Texas Instruments was not alone in following this path. In 1968 Bill Hewlett asked his research staff to produce a pocket-sized calculator which could perform all the calculations of a slide-rule. The calculator was completed in early 1971, followed by the scientific calculator, the HP-35 priced at $395 ($1,400) in 1972. Meanwhile TI was negotiating with Canon in Japan for the commercial production of its prototype ‘Cal-Tech’, produced by Kilby. This was introduced as the Pocketronic in 1971, selling at $345 ($1,300). Thus arrived the calculator revolution, a revolution which seemed to take pace at whirlwind speed. The early profitability and mass-market appeal of a device which replaced archaic manual calculation devices attracted hundreds of competitors.
For Intel the attraction of the market was unavoidable. Its great need was to find a group that would finance the development of complex chips and guarantee volume production and sales. The new calculator market was not only a growing volume market, but one in which chip design was the key to success. Unfortunately for Intel, the main producers of calculators were already either self-sufficient in chips or had established relationships with other chip suppliers. As a consequence, Intel ended up in a contract with the Nippon Calculating Machine Company, a Japanese company which operated the Osaka factory for Busicom. Busicom had a number of business strands: on the one hand it imported computers and software from France to Japan. On the other it sold Busicom calculators to the USA on an original equipment manufacture (OEM) basis under the NCR brand.
Prompted and quietly funded by Sasaki, Busicom sought to enter the new small calculator market by developing a new calculator in collaboration with one of the few US semiconductor manufacturers not exclusively tied to a producer. The agreement reached in February 1970 reflected the bargaining powers of the two companies. In return for development finance and a guaranteed output volume, Intel granted the intellectual rights in the chip to Busicom. Article one of the agreement defined the minimum properties of the chip and article two assigned exclusivity on the chip to NCM or Busicom. Having obtained a contract for a processing chip, all that Intel had to do was design and manufacture one.
The development process was accompanied by headaches. The Intel chip designer, Ted Hoff, did not subscribe to Busicom’s detailed design plans. He felt they were out of date, and reflected thinking that had been superseded by technological advances, many of which he had had seen for himself through his work on the DEC PDP-8. He believed that Busicom’s plans would lead to a product that was overly complicated and too expensive to provide the core of a competitive calculator. He was correct, a fact that the Busicom representative graciously conceded.
9.3 – An expensive scientific instrument: early calculator advertisements (Sinclair and HP35)
Source: G. Ball and B. Flamm, The Complete Collector’s Guide to Pocket Calculators, California: Wilson/Barnett Publishing, 1997.
However, being correct in ascertaining the flaws in an existing design and being able to produce a superior replacement are two different things. Eventually, under the auspices of Federico Faggin, with the assistance of the Busicom representative Masatoshi Shima, a new improved chip, named the 4004, was produced. The 4004 contained more than 2,000 transistors and could perform over 60,000 executions a second. The new chip was still not viewed within Intel as a mainstream product, despite the fact that the funding from Busicom was vital to the company. Intel continued to concentrate its resources on the memory-chip side of its business.
Whether Busicom’s exclusivity contract devalued the chip’s commercial importance within Intel is a moot point. At the time the memory business was believed to be the main future profit centre. Microprocessors were little more than a sideline, a means to assist the sales of memory chips. Microprocessors were to become the cornerstone of the industry over the 30 years that followed, and so it is difficult to find too many of the early participants willing to admit their opposition to the development of that particular business with the company. It is clear that the potential of the microprocessor was not apparent at the time, and there was a great deal of unease about the amount of resources devoted to a product with no obvious market at a time when the sales environment was very difficult.
Uncertainty over the correct course of action ran throughout the company, up to and including the board of directors. Even when Busicom hit financial problems, and was forced to give up its exclusive rights over the new chip, it was by no means a unanimous decision of the board of directors to develop and promote the product. A division occurred between those wishing to stick to memory chips, and those who wished to pursue the microprocessor. The case for the microprocessor was helped by the argument that, should more be sold, it would inevitably increase demand for memory chips, the main product. As a note of side interest, the cost to Intel of regaining control of the 4004 chip was just $10,000 ($40,000). This only emphasises the lack of understanding of the potential for microprocessors. This was not to last long. Intel began to market the new chip in late 1971 and by 1972 the potential was becoming apparent, with both the technical press and customers displaying interest. Competitors such as National Semiconductor and Rockwell were also beginning to follow the same path.
The marketing concerns within Intel about the 4004 chip revolved around whether the size of the potential market for the microprocessor justified the expense and effort required to support it. The biggest effort needed was in the development of applications and the need to teach potential customers how to program these applications on the chip. By now Intel also had a second powerful new chip, the 8008, to consider. This chip had been developed for Computer Terminal Corporation, but had later been orphaned when CTC switched the contract to Texas Instruments.
