chapter 9
Processing Power for All
The rise of the PC
“What the hell is it good for?”⁸⁴
Robert Lloyd, engineer at the Advanced Computing Systems Division of IBM, on the microprocessor, 1968
“There is no reason for the individual to have
a computer in their home.”⁸⁵
Ken Olsen, president of Digital Equipment Corporation,
Convention of the World Future Society in Boston, 1977
“640k [of computer memory] ought to be enough for anybody.”⁸⁶
Bill Gates, founder and CEO of Microsoft, 1981
The roots of the PC
DEC had exploited the technology of integrated circuits with the PDP–8. In doing so, it provided a solution to the needs of a whole class of user. The market gap that it filled had been created by the industry itself. In economic terms, the demand for computing power was highly elastic – that is, highly responsive to price. Demand also responded to simplicity and product portability, but the initial reaction was to price, particularly price points that allowed the displacement of timeshare access to mainframes. This had been the secret of the success of DEC, but the logic of this success was not taken further by either DEC, IBM or any of their major competitors. The logic suggested a potentially enormous market for new products at cheaper prices, but it was not pursued.
The development of the personal computer market was therefore effectively left to chance and grew out of seemingly unrelated efforts in a somewhat haphazard manner. Unlike some technological advances, where replacement of existing methods was the target and hence the commercial yardsticks were fairly obvious, the story of the personal computer is more akin to the development of broadcasting as a by-product of the radio. The original intent of Marconi and other early pioneers was to replace wire-based one-to-one communication. The radio as a medium for broadcasting was recognised and developed initially by ‘amateurs’ before the potential commercial attributes became obvious and attracted existing media players to the market.
In the case of the radio, amateurs required the improved productive efficiency and hence the lowered cost of valves to pursue their hobby. For the personal computer, a number of elements were vital for the market to emerge. Firstly, processing power and memory capacity had to be expanded and reduced in cost. Secondly, software had to be developed. Thirdly, peripherals had to become available at reasonable cost. These requirements were no different from those that faced producers in the 1950s and 1960s, but the difference was that they no longer needed to be supplied by a single company. This represented a sea change in the industry since the availability of capital – either corporate or through government contract – was no longer one of the pre-determining factors for entry. Capital was still important, but the scale of what was required was substantially reduced.
Intel, the company which became synonomous with the semiconductor or ‘chip’, was instrumental in making possible the development of personal computing. It emerged from work that was done by two other pioneering semiconductor companies, Shockley Semiconductor and Fairchild Semiconductor. The ‘mother ship’ of the semiconductor industry was built in 1957 in Mountain View, California, and was the result of Robert Noyce and Gordon Moore becoming dissatisfied with the direction of William Shockley’s company. They were unable to persuade the principal financier of the merits of their vision of the future and found that their only alternative was to set out on their own.
Their first attempt to raise capital consisted of drawing up a list of potential companies that might conceivably have a desire to enter the semiconductor business. Led by the investor Arthur Rock, the investment bankers contacted all the companies on the list. The success rate was low. No funding was forthcoming; moreover, not one of the companies was willing to even see the group. The search was broadened, and eventually funding was provided by Sherman Fairchild and the Fairchild Camera and Instrument Corporation. On the back of its new semiconductor division, Fairchild was to become a huge commercial and stock market success. Its semiconductor operations rapidly proved successful, leading Fairchild Camera and Instrument to exercise its option to take 100% control in late 1959. That year the sales of Fairchild-produced semiconductors had grown tenfold. The years that followed demanded substantial investment in R&D and production facilities, and Fairchild retained its lead as the largest producer of silicon transistors in the world. The company would continue to prosper with the impetus of this new growth area and the value of its shares on the stock market more than reflected this achievement.
9.1 – Fairchild Camera and Instrument: glory to despair
Source: CRSP, Center for Research in Security Prices, Graduate School of Business, University of Chicago, 2000. (Used with permission. All rights reserved. www.crsp.uchicago.edu.)
The company’s commercial viability was sustained in the early years in much the same way as that of the other computer companies had been: by government contract. Fairchild benefited directly from the Cold War, supplying integrated circuits for the electronics of intercontinental ballistic missiles (ICBMs). Its first major contract was for the Minuteman project. The operations at Fairchild represented the cutting edge of semiconductor design and manufacture; as such, they effectively became the industry training ground for semiconductor engineers and entrepreneurs, who would hone their skills at the company before moving to join or set up competitors. Among the companies whose operations emerged in what became known as Silicon Valley were National Semiconductor, AMS, Teledyne, Rheem and Signetics.
9.2 – The rise and fall of Fairchild Semiconductor
Source: New York Times (dates in image itself).
