending four years at National, where there was a distribution of shares.
By 1971, Hoerni, in his third startup since leaving Fairchild, invited an increasingly well-known Smullen to run the digital memories division of the newly united Intersil and AMS companies, building RAM cards for IBM mainframes. Smullen worked with Jack Gifford, who ran the analog division of Intersil, before leaving Intersil in 1979 to take some time off. Smullen then joined Franklin "Pitch" Johnson in helping to revive Applied Micro Circuits Corporation (AMCC), a San Diego-based bipolar manufacturer that has since helped breath life into, and also benefited from, the return of bipolar technology for digital telephony. Clearly, Sporck had chosen well in hiring Smullen, and, just as clearly, Sporck had been right in abandoning Fairchild Semiconductor.
During the time between Sporck's departure from Fairchild to take over at National and Noyce and Moore leaving to found Intel, Fairchild found itself largely leaderless. It's too simple of a telling, but essentially true, that the manager of arch rival Motorola's semiconductor division, Lester Hogan, was chosen to run Fairchild; Darth Vader was invited into the company and shown all of its secrets. Rather than saving the company, many felt Hogan betrayed it. That was the downfall of Fairchild Semiconductor as the industry leader, and, although Fairchild is successful today, it is a largely reborn company, after two tortuous decades. Included in those two decades is yet another saga, National Semiconductor's acquisition and sale of Fairchild, but moving on to that story would force us to ignore the story of others at Fairchild, such as Wilfred Corrigan.
For a brief time, Corrigan, the eventual founder of LSI Logic, Jerry Sanders, the eventual founder of AMD, and others worked together under Hogan at Fairchild. Corrigan was a classically trained chemist, and chemists were greatly needed in the early days of semiconductors. He had worked in Boston at Transitron - the second-largest manufacturer of semiconductors, after Texas Instruments, at that time - before he and his new Norwegian bride jumped on a plane to Phoenix. Corrigan worked for Motorola for eight years, becoming a senior manager of Motorola's semiconductor operation, and then Hogan selected Corrigan to go to California, where he and the other "heroes," as they were called, would attempt to save Fairchild Semiconductor. Corrigan is one of the unsung technologists, and a much-recognized CEO, who helped a nascent industry develop the basic techniques, materials, and processes for semiconductor-wafer manufacturing. His area of specialization was in the use of RF microwaves, and similar technologies, to turn solid materials into gases and deposit them on ultra thin layers once again as solids, over the surface of photo masks that covered the wafers to produce the "integration of circuits."
Every semiconductor manufacturer in the late 1960s was fully integrated: grew its own silicon or geranium ingots, designed and manufactured its own circuits, performed its own testing, and integrated circuits into packaging suitable for their end use. Corrigan recalls that he couldn't even buy the exotic liquid mixtures that he needed for the processes he was attempting to develop and make precise. Instead, he had to gather the raw materials and mix them together, somehow managing to live through all of the explosions that occurred in his development labs, although his arms are permanently scarred from his work. After years of work at Motorola, the end result was the creation of repeatable techniques that reliably produced products of remarkably high quality. Hence, Corrigan's invitation to join Fairchild and his eventual uplifting to president of the semiconductor division, until that division was sold to Schlumberger.
It seems like hyperbole, but you look at Corrigan and think that he is a cross between Thomas Edison (Corrigan has several patents in his name) and John D. Rockefeller (Corrigan helped make an industry and now has the wealth to demonstrate his record of accomplishments). From humble chemist to founder of a company that, 20 years later, has nearly $3 billion in revenue, he's still here - just like his colleague Jerry Sanders. If a movie were to be made about Silicon Valley, the Hollywood studios simply wouldn't know what to do with a character like Corrigan: a scientist, entrepreneur, businessman, and marketer, with his great breakthrough at LSI Logic. He was the first to attempt, and succeed at, a commercial business based on building full-custom integrated circuits for others at a time when the industry had different models. Corrigan had the knowledge and the gonads to attempt to change the basic rules of operating in an industry created and populated by giants. Compare him to Edison, and he blushes and says, "Nonsense." Still, there's a Bentley in the parking lot, and, today, thousands of companies can afford to design and build application-specific integrated circuits largely as a result of one modest man who thought he could change the rules.