What the chips lacked most of all was a language to convert instructions into signals the chips would recognise and implement. In 1972, a professor at the Naval Postgraduate School in California named Gary Kildall wrote a language capable of taking instructions through an IBM 360 and translating them into commands for the 4004. Kildall was employed to produce a high-level programming language – PL/M – which could be applied not simply to the 4004 and the 8008, but also to a future family of Intel processors, should this eventually arise. Over time, Kildall’s efforts eventually turned into an operating system called CP/M. The future for microprocessors was still viewed with some ambiguity within Intel. This was demonstrated by the fact that Kildall’s request to sell CP/M on his own met no resistance and was not deemed to have any meaningful commercial significance.⁸⁹ Even the pioneers who had backed their own judgement by creating a new company and a new product were unable to envision how the industry was to unfold. Kildall was also unaware how rapidly the industry was set to explode. In the late 1970s, exasperated by the work of running a commercial organisation, Kildall offered his partners his interest in the company, Digital Research, for $70,000. Within a few years of this offer, Digital Research had CP/M installed in over 200,000 computers and revenues in excess of $6m.⁹⁰
Intel
The early years at Intel were ones of capital consumption as the company sought to produce new products. In 1969, for example, R&D plus selling, general and administrative costs (SG&A) amounted to more than five times gross revenues. In 1970, sales had grown tenfold, but although losses were still being made, any concerns investors might have had would have quickly dissipated. The rest of the 1970s were years of pretty much untrammelled growth. Moreover, the growth in revenues was not achieved at the expense of margins, which at an operating level consistently exceeded 20%. Net income also grew at a rapid rate. As a consequence, after the initial years Intel’s balance sheet showed none of the weakness frequently experienced with new technology companies.
One reason was that Intel was a spin-off from an existing successful business, and its founders were pioneers of the industry. The technology risk was therefore minimal, the critical question being more one of commercial exploitation and the grasping of new opportunities. New opportunities certainly existed and the compan
y attempted to exploit them, whether they resided in memory chips, digital watches or processor chips for calculators. The annual reports of the company showed that the microprocessor was not the main focus of the company in the early years.
Eventually the personal computer appeared and began to lift Intel to a new level of growth. Although growth had been rapid during the 1970s it had not always been smooth, reflecting the changes taking place in the industry and the poor underlying economic conditions. By 1980, the outlook for the company might not have looked as exciting as it had in prior years. Competition was increasing, and sales growth had slowed, bringing lower margins and reduced profits. The economic backdrop was one of high inflation and volatile interest rates. The reaction from Intel was to try to raise barriers to entry.
The company strategy had been based in part on timely delivery of product, but this was augmented by the use of litigation, as a defensive tool in reaction to copyright infringement, and as an offensive tool to scare away potential competitors. For possible new entrants, this was not an insignificant threat. Pure technological superiority was less important. There were numerous occasions where Intel managed to retain its market lead without having the leading product, not least in its battles with Motorola.
9.4 – Intel: the payoff from Moore’s law
Source: Intel annual reports. 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.)
Just as competition in memory chips hotted up, the personal computer industry was about to be transformed by the introduction of the IBM PC. What followed was rapid and accelerating growth for Intel in microprocessors but gradual deterioration in the memory market. Intel’s supremacy had been usurped by Asian and Japanese producers, who produced better memory chips with lower wastage rates. Moreover, monetary conditions in Japan had fostered an environment where capacity expansion was no longer constrained by lack of capital.
The net result was that Intel saw a rapid decline in profits as competition increased and memory chip prices fell. The industry might have been a growth one, but this did not protect it from exposure to the economic cycle and fundamental supply-and-demand conditions. In the microprocessor area, IBM had required dual supply, as a result of which AMD obtained a contract to be the alternate supplier to Intel, a position ultimately led to costly litigation on both sides. In the mid-1980s, increased competition took its toll and Intel plunged into loss. For the investor, conditions would not have appeared appetising, with increased competition in most product areas on the one hand and higher capital expenditure requirements on the other.
In practice, this difficult period presaged the beginning of another chapter in the Intel success story, with a move from PCs being branded to a situation where each PC box contained new branded products: the Microsoft operating system, Windows, and the Intel microprocessor. This shift was decisive in making PC production a commodity business and microprocessors a branded one. It is no coincidence that Intel stopped labelling chips by number and started giving them names. The growth of Intel was fuelled by the overall growth of the PC market, and enhanced by ever lower costs of production and the eventual expansion of the Internet and home computing.
This is not to say that the industry’s cyclicality was removed. Indeed, for Intel, it probably increased, there being no hiding place for such a dominant supplier should there be a turn down in end-user demand. Furthermore, not only was Intel a proxy for the general condition of the personal computing market, the capital intensity of production continued to increase sharply, meaning that overlaid on cyclical sensitivity was a higher level of operational gearing. This means that in a downturn, the impact on profits will be very similar to companies in the traditional cyclical industries.
Not only has Intel become more cyclical, it has largely missed out on the growth area of mobility, allowing ARM to gain leadership. Moreover, it now faces the potential move to ever more specialist chip designs reflecting the growth in data science which flows from Internet usage and data collection. Increasingly data companies such as Google and Facebook are likely to want bespoke chip design. It is not hard to see Intel eventually morphing into primarily a chip fabrication business – highly capital-intensive, profitable but cyclical.
