Chapter 2 - Background
2.20 The Integrated Circuit -- 1959
Transistors represented a major improvement over vacuum tubes, but were not without problems of their own. Transistors came packaged as one transistor per each small “pot.” The pots were much smaller than vacuum tubes, hence more devices could be squeezed into the same space. But as the desired complexity of device interconnections kept growing, wiring all these small devices became an interconnection nightmare, and very costly. From the years 1952-1959, firms and governments around the world searched for an answer to the problem of interconnections.
The story of the revolutionary solution, the integrated circuit, begins at Bell Labs in the years following the invention of the transistor. As discussed, Bell Labs followed an open policy of diffusing transistor technology. Unwittingly, they also served as the most important source of trained personnel; personnel who saw the economic implications of the transistor. Two of the most celebrated to leave were Gordon Teal in 1952 and William Shockley in 1956.
Patrick Haggerty, the CEO of Texas Instruments (TI), recruited Teal in 1953 to be director of TI’s Central Research Laboratory. The charge was to innovate mass produced inexpensive transistors.79 Teal who had developed some of the earliest techniques to purify germanium, started work immediately to develop sufficiently pure silicon and with it to fabricate silicon transistors. In 1954, TI began selling silicon to others, and by 1956, had perfected a relatively inexpensive silicon transistor.80 (In 1954, a single silicon transistor sold for $23.95, while the only alternative, a transistor made of germanium sold for $3.56. TI’s goal of a $2.50 transistor took until 1963.81 )82
In 1958, TI made the propitious decision to hire Jack Kilby. Within two months, he conceived of the solution to interconnecting large numbers of transistors and other components. Kilby’s idea would come to be known as the”Monolithic Idea” – “a complete circuit would consist of one part – a single (“monolithic”) block of semiconductor material containing all the components and all the interconnections of the most complex circuit designs.”83 By September 1958, Kilby had by hand fabricated an “integrated circuit.” TI filed for a patent in February 1959. But Kilby was not alone. For best deserving the credit for this revolutionary innovation is Robert Noyce.
Robert Noyce’s story also begins at Bell Labs. In early 1956, William B. Shockley, left Bell Labs to start Shockley Transistor Laboratories in Palo Alto, California – the future Silicon Valley. Shockley then recruited legendary names in the history of semiconductors, including Robert Noyce, Gordon Moore and Jean Hoerni, to join his firm. But Shockley was no executive, and before long, eight of the recruits were terribly dissatisfied, and made it known they would prefer a new home. The new home came not, however, from being employed by an existing company, but as owners of a new firm, one funded with venture capital. It was to be the legendary, virtual fountain of future semiconductor companies, Fairchild Semiconductor Corporation, formed in early 1957.84
Noyce is considered the father of the integrated circuit because he not only conceived of the Monolithic Idea, as had Kilby, but also the means of manufacture - the planar process. Fairchild’s patent was filed in July 1959. The problem of interconnecting transistors had been solved. Ever since, the path of innovation has been to make device and interconnection features smaller, and the resultant integrated circuit, or chip, bigger.
The integrated circuit was not an overnight success for one simple reason: they cost too much to make. Then in May 1961, President John F. Kennedy challenged the imagination of the American public to put a man on the moon. To do so would require the use of integrated circuits. Through 1964, purchases of integrated circuits for the Apollo space craft computer and the Air Force Minuteman ICBM [Intercontinental Ballistic Missile]guidance computer drove the market for integrated circuits.85 See Table below. Once again, government support proved essential to market lift-off.
Exhibit 2.23 Government Purchase of Integrated Circuits 1962 - 1968
|Year||Total Integrated Circuit Shipments ($ millions)||Shipments to federal government($ millions)||Government share of total shipments (%)|
Source: Richard R. Nelson, Government and Technical Progress: A Cross-Industry Analysis, 63.
Even though the government had committed two critical programs to integrated circuits, into 1963 there remained sharp debate as to whether integrated circuits were the ultimate solution. But by then the costs of manufacturing integrated circuits were in steep decline, due to the volume purchases by the government, and any doubt as to their being produced or reliability was dispelled.
A new computer start-up, Scientific Data Systems, founded in 1961 by Max Palevsky, was the first to introduce a computer using integrated circuits. The SDS 92 shipped in 1964.86 IBM did not ship a computer using integrated circuits until 1969.87 The stories of both SDS’s and IBM’s decision to use integrated circuits will be told below.
Meanwhile, corporate executives – the customers leasing or buying computers from the growing computer market-structure – were being challenged to change how they managed the “modern” corporation, and how to use computers to do it. That story is next.
Reid, p. 62
Ibid., p. 43
Nelson, p. 36
The productivity of semiconductor technology can be appreciated when 40 years later, one million transistor microprocessor chips were selling for around $1,000 – not $23 million!
Reid., p. 23
Nelson, p. 73
Nelson, p. 62
The state of integrated circuit technology in 1964: “The early chips packed half a dozen or so active elements in an area half an inch square.” Business Week, Jan 1970, p.71
The Chip, p. 126