Another Transitron alumnus and transplant to Silicon Valley's semiconductor industry, Robert Swanson, founded a company the same year as Corrigan. But Swanson's interim learning ground was at National Semiconductor with Sporck. Today, Corrigan's company, LSI Logic, sits across the street from Swanson's Linear Technologies. Trained as an industrial engineer, Swanson found his way into Transitron because of his knowledge of statistics. Transitron was making linear-analog products as well as transistors, but it really had no way to track what was happening with different batches of materials that were cooked at different temperatures and produced varying quantities of usable products. Swanson's job was to figure all that out and improve yields. The statistics he knew cold; the rest this still-garrulous, twinkling-eyed Irish son of Boston figured out on the job. In the process, he became the manager of entire manufacturing lines, then multiple product lines, and then entire factories. By the time he went to work for Sporck at National Semiconductor, Swanson was starting up factories in Scotland and Germany, and he later ran the linear product line at headquarters.
"Charlie sued me for seven years after I left. I guess he figured that National was the only company that was supposed to be in the linear business. It was the biggest piece of business, and I did take some of the brightest National guys out with me," Swanson recalls. However, Sporck had done much the same to Fairchild Semiconductor before Moore and Noyce left. Swanson watched as National Semiconductor entered the watch business, minicomputers, and every new business on the block, while his own division, and the old-fashioned analog devices like power amplifiers or rectifiers that his division produced, got little of the limelight. He left and, 20 years later, runs a $2 billion business he founded producing the same kinds of parts that he once built for National Semiconductor. No bones about it: Swanson, the most youthful 50-year-old that you can imagine, built a global company in the face of disbelief, but with the support of VCs like Don Valentine, Thomas Perkins, and others who put up the $4.5 million Swanson and his partners wanted for 30 percent of their nascent company. In speaking with Swanson, you know that he's a guy you'd like to go drinking with; in fact, he looks like he might be pretty good in a bar fight, too.
Today, Linear Technologies builds the products in your personal computer that take the power delivered by the battery and separate it out into the individual power voltages required by the keyboard, the hard drive, the display, the input/output channels, and everything else. This is pretty complicated stuff that is still done best by the analog products that Linear Technologies sells around the world.
The final stellar graduate of the original semiconductor companies is the now more mature-looking T.J. Rodgers. In 1981, on the day after Linear Technologies' IPO, Rodgers' Cypress Semiconductor went public, with Morgan Stanley as the lead banker. Talk about a lack of coincidence; talk about a self-made man. Speaking of which, this man can talk. When he starts talking, after a while - when you get past the in-your-face presentation and the "I have an opinion on almost everything" personality - you hear the story of a young football player for Dartmouth College who was too smart to remain a jock and who came to Stanford University to earn a master's degree and a Ph.D. in electrical engineering. Rodgers also remembers William Shockley, who was Rodgers' professor at Stanford - the kind of professor who would invite his graduate students up to the blackboard to outline an answer to his question, let them write away for half a class, and then tell them how stupid they were. Tell Rodgers that he's a little like Shockley, and he won't disagree. He probably even likes the comparison in some ways. It's Shockley's brusqueness, forthrightness, intelligence, and impatience that he shares, and not Shockley's inability to lead or manage.