While many of these companies became important, the defining moment came in 1968 when Robert Noyce and Gordon Moore left to form Intel. For Noyce and Moore, Fairchild had become a frustrating place to work. A series of management changes had left them unable to control either their own destiny or the technological thrust of the company. Fairchild had become a victim of its own success. While it might have spawned an entire industry, its ultimate fate was to be a distressed sale to Schlumberger, the oil service and equipment company, in 1979, and after further decline, a further sale to National Semiconductor in 1987. Fairchild’s shares had been one of the top-performing securities of the period; its later decline reflected the decline in the company’s fortunes.
The birth of Intel
Moore and Noyce, colleagues since the Shockley days, left Fairchild in order to achieve greater control over the direction of their efforts. After discarding ‘Moore Noyce Electronics’ because of its phonetic connotations, the partners settled on Intel, an abbreviation of Integrated Electronics, for their new creation. Intel was formed in 1968 with seed capital of roughly $250,000 ($1.2m) provided by Moore and Noyce, augmented by a further $2.5m ($12m) from Arthur Rock. Rock had been involved in numerous industry start-ups in the nascent industry, including Fairchild, and was an enthusiastic advocate. From a very early stage he was convinced of the future success of the company.
The hurdle of perception was a relatively low one because Moore and Noyce had already demonstrated their ability. Perhaps more importantly, the commercial feasibility of what they were proposing had already been accepted by the financial markets and reflected in the Fairchild share price. They also had the support of Rock, whose track record was such that he was able to raise the funds for the new venture in two days simply by phoning his established business contacts.
Intel’s original intent was to focus on memory chips, but the one-page ‘business plan’ was more a statement of intent than a fully worked out document. What Intel was selling was the ability of its founders to produce new and complex integrated circuits. One avenue was memory chips, but the main driver would be from the emergence of a new use for integrated circuits. Intel was not alone in its desire to find new uses for integrated circuits which would increase the volume of production and hence improve productivity. Nor were they alo
ne in recognising the need to make use of the chips in such a way as to demonstrate their use and act as a powerful marketing tool for further developments.
What Intel really needed was a source of volume demand for its proposed products. The source of this volume demand would emanate from the demands which had originally stimulated the work of Charles Babbage, namely the need for accurate computation. Adding machines had increased in size and complexity as the technology shifted from analogue to digital computation. Mainframes and later minicomputers were able to undertake complex calculations with unprecedented accuracy and speed. This provided great assistance for government with its vast data processing requirements in the areas of defence and population study. However, for the rest of the population, the value was limited. Some electronic adding machines had been produced, but they possessed only very basic functions and, as pieces of precision equipment, were expensive to produce and purchase. This left a potential enormous market if an adding machine or calculator could be produced at a reasonable price.
The calculator – accidental mass market product
In many ways the calculator can be seen as the natural culmination of the efforts of the scientists of the previous century. Two great tasks of government – public finance and defence – relied upon mathematical calculation, whether it was the calculation of taxes under Napoleon, or code-breaking under Churchill. As a consequence, governments throughout the world had a powerful interest in funding and encouraging the development of the ability to compute. The science of ballistics during and after World War II took this need to new heights and the level of sophistication required increased in parallel. Huge, expensive mainframe computers were built and immediately the search began for improved versions. The use of computers spread from the government, military and academic sectors and began to play a role in private companies, just as adding machines had done 50–60 years before.
With the advent of transistors, and then semiconductors, the way was paved for a reduction in the physical size of computers. DEC brought in the minicomputer to rival IBM’s dominance in mainframes. Semiconductor research became an industry in its own right. Grosch’s law had been undermined by the emergence of the minicomputer and was to be made obsolete by the development of the microprocessor. As scientific advances followed upon each other, a new law was to emerge which was the antithesis of Grosch’s law. The rate of progress embodied in these advances was neatly summarised as Moore’s law, which first saw the light of day in the anniversary edition of Electronics magazine in 1965. Roughly stated, this asserted that the complexity of computer chips would double roughly every year. As one of the founders of Intel, Gordon Moore himself helped to bring this to pass, but the latent demand for computing power was the decisive factor in making it happen.
The electronic calculator is an example of a technological advance that started out as pure science, and was then forced to find a commercial outlet which had not hitherto obviously existed. Before the calculator, the slide rule was the principal available tool for small-scale calculations. This was an aid, but using it remained laborious. The earliest electrical calculators were pieces of precision engineering requiring complex gear and transmission systems. The main manufacturers were based in developed economies – the USA, Germany, Italy and the UK – where the necessary skills and equipment existed. Japan had no real presence until the Kashio family sought to move its business into what they perceived as a new growth area. Hitherto their main product had been a metal ring which fitted over the finger and held a cigarette. This allowed the smoker to puff away at his cigarette while remaining free to smoke and work at the same time. The key selling point was that, given the high cost of tobacco, smokers could now smoke their cigarettes down to the butt without burning their fingers! Despite such improvements as plating the rings with chrome to enhance their appearance, demand for this curious product eventually disappeared.
