a chance of winning the contract, Westinghouse had decided to redesign the radar system around the equivalent Fairchild part. Having been bitten once already, it wanted a second source in case Fairchild in turn failed to deliver.
Simonsen realized that AMD had just been offered a great opportunity. ‘Don’t do that,’ he replied. ‘You don’t need to redesign your system. Give us the specifications [for the TI part], and we’ll build you an adder that works with your existing system design. The component price will be high, but it will cost you less than the redesign.’
The Westinghouse manager was sceptical. Surely, he said, it would take too long for AMD to develop a second-source TI part from scratch. After all, there were only three months to go before the radar system test.
‘We’ll do it in six weeks,’ said Simonsen.
Six weeks later, to the day, Simonsen delivered the first production sample to Westinghouse. It was a smart move. Not only was AMD guaranteed far higher margins than it would have received as a second source to an existing Fairchild part. More importantly, the Westinghouse guy was now convinced that everyone at AMD walked on water. The goodwill that the company had earned would be worth many hundreds of thousands of dollars in future contracts.
* * *
The ideal kind of second source, people used to say in their cups at the Wagon Wheel, was a company that was just good enough to make customers believe it could build the part, but not quite good enough actually to build it. That was the kind of second source that Intel was lucky enough to find.
The product that proved the point was the 1103, the world’s first 1K DRAM. Intel had advertised it in the trade rags under the screaming headline: cores lose price war to new chip. The ads went on to invite potential customers to call Intel collect in Mountain View, tell its sales department what they were currently paying per bit for core memory, and hear from Intel how they could switch to smaller, faster semiconductor technology without spending a penny more.
So powerfully did this pitch grab the industry that Intel was soon inundated with calls from computer companies, none of whom wanted to be caught out by being last to switch to semiconductor memory. Five minutes into the conversation, however, would always come the dread question: who was Intel’s second source?
Andy Grove was temperamentally opposed to the principle of second-sourcing. As anyone who had seen the charts on his wall could tell, he saw it as one of the greatest failures of the electronics industry that its members never seemed to be able to deliver on their commitments. The solution to the problem, he believed, was not to sacrifice revenues by allowing a know-nothing competitor to rip off your designs. Instead, he maintained, Intel should get its manufacturing working with sufficient reliability so that customers did not need to ask for a second source.
There was also a concrete issue at stake. The industry’s graveyard was littered with the bones of companies that had come up with a couple of good ideas but then failed to mass-produce them at the right time and the right price. The great risk of signing a second-source deal was that if the chosen partner was one of the industry’s leading names, Intel might hand over to a well-funded and well-organized competitor the very secrets that could be used to drive it out of business.
But Grove was before his time. Early in 1970 Intel had received a request for second-source rights from a Canadian telecommunications company called Microsystems International Limited. After extensive discussions, Intel’s board of directors overrode Grove’s objections and authorized Noyce and Moore to go ahead and negotiate a deal.
The MIL deal was fortuitously timed. It began to produce a stream of fees and royalties – starting with an up-front cheque for nearly $1m – that kept the company afloat while Intel was suffering from the recession at the end of 1970. But the pain came in the winter, when Intel was forced to send teams of engineers out to Ottawa, braving the worst snow and ice that a Canadian winter could throw at them, to hand over to MIL the mask sets for the 1103 and to make sure that its fab lines were running properly. The job proved unexpectedly difficult, but Intel was contractually obliged to do it, and the Canadian company had astutely inserted a provision in the agreement which meant that Intel would receive bonuses if it helped MIL to achieve certain yield levels by a specified date. The painful process of managing this, and the wasted time of repeated engineering visits, left Grove with a bad taste in the mouth. It was ‘not as emotionally satisfying as getting a product to manufacturing,’ he recalled, ‘but the degree of work was really quite the same’. But he worked loyally to implement the decision that he had opposed, and did his best to make it a success.
Three months later, as MIL began to offer its own version of the 1103 in competition with Intel, it began to look as though Grove had been right. Every time the subject came up for discussion, his face gave away his thoughts. You should have listened to me. Was it really worth the front money? Look, we’re losing money left and right here.
But then MIL got too clever. In an attempt to get more chips out of each wafer on the fab line, the company made a change to its production system without consulting Intel: it tried to shrink the chip, and to increase the wafer diameter from two inches to three. The change was a disaster: yields dropped immediately to zero. To make matters worse, the Canadian company built up many months of useless inventory before the problem was properly detected, let alone solved. The result was a stream of former MIL customers, suddenly strapped for memory circuits, coming back to Intel begging for orders of 10,000 units for delivery as soon as possible. Intel profited handsomely, and MIL’s failure helped to reinforce the company’s reputation as an organization that could deliver on its commitments. Even Andy Grove was forced to admit that this first experience of second sourcing had proved to be a success.
AMD’s public relations were handled by Elliot Sopkin, a journalist who had come to California from Atlanta to work on a local paper in Oxnard, a beach town just north of LA. Sopkin had lived next door to Ed Turney, and had been impressed by the salesman’s raffish lifestyle. Every day, as Sopkin went out to work, he saw Turney stroll out on to the sand in a swimsuit, carrying a telephone with an enormously long extension cable. He would then sit down in a deck-chair, telephone machine at one elbow and adding-machine at the other, and start making loud sales calls to clients. When Sopkin returned from work at midnight, Turney would be back on the beach, full of drunken talk about the topless bar he had just visited in Santa Monica and the $100,000 sale he had just made to the purchasing manager he took there. It certainly seemed more fun than journalism. When Sanders appeared in Oxnard and offered Sopkin a position at AMD, he accepted without hesitation.
Six months into the job, Sopkin presented Sanders with a cartoon from the New Yorker depicting a king, obviously King Arthur, sitting on a throne at a large table surrounded by knights. ‘The round table symbolizes our equality,’ said the caption. ‘The high-backed chair and the funny crown symbolize that some of us are more equal than others.’
The cartoon captured Sanders’s style perfectly; his colleagues soon began to refer to him as ‘the king’. While Intel was non-hierarchical and resolutely insistent on the primacy of ‘knowledge power’ above ‘position power’, Sanders kept aloof. Even before the financing was in place, he made it clear to his seven colleagues that he would prefer to deal with investors himself – and followed this up by excluding the other founders from the company’s board of directors. Once the company was formed, he awarded himself the titles of president, chairman and chief executive officer. Nobody complained: they all had plenty to do themselves, and Sanders evidently had the capacity to lead.
He certainly looked the part. Sanders drove a Mercedes-Benz, and wore a huge range of clothes, from dazzling white suits to cowboy hats, that were evidently made to measure. Some of his colleagues thought he was the height of style; ‘always six months ahead of GQ,’ said one. Others thought Sanders’s clothes simply revealed that he was a poor boy made good. But everyone