Rodgers was interested in a new form of MOS technology, VMOS, which promised to provide some significant advantages to semiconductor products. American Microsystems spotted Rodgers' dissertation, bought the patent on his VMOS idea, and hired him to boot. Rodgers spent the next five years discovering why VMOS was a dead end - lessons that he calls his MBA. The Rodgers went to work for Sanders at AMD. Thinking about these two men at one company presents some amusing images. Still, you can see that Rodgers would have contributed mightily to Sanders's success with AMD. But, like Corrigan and Swanson, Rodgers was destined to run his own show. Rodgers builds, as he proudly points out, static RAMs, the product business that Intel largely abandoned to the competition around 1984. If you could ever get this guy to shut up, you'd maybe tell him you think he's one of the smartest sons of a bitch you've ever met - and, in the context of this issue, that's saying something - but he's not about to give you the chance. Instead, Rodgers tells you he hates ass-kissers and is a reluctant republican, and then he shoves a copy of his latest ass-chewing memo to the company's engineers, regarding a recent mistake, under your nose.
We started this story about semiconductors with Moore, Noyce and Last, wondering about the odd, boyish enthusiasm that is still visible 30 years later. The story of Marcian Edward "Ted" Hoff and the making of the first microprocessor, the Intel 4004, is similar.
"Him, oh, he's some marketing guy," one person said to me, making me immediately think of William Davidow and preparing me to discover a young superachiever with a bachelor's, a master's, and a doctorate in electrical engineering, with an emphasis in what we have come to call artificial intelligence. Hoff was employee 12 at Intel, where he remained for what would be a wide and rich career for some, but was just the beginning for a self-admitted 60-year-old inventor who keeps lasers and other technical wonders in his basement.
If Noyce was the grand old man, Moore the technical resource, and Vadasz the semiconductor wizard, then Hoff was the local postdoc, hired straight out of the labs at Stanford University, who was going to settle onto a workbench and make stuff happen. Hoff figured out how to push beyond the bipolar technology and products mandated by Noyce and into CMOS. And, by burying himself in the circuit lines and logic equivalencies of the semiconductors, he figured out that there were many different parts that Intel and other companies were designing and manufacturing that could be combined to make a device that served many general-purpose functions. This device became known as the microprocessor and, more specifically, the Intel 4004 and then the Intel 8008 microprocessors. What kind of man is capable of pushing a company full of already-acclaimed geniuses into further achievements?
As you wander through a rambling conversation with Hoff, he is more interested in pointing out all of the individuals who did this or that work: how Federico Faggin's appearance at Intel made the microprocessor that he imagined possible and how the guy at the next workbench, Dov Frohman, invented the erasable programmable read-only memory, and introduced the concept of upgrading semiconductors once they left the factory. Bemoan the loss of the old Silicon Valley and the wonderful old places where you could push through dusty bins of spare parts and components - the kind of places that fired up a later generation of young inventors, like Steve Wozniak - and Hoff puts you straight, rattling off a list of such places and their locations. He then tells you which place has what kind of parts before going into a diatribe about how the big argon laser that he bought at one of these places eats up a bunch of power and he can get one with what he is really interested in. At 65 years of age, Hoff is as much of a "Hardy Boy" today as he was at 20, and he probably will be until they put him in a box. Then we'll only have a few recollections of this kind of man, yet another of the special people who make Silicon Valley a land of dreams.
Hoff was managing a group at Intel when he hired a young Federico Faggin, straight from Italy with a doctorate in solid-state physics. According to Faggin, on his first day at Intel, an engineer from Japan Business Systems showed up and asked him about the product that he was building. When Faggin tried to explain that he was new to Intel, the Japanese representative flew into a tirade, telling him, "You bad," among other choice "jinglish" criticisms. After three days of tirades, Faggin convinced Masatoshi Shima, the Japanese representative, that he would design the custom circuit that the company had contracted with Intel to build, but he had to have a fair chance. So began a long-standing collaboration between Shima and Faggin, who would later hire Shima to work with him when he founded Zilog. Shima helped Faggin design the Z-80, the world's most produced microprocessor, which is still under license and being built today, but that's a story for another time.