In 1950 Toshio Kashio designed a solenoid-based electrical calculator, but abandoned this in favour of an electrical relay version that avoided the noise problems of existing electro-mechanical machines. After seven years of perseverance, his model was completed. It was approximately three feet square and weighed about the same as a sumo wrestler. The calculator was priced at a discount to its imported competitors, and as it became cheaper, and the Japanese economy started to grow rapidly, sales took off. Growth continued apace and the corporation broadened its product range to include scientific calculators. The company, Casio, became increasingly prosperous. Its future seemed rosy.
This idyllic situation was shattered in mid-1964 when Sharp introduced the first electronic calculator to Japan. Overnight, Casio’s relay calculators had become technologically redundant. Had the company stayed abreast of technological developments taking place in America, this would not have happened. However, as the chairman of the company subsequently admitted, success and the profits this generated had bred a degree of complacency, which manifested itself in thrice-weekly family fourballs at the local golf course, rather than attention to their own engineers’ warnings of the growing threat from the new electronics industry. Drastic retrenchment allowed the company to survive. Many of its competitors proved unable to react with such speed and disappeared. In 1967 Casio produced a desktop electronic calculator. This had now slimmed down in size from the weight of a sumo wrestler to that of a small child.
The product that Sharp brought to the market can also be traced back to skills developed during World War II by both the Allied and Axis powers. During the war an engineer named Tadashi Sasaki was employed by the aircraft manufacturer Kawanishi Kobe to work on antiradar devices. During the postwar period, this company merged to become Kobe Kogyo. It later became part of Fujitsu in the early 1960s. Sasaki was a key figure in the development of the Japanese semiconductor industry. He was a visitor to Bell Laboratories and corresponded with John Bardeen, co-inventor of the transistor. Sasaki understood immediately the significance of the transistor and instituted a research programme at Kobe Kogyo.
Kobe Kogyo was also a supplier of electronic components to Sharp, at that time a manufacturer of domestic appliances. At Sasaki’s instigation, Sharp sent a number of employees to retrain as part of an attempt to enter the calculator business. In 1964, when Kobe Kogyo had been absorbed into Fujitsu, Sasaki moved to a senior position at Sharp, where he could oversee its entry into the calculator industry and the production of the first transistorised calculator. Sasaki kept a close eye on developments in the US semiconductor industry, but was unable to break the consensus within Sharp that it was better to proceed with a programme of incremental improvement to existing circuits than to try and reduce all the required functions to a single chip. Sharp’s supplier of chips, Rockwell, refused Sasaki’s request to develop such a chip on the basis that it would divert valuable resources away from its existing profitable semiconductor business lines.⁸⁷
Sharp had been visited by Robert Noyce of Intel in 1968. Given Noyce’s background with Fairchild, Sasaki was keen to see if Intel could produce under contract the technology that Rockwell did not wish to pursue. The suggestion was made, but Rockwell, which had an exclusive contract, rejected the request from Sharp. Frustrated by his inability to promote this line of development within Sharp,⁸⁸ Sasaki decided to provide financial assistance to a small Japanese company to whom he had provided technical input in the past. The president of this company, Business Computer (Busicom), was an alumnus of the same university department as Sasaki. Sasaki provided ¥40m (equivalent to roughly $120,000 at the time, or nearly $0.5m now) for Busicom to contract Intel to produce the chip that Sasaki had failed to promote within Sharp.
Desktop electrical calculators had been produced through the 1960s in America, but their cost meant that their use was limited to scientific and high-level business applications. The Mathetron, a programmable desktop electrical calculator produced by a division of the Barry Wright Corporation, was one of the first models in the United States. It sold for nearly $6,0
00 ($33,000) in 1964. The Hewlett-Packard 9100B, designed for scientists and engineers, appeared in 1968 at a price of $4,900 ($23,000) and the Sharp Compet 361R shortly afterwards. These machines were invaluable to the scientific and business communities in that they greatly aided computational tasks but avoided the need for access to computers. These machines, like the majority of the market, were targeted at particular subsets of customers, users whose day-to-day existence typically involved heavy computational tasks.
Economic imperatives
The broadening to a more mass-market appeal that would underpin the success of handheld calculators arose not from the understanding of their latent appeal as much as the need to find ways to improve the economics of chip production by creating volume markets. At Texas Instruments (TI) in 1965, Jack Kilby, one of the inventors of the integrated circuit in 1958, put forward with two of his colleagues the suggestion that a handheld calculator be produced. The reasoning was that such a device would do much to broaden the appeal and demand for the integrated circuit. Two years later, in August 1967, a battery-powered calculator, the ‘Cal-Tech’, which could produce results on thermal paper, was produced. This signalled the start of a brand-new industry. The original impetus came mainly from the industry’s drive to produce complex circuits at low cost.
Engines That Move Markets (2nd Ed) Page 44