Inside Intel. Tim Jackson

every Tuesday morning at 9 a.m. to hear a succession of predictions, announcements, questions, instructions. Sanders used the meetings to keep track of how different departments were progressing with different projects. But he gave his colleagues little opportunity to raise issues of their own at the meetings. These he preferred to deal with one to one, in his own office.

      But it was in the board meetings that Sanders’s instinctive political skill was most evident. Skornia, who served as company secretary as well as general counsel, began to notice how well the AMD chairman handled his fellow-directors. To ensure that his proposals were smoothly and speedily adopted in the board meetings, he would prime key board members in advance of a difficult decision. Only once in the first year, when he made an ill-judged proposal for AMD to take over a smaller electronics firm, was Sanders forced by his fellow-directors to retreat.

      In 1971, however, Sanders began to shake out his organization. One day he summoned Sopkin into his office and began asking him questions: ‘Elliot, do we have security guards on duty during the day?’

      ‘Sure, Jerry. You know I look after that.’

      ‘You know what, it wouldn’t hurt if we had a guard on duty right up here this morning.’

      ‘Where would he be?’

      ‘I think he’d be walking between your office and mine, a bit closer to yours.’

      ‘When?’

      ‘About ten.’

      At 10 a.m. precisely Sanders called Frank Botte, one of his seven fellow-founders, into his office, and closed the door behind him. He told Botte that he was performing below par, and fired him on the spot.

      Five minutes later a strangled sound emerged from the office. Blazing with anger, Botte had grabbed Sanders by the throat and was squeezing as hard as he could. It required all the strength of the security guard, whose presence Sanders had carefully prearranged, to separate the two men.

      Later in 1971 a similar fate befell Jack Gifford, the leader of the original group of four who had come to Sanders with the idea of building analog circuits. This time, the sin that the victim was allegedly guilty of was treason, pure and simple. Sanders believed, rightly or wrongly, that Gifford had tried to organize a palace coup against him and to take the chairmanship of AMD into his own hands. With a few well-placed phone calls, Sanders made sure that AMD’s directors realized how important it was to remove this potentially divisive influence – and with the board’s approval, it was not long before Gifford, too, left the company.

      Thanks to the success of the engineers’ second-sourcing efforts, and Sanders’s indefatigable sales campaign, AMD was able to go public less than a year after Intel. Its prospectus, issued at the end of September 1972, showed that the company was making money – just. But there were a few surprises. Because Sanders had fired him without claiming cause, but then tried to force him to sell his AMD stock back to the company at ten cents a share, Botte had sued – and the case, worth over $1m, was still pending.

      More interesting was what had happened to the company’s share capital. Unable to make its startup funding last all the way to the IPO, AMD had gone back to its investors in a second round for another $650,000 – ostensibly to follow Intel’s lead, and diversify from AMD’s basic bipolar technology into a new line of MOS semiconductors. The prospectus for the IPO also revealed that the balance of power between the founders had also changed. Originally, seven of the eight founders had each received 76,500 shares, and Sanders 102,000 in recognition of his senior status. But in June 1972 the company had granted him options over a further 48,000 shares.

      Four of Sanders’s five remaining founding partners were astounded when they got news of this. The original equity splits had been carefully negotiated; how had Sanders managed to arrange an extra slab of shares for himself? The fifth founder, John Carey, was probably less surprised: the prospectus showed that he had been promoted to vice-president in 1972, and had been awarded over 24,000 shares in recognition of his enhanced value. Carey and Sanders soon found themselves facing furious demands by the other four to rescind the arrangement. But Sanders brazened it out. Without apologizing or explaining, he made it clear that the arrangement had been made, and that was that.

      In the eyes of his partners, the issue of those stock options marked the beginning of what they believed would later prove to be a trend in Sanders’s stewardship of AMD: a grabbing for himself of a share of the company’s success that was considerably greater than what comparable managers elsewhere were being awarded.

      At first sight, there seemed little need for this. At the time AMD went public Sanders was already paying himself $34,000 a year in salary – considerably more than both Noyce and Grove at the larger, older and more profitable Intel. But Sanders had always lived just a little beyond his means. Back in his Fairchild days he had inhabited a house on a good street in Los Altos Hills, one of the smartest places in the Valley. At AMD he had moved further north to the even more prestigious community of Atherton – and after the IPO he would move to an imposing house on a seven-acre estate so large that it required a full-time gardener and a housekeeper to run it.

      In January 1973, when the wage controls imposed by Richard Nixon’s government were removed, Sanders set to work crafting a new contract covering his employment at AMD. The new terms included tight restrictions on the circumstances under which the board would be allowed to fire him, and salary and fringe benefits lavish enough to create a stir when they were published in a trade magazine the following year. Getting this new contract past the board was one of Sanders’s toughest showdowns yet, but once again he succeeded by brazening through. The AMD chairman’s favourite phrase was: ‘A man’s reach should exceed his grasp.’ Never had a life been lived so literally through that maxim.

11

       Turning Failure to Profit

      INTEL’S 1103 CHIP was the world’s first commercially successful semiconductor memory device. But engineers in the computer industry designed it into their products because its high performance and small size forced them to – not because they wanted to. As Intel would admit later on, the first version of the 1103 that Intel produced was a pig to use. A decade after its introduction Gordon Moore made the extraordinary admission that it was ‘the most difficult to use semiconductor ever created by man’.

      It wasn’t simply that the device required a host of peripheral circuits to make it work. It also needed three different power supplies, and it had a capricious timing cycle that any system that used it had to accommodate. Once the feeling of triumph and excitement at simply being able to manufacture the chip at all had subsided, Intel’s top management realized that the company faced a crisis. Unless the flaws of the 1103 were quickly corrected, they might prevent the chip from making the transition from technological curiosity to commercial triumph.

      Intel came up with two responses to the problem. One was to start, immediately, a programme of looking again at the chip design to see how it could be simplified and improved. But it would take time to prepare the new masks, prove a new process, and move from samples to full-scale manufacturing, so the redesign could not be expected to bear fruit for nearly two years. Intel’s second response to the problem was more short term and small scale. Ted Hoff and Stan Mazor, two of the company’s engineers, took a few weeks off other projects to build some demonstration systems. Their hope was that when they saw these systems, customers would be more easily convinced that designing the 1103 into a computer system was less difficult than it looked. But Intel’s customers didn’t behave as expected. Instead of nodding their heads wisely and putting in their orders for the 1103s, the computer company engineers who saw the demo systems began to ask to buy them in mass production quantities, so that they could leave to Intel the job of designing the entire sub-system that accommodated hundreds of individual chips.

      While Intel was pondering whether to manufacture these integrated memory systems, a telegram arrived at Bob Noyce’s house. It was from Bill Jordan and Bill Regitz, two senior staff at Honeywell, the computer company that