When these two stood facing each other at Intel, Shima had the logic design for a chip that was to be included in a new calculator that Intel was supporting by replacing a handful of discrete-logic parts with a single chip. Faggin recalls that Noyce, Grove, and Vadasz were "all out worrying about memories. Intel was a memory company at that time, and the competition was thick. This full-custom thing was a side project for them; they were worrying about how to [recover] from a huge and recent failure in memories. This single-chip computer project wasn't even on their radar. They weren't concerned with it in the least. So I had to design and build the thing. Shima knew what functions he wanted in the calculator, and he had logic diagrams from the company. I had to figure out how to put that in silicon."
It's a long and convoluted story, but Faggin succeeded in building what was to become the first mass-production microprocessor at Intel. However, it was not until Japan Business Systems went into financial difficulties and released Intel from exclusive production limitations on the product that Faggin began to try to drag the Intel executives' attention toward what he had just created: the first microprocessor. "They just didn't get it," Faggin says. Well, they got it enough to allow Faggin to push ahead and design and develop the Intel 8008, a precursor to the current generation of microprocessors that is Intel's claim to fame, but it wasn't enough to keep a frustrated Faggin motivated. He left Intel, taking Shima and a young Ralph Ungermann (who later found the first Ethernet company) with him to Zilog, where they built the first microprocessor used widely in personal computers.
The semiconductor industry is rife with stories like this, and this is only a very truncated version of the tale. Still, if the Nobel Prize committee, in its wisdom, should decide to award a prize for a little piece of revolutionary silicon, Hoff - who, Faggin says, is the only person at Intel who supported his ideas and work, the only manager who "got it" - and Faggin would probably be named.
But, before we leave the semiconductor industry in the early '80s, let's spend a few moments with one other leader - a young man who grew up with 10 brothers and sisters in the cornfields of Indiana; attended the local technical university; graduated with a degree in electrical engineering; entered the U.S. Navy before the war had ended, but too late to fight; and started his first career at RCA. Bernard Vonderschmidtt was one of the early employees at RCA's solid-state division, during the company's heyday, and was elevated to run the division when it was making $500 million per year. Given his life and career thus far, building a hugely successful semiconductor division with a multinational company on the East Coast, Vonderschmidtt may not, at first glance, seem to belong in this pantheon of West Coast heroes. But allow me to explain.
Vonderschmidtt came to California in 1981 to run Zilog for Exxon, which had provided its founding capitalization. He stayed for three years before leaving, in disgust, as Exxon imploded in the computer field. One of the company's engineers, Ross Freeman, left with Vonderschmidt, and, together, they raised $4.25 million from John Doerr and several of Doerr's contacts. With that money, Vonderschmidtt and Freeman founded a new kind of semiconductor company called Xilinx, based upon the premise that the design of the microprocessor - or, more specifically, the logic employed by a designer in a chip's configuration - could be customized, as done with a full-custom chip (the freedom first given to designers by Corrigan and LSI Logic), but through the use of a common substrate of semiconductor product, more accurately referred to as a field-programmable gate array. In short, the designer takes logic-design software from Xilinx, applies a design to a base chip (the gate array), and programs the design onto a chip in a matter or days.
Similarly, the test function that often proves a full-custom design process has erred in some way is reduced to a matter of hours. In effect, the semiconductor revolution evolved into microprocessor, and, at Xilinx, the microprocessor has seemed to evolve to a stage where the customer, not the semiconductor company, determines to what use the silicon will be put. Today, Vonderschmidtt's company, which started with 600 gates (logic operations) on a chip, is about to produce gate arrays with 10 million transistors on a single chip. The company's four buildings, housing 2,500 employees, seem impervious to the current downturn faced by the general-purpose microprocessor-based companies, because the design of the microprocessor, according to Vonderschmidtt, is finally moving into the hands of those who will drive its future: the customers. It's an odd way to end this discussion of the microprocessor industry, thinking that the industry is about to change dramatically. But, after talking with the people in this story, you can well imagine that this is an industry likely to be reborn from within.
No hay comentarios:
Publicar un comentario