From calculators to the PC
In the early stages, the marketing department of Intel was correct. The principal use of microprocessing chips lay in the increasingly competitive calculator market. The Japanese companies Busicom, Sharp, Casio and Canon all manufactured or assembled calculators in association with US companies, such as Intel, Rockwell and Texas Instruments. They were joined by North American companies such as Hewlett-Packard, Commodore, Bowmar Instruments, and the Micro Instrumentation Telemetry Systems Company (MITS). Some of the microprocessor manufacturers, such as TI, also produced their own models directly. In Europe, a slide-rule manufacturer – Aristo, Dennert and Pape – sought to protect its 100-year-old heritage by joining the handheld calculator market decimating its traditional business. Its first calculator was based on a TI chip and retailed at DM480 ($200 or over $600 now).
Demand in the early years was high; Hewlett-Packard found that its first scientific calculator, the HP-35, could retail at $395 ($1,500) and still result in a four-to-five-month order backlog. The price might have reduced from the equivalent of a family car, but it remained roughly 10% of the annual income of a skilled employee. The early chips held an equivalent processing power to the massive ENIAC machine completed barely a quarter of a century before. The demand quickly brought in new competitors and the calculator market became a ferociously competitive one as advances in chip design allowed manufacturers to lower their prices repeatedly.
9.5 – All about the cost curve: the inexorable decline in calculator prices (1968–2000)
Source: G. Ball and B. Flamm, The Complete Collector’s Guide to Pocket Calculators, California: Wilson/Barnett Publishing, 1997.
It was not from technology advances alone that the problems stemmed. The chip manufacturers realised that the companies selling the calculators were making profits from a product which, in reality, was simply their chip in a small plastic wrapping. Not surprisingly, some manufacturers decided to try and take this profit for themselves by producing their own product, backed up with aggressive pricing. In a market where volume production was so important, casualties were unavoidable. The initial feast had simply brought too many hungry suppliers. The price of a scientific calculator was to halve three times in less than three years, from nearly $400 in 1971 to $200, then to $100 and in 1974 to under $50. MITS found itself trying to sell a self-assembly kit for $99 against an equivalent assembled product selling at less than half the price. The financial consequences were inevitable.
So far as MITS was concerned, an alternative product had to be found if the company was to remain solvent. Ed Roberts, the president of the company, decided that there was no future in calculators and sought to find other uses for the Intel chips that had formed the guts of his calculator business. The target was to build a new product, differentiated from what already existed, where there would be, at least for a period, a degree of market protection and better profitability. Rogers rejected the earlier Intel chip designs as unable to provide the capability he required. When Intel produced the 8080 chip, however, it appeared to have overcome the problems of execution speed associated with its predecessors, and thus had the potential to become more than just a programmable calculator. Out of the near extinction caused by the oversupply of chips and the calculator price wars was to emerge a product of intensely more lasting significance. This was the personal computer.
The big computer manufacturers had a number of things in common. Firstly, most of them had the capability to produce a personal computer. Secondly, most of them also had within their organisation at least one champion of such a machine. In many cases, that proponent had produced either plans for a small computer or a rudiment
ary working model. At Hewlett-Packard, a design by a young engineer named Stephen Wozniak was rejected, not because of concerns over its feasibility but because the potential market was not immediately apparent. Equally, at Digital Equipment, David Ahl, the former head of the educational products division, put forward a proposal to develop a machine which could be sold to schools and other educational users. A prototype was developed but foundered due to lack of support from senior management.
Again one of the main obstacles was how support for such a machine could be supplied economically. Large companies or public organisations had their own internal support groups which could program and operate the computers. Small ones did not and companies such as DEC, HP, IBM and Control Data could see little possibility of a viable business model being developed. Early development of the PC did not therefore take place at the organisations best placed to move it onto a commercial footing. Rather, it was left to enthusiasts to develop their hobby into a viable business. In this, there were many similarities with the development of the radio, where for periods the amateur enthusiasts effectively sustained the non-military side of the business.
Creating an industry
In terms of market impact, it was Ed Roberts’s failing calculator self-assembly company, MITS, which set the early pulses racing. As the calculator price wars decimated the industry, Roberts was faced with two options. He could watch his company slip into liquidation, or he could develop a new product which had some protection from the vicious price competition. He chose the latter course. Like all new product launches, success depends in part upon the product itself. However, marketing is also vital. If a product is in any sense ‘groundbreaking’, as Roberts intended his to be, marketing moves from being important to critical. During the mid-1970s, a number of magazines were avidly read by computer disciples. One of these was Radio Electronics (RE), another Popular Electronics (PE). In July 1974 RE published a feature about a computer called the Mark-8, constructed around the Intel 8008 chip. Roberts had examined the possibility of using the 8008, but was dissuaded by the microprocessor’s limitations. These limitations did not dampen the public reception to the Mark-8, but did fatally undermine its commercial appeal.
Engines That Move Markets (2nd Ed) Page 45