|
Continued
Previous | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
For each new generation of chips, Intel has had to develop the parallel technologies of the chip itself and the means to produce it commercially. That's where all the research and development money goes.
The power of the microprocessor increased exponentially over its first two decades because engineers found ways to make the main components, the transistors, smaller and smaller and smaller. Transistors were no longer bits of metal and plastic, as they had been when most people first saw them inside a pocket radio. They were specks of chemicals, and it would take hundreds of them to make a ring around a human hair. Arranging millions of them in effective circuitry telescoped the capacity of a much larger machine onto a chip the size of a postage stamp. But even as specks, transistors had to be precisely drawn.
On an old-fashioned transistor, an impurity the size of, say, a cookie crumb, could interfere with performance. Next to a transistor measuring less than one micron, a germ - a single bacterium - looks like a boulder and renders the whole chip worthless. Intel had to design production rooms in which all the air was filtered every few seconds, leaving less than one such particle per cubic foot. Humans, those roving dust storms of dandruff, viruses, spit, and lint, had to be sealed inside special suits in order to work in "clean rooms."
Inside the clean rooms, the photographic process of imprinting arrays of transistors was extremely sensitive as the detail shrank. Site selection for production facilities became a matter of geology: the slightest tremor, imperceptible to humans, would distort the circuitry being exposed. Most places on earth shake
almost continually at extremely low levels. A crucial factor in Intel's expansion to places such as Ireland and Israel is that the ground itself is stable in those places.
According to Gordon Moore, Intel has to produce a new generation of processor technology every three years in order to maintain its advantage in chip engineering. "We just continue to push narrower and narrower line widths and more and more complex processes, so we can increase the density of electronics on the wafer," he said in 1993. "Get more and more stuff on a chip, essentially." a bug. At first Andy Grove dismissed it as a remote, statistical problem:
"If you know where a meteor will land, you can go there and get hit," he responded, when asked if even a rare problem was still a
problem. Eventually, Intel offered to replace Pentium Processors at
no charge. Grove later admitted that he had a lot to learn about
dealing with the public.
Even as Intel began to ship the $995 Pentium, in March 1993, the
pace at Intel continued to quicken. In December 1993 Intel said it
would double the capacity of its chips in the next year, slashing the product development interval of eighteen to twenty-four months
down to one year. "The operative word is focus," Grove said in 1993. "You have to put all your effort behind the thing that you do better than the other people in the business, and then not hedge your bets. . . . If you focus and you're wrong, you lose-but if you're right, you win big time."
In the first week of April 1994, Intel announced that it planned to
spend $150 million to market the Pentium-a stunning amount of
money for a piece of silicon. Pentium, though dominant, was competing with a growing array of products made by companies like Advanced Micro Devices, Cyrix, and IBM, along with the new Power PC chip made by Apple and Motorola. The purpose of Intel's latest marketing campaign, with the tag line "Intel Inside," was to make the chip's brand name a household word, as familiar to the American consumer as McDonald's or Coca-Cola. As an inducement, the company offered to pick up half the tab for advertisements its computer-producing partners ran that used Intel logos in the ads.
In 1995, when Intel had $16.2 billion in sales and $4.9 billion in
profits, Noyce and Moore's bold prediction of a quarter century earlier seemed, if anything, a major underestimation. Gordon Moore's net worth, measured largely in Intel stock, was pegged at over $2 billion by Forbes.
Under Grove's leadership, Intel was still paranoid in a healthy way. After all, staying at the vanguard of the computer revolution requires a sort of perpetual motion. Like a jogger running on an accelerating treadmill, Intel has had to run faster just to maintain its position-and even faster just to stay ahead of everybody else.
Continued
Previous | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
|