-
7. IBM: MOS AND THE VISIBLE HAND, 1967 – 1975
- Johns Hopkins University Press
- Chapter
- Additional Information
210 SEVEN IBM MOS A N D T H E V I S I B L E H A N D , 1 9 6 7 – 1 9 7 5 Intel, a small start-up, had a very tight organization with very few technology transfer problems. Its MOS process was developed in the same run-down Mountain View building where it would be manufactured. At IBM, the establishment of MOS memory technology involved at least five different sites and the transfer of people and technologies across thousands of miles. IBM’s MOS memory program achieved high-volume production at roughly the same time as Intel’s, but the path to that point was completely different. IBM is a classic case of the oligopolistic, vertically integrated company analyzed by Alfred Chandler. In IBM’s semiconductor operations, management coordination replaced market coordination. IBM’s semiconductor work had no direct contact with markets, allowing managers an extraordinary degree of freedom to direct work as they saw fit. Intel developed both bipolar and MOS technologies; it had no idea beforehand whether either would be successful. The success of MOS in the marketplace dictated that Intel would be primarily an MOS company. At IBM, the MOS versus bipolar decision was one made by a manager. Once IBM made the decision to use MOS memories in its computers, the success of the program was virtually guaranteed. IBM would pour whatever 211 IBM resources necessary to make the program a success and would become one of the world’s largest producers of MOS memories.1 At IBM, questions of coordination loomed large. It had to manage the transition from the older magnetic core memory technology to semiconductor memory without jeopardizing its enormous revenues. Managers had to determine how to marshal the various organizations and resources at their disposal. IBM did not use its resources with perfect efficiency and had many duplicative efforts and false starts, but even these served a purpose, by helping IBM to manage the risks of moving to the new technology and satisfying various constituencies within the company. When it was over, Burlington, Vermont, had emerged as the center of MOS work in IBM. Although IBM’s MOS program could be called a success, that success was qualified in two important ways. First, IBM’s philosophy of semiconductor production could only be practiced within the IBM environment, as shown by the failure of a start-up formed by IBM engineers. Second, IBM’s MOS success was confined to memory. MOS memory had clear relevance to IBM’s most profitable line of business, mainframe computers. But MOS logic was too slow for mainframe computer applications, and while IBM had many smaller systems where it could have been productively used, bipolar technology proved to be a satisfactory alternative. The Joint Program In 1967 managers from IBM’s Components Division (CD) and Research signed an agreement to conduct a joint program in MOS memory development , involving participants from three different divisions and four different sites within IBM. The main work in the program would be carried out by those outside the centers of power of IBM’s semiconductor work. Research was in this position, for it had made practically no contribution to the semiconductor technology currently being manufactured by IBM. It was joined by another group with a marginal mission, located in IBM’s Boeblingen, West Germany , facility. IBM Boeblingen had been established as a development site in the early 1950s to work on solid-state electronic devices. Germany was chosen to take [54.163.221.133] Project MUSE (2024-03-28 18:00 GMT) 212 T O T H E D I G I TA L A G E advantage of native physics talent, Boeblingen because of its proximity to an IBM manufacturing facility. The history of IBM Boeblingen’s electronics efforts up through 1963 paralleled IBM Research’s, with an unsuccessful foray into gallium arsenide.2 In 1963 IBM Boeblingen transferred the silicon planar technology used in solid logic technology (SLT) from CD in Poughkeepsie (later to become East Fishkill), New York. Boeblingen then used that base to develop products meeting local needs for variant semiconductor technologies, while the most critical and most widely used semiconductor products still came from IBM East Fishkill . Boeblingen developed a semiconductor crosspoint switch, to be used in telephone communications systems.3 Wolfgang Liebmann, who had led the Boeblingen effort to transfer the silicon planar technology from East Fishkill, came to Yorktown to lead the MOS joint...