To the Digital Age: Research Labs, Start-up Companies, and the Rise of MOS Technology

6. The End of Research: INTEL AND THE MOS TRANSISTOR, 1968 – 1975

167 SIX The End of Research I N T E L A N D T H E MOS T R A N S I S T O R , 1 9 6 8 – 1 9 7 5 The year 1968 was pivotal in American history. The country was in almost constant turmoil for the first eight months of the year, from the Tet offensive beginning in January, to the assassination of Martin Luther King Jr. in April, to the assassination of Robert Kennedy in June, to the riots at the Democratic convention in August. The election of Richard Nixon marked the end of a period when Democrats had held the White House for 28 out of the previous 36 years, and it ushered in a period where Republicans would hold the presidency for 20 out of the next 24 years.1 While the general public did not realize it, 1968 was also a year of turmoil and reorientation in the semiconductor industry. The turmoil, part tragedy, part comedy, came through management upheaval at Fairchild that rippled through the industry. The year marked the end of Fairchild’s domination of the semiconductor industry and the formation of Intel, which would remain the most important semiconductor firm up to the time of this writing.2 Intel’s success was built on MOS technology, and so it is important to be precise about Intel’s achievement. Intel was not the largest producer of MOS integrated circuits at any time until 1975, but Intel was the first to put into production the silicon gate MOS process, an extremely powerful method for 168 T O T H E D I G I TA L A G E making MOS transistors. And Intel was the leader in the most important market for MOS technology in the 1970s: semiconductor memories. In that area it achieved first mover advantages.3 Although these advantages would last for only a single generation in that market, Intel was also extremely effective at creating a variety of new products that exploited the possibilities of the silicon gate MOS process. These new products, including the erasable programmable ROM and the microprocessor, formed the foundation of its later success. Intel had an ambivalent relationship to its Fairchild past. Fairchild provided a large proportion of the people who worked at Intel, and it was at Fairchild that they had learned their skills. But the founders of Intel made a conscious attempt to transcend the organizational problems that had plagued Fairchild, most notably by not having a separate R&D lab, even though the first four people at Intel included the director and assistant director of Fairchild R&D. Intel’s work on silicon gate MOS technology represented a synthesis of R&D and manufacturing approaches. Although the silicon gate process had been a project of Fairchild R&D, the key figure in developing Intel’s silicon gate process was from manufacturing, and his experience in manufacturing proved crucial to the success of Intel’s process. Intel’s early history is a classic story of a Silicon Valley start-up, serving as an exemplar of Gordon Moore’s claim that start-ups are not better at creating new things than existing companies; instead, start-ups are better at exploiting new things. When Intel started, a great deal of uncertainty surrounded MOS technology —no one knew if there would ever be successful MOS products or what they would be. Flexibility was key. Intel was able to conduct a series of economic experiments by quickly producing a range of products and then pouring resources into those that succeeded. The resources available on the San Francisco Peninsula enabled Intel to move quickly and conserve capital.4 Moore’s Law, the Disintegration of Fairchild, and the Founding of Intel In 1965 Gordon Moore wrote an Electronics article that subsequently became famous as the original statement of Moore’s Law. While this article on the future of integrated circuits included such optimistic predictions as home computers and personal communications equipment, it also implicitly recognized 169 The End of Research a daunting problem for companies that wanted to make money producing integrated circuits. The first eleven years of Fairchild’s history revealed this ambivalence, for even though Fairchild had made a string of major innovations , it was unable to secure a lasting competitive advantage in the industry.5 A key section of Moore’s 1965 article was based on an analysis of the cost of integrated circuit production. He claimed...