(FORTUNE Small Business) – Gordon Moore is the Lou Gehrig of Silicon Valley. In the same way that the Baseball Hall of Famer, sandwiched between Babe Ruth and Joe DiMaggio, is sometimes overlooked in the Yankee pantheon, Moore is overshadowed in Intel's history by his flashier counterparts Robert Noyce and Andy Grove. Yet Moore helped make Intel, the company he co-founded with Noyce in 1968, into a murderer's row for semiconductor chip development. Within three years of starting Intel, he had overseen the invention of the industry's first significant memory chips, as well as the earliest microprocessor, ushering in the computer age. In time, he and Grove established Intel as a pillar of the dominant Wintel duopoly with Microsoft, building what is today a $27 billion company.

Moore was always Mr. Inside, letting Noyce and then Grove act as Intel's public face while he oversaw the technical vision. But he was CEO from 1975 to 1987 and chairman until his retirement a couple of years ago. Moore, 74, is far too modest to take credit for the kind of societal changes he famously predicted in 1965, when he not only observed that the number of transistors that could fit on a piece of silicon would double each year (known as Moore's law) but also anticipated that those pieces of silicon would lead to cellphones and personal computers. He'd rather chalk up Intel's success to serendipity. But as Moore describes the birth of the company he named Integrated Electronics (Intel for short), he reveals the source of its success. --DAVID LIDSKY

"I was an accidental entrepreneur, and that accident happened a startup before Intel. In 1957 a group of us were working at Shockley Semiconductor, the organization that brought silicon to Silicon Valley. William Shockley had won the Nobel Prize for inventing the transistor at Bell Labs. He had phenomenal physical intuition: A colleague of mine said he could see electrons. But he didn't have nearly as good an idea about how people worked. And we had enough problems internally that a group of us wanted to go around Shockley, go to the source of his money, and get him pushed aside and have somebody else brought in. We discovered in the process that a group of young guys had a tough time pushing aside a Nobel laureate.

We thought we had burned our bridges so badly that the group of us would have to look for new jobs. One of the group, known later as the Traitorous Eight, wrote to a friend of his father's, who was an investment banker, saying, "Hey, there's a group of us here that like working together. Do you think there's a company that would like to hire the gang?" The investment banker sent out a senior partner and a young Harvard MBA named Arthur Rock to talk to us, and after an evening's discussion, they said, "What you should really do is start your own company. And we'll find you the money." Gee, we wouldn't have to move--because we'd all bought houses around here--and that's how I became an entrepreneur as a co-founder of Fairchild Semiconductor, which was staked by New York-based Fairchild Camera & Instrument Corp. and later became a division of the company.

We went back to Shockley's original idea--making a silicon transistor. We didn't have a very clear picture of what type of business might come out of it. Our first transistors were used in military computers for driving the memory, and IBM was an early customer. My future Intel partner Bob Noyce, whom I had known since we started together at Shockley, invented the way to make the first practical integrated circuit, successfully putting more than one transistor on the same piece of silicon. This major advance let the business grow to more than $150 million annually.

By the late '60s we had some problems at Fairchild. We went through two CEOs in six months. At that time most of the company reported to Bob Noyce, and Fairchild Semiconductor was generating over 100% of the parent's profits. Rather than offer Bob the CEO job at the parent company, the board decided to go outside. This ticked Bob off, as you might imagine, and he asked me if I'd like to try to set up another company. My reaction was no; I thought I had the best job in the industry as director of R&D. A couple of months later he came back with a much stronger message, saying, "I'm leaving!" And I gulped. I could see things were going to change. I'd rather take action before than after, so I said, "Okay, I'll leave too."

The idea of semiconductor memory was just coming around. I think I had mentioned to Bob once that it was the first idea I'd seen in a long time that I thought you could start a new company on. I was probably trying to justify the work we were doing in the lab. But this was the place where it looked as if there was a new opportunity. We wanted to take different approaches at Intel than the one Fairchild took. What we wanted to do really was make very complex circuits. The industry had gotten to the point where you could build something much more complex than anybody wanted. And things became unique as they became complex. Semiconductor memory was one thing where it looked as if you could build the same product in large volumes even though it was complex. I can't tell you how much of that was just our inclination to push the technology and how much was our real understanding then that this was going to change the leverage in the semiconductor industry. I suspect that it was a bit of each.

It was one of the rare periods when money was easy to raise. We had enough of a reputation from the Fairchild success that we were able to call Arthur Rock, now a venture capitalist, and say, "We're setting up a new company, and we want you to raise $2.5 million for us." And he said, "Okay, give me something I can talk to people about." So Bob typed up a one-page business plan that said nothing except that we were going to work in silicon and do interesting things. Arthur called a few friends and actually raised the money in an afternoon. Bob and I put up half a million and got half the company, and they got the other half for 2 1/2 million.

Bob and I set up Intel to avoid many of the problems we'd had previously. One thing Bob had felt frustrated with at Fairchild was not having anyone to talk to. We set Intel up as a partnership of equals so that we could discuss any of the problems that arose. By that time he kind of liked the outside focus. He liked to talk to the investment community, some of the bigger customers. And I focused on the internal stuff. We hired Andy Grove as our director of operations.

The problem I had increasingly had at Fairchild was the frustration of having stuff we came up with in R&D not find its way into production. To get around that, we never set up a separate research laboratory at Intel. We'd take the inefficiency of doing the development work right in the production line to avoid the inefficiency of transferring technology from one organization to another. If you look around, you'll find very few central research laboratories ever got anything transferred into production.

Intel Incorporated on July 18, 1968, and got started on Aug. 1. That first year we had revenues of about $3,000--interest on the money we had in the bank. It took us a year to get the first product out. It wasn't a barnburner, but if the demand had been much higher, we couldn't have met it. Then we developed the dynamic memory (DRAM) chip, the 1103, in 1970. That was the first chip cheap enough to be used for memory in mainframe computers, and it became a very important product for us.

The Intel startup was amazingly smooth. We beat our budgets. Our initial idea was that we had five years to get to $25 million in revenue; we actually got to $63 million. We went public in 1971, the first year we had a profit. Because things more or less happened on schedule, I was pretty relaxed about it. If you ask Andy Grove, he says that it was the worst time of his entire life. He was afraid we were going to go bankrupt every week. Either I didn't understand what was going on, or he's more paranoid than I am.

After we did some memory chips, we were looking for some other large-volume, complex chips we could build. The electronic calculator was just coming on, but we were a little late to the party. The established calculator companies--Hewlett-Packard, Texas Instruments, and Casio--had all made deals with established semiconductor companies. Here we were--a startup. Busicom was a Japanese startup, and it had 13 chips it had designed to make a family of high-end scientific calculators it wanted us to build. Our small engineering department was completely full building a couple of memory chips. While we wanted to get our feet wet in that market, we didn't see how we could do it.

Ted Hoff, our systems guy, looked at them and said, "You know, I could do all of these with a general-purpose computer architecture, and I don't think it would be much more complex than the memory chips we're building." Gee, I thought, maybe we can do it after all. Our job then became convincing Busicom that it ought to throw away its design work and do it our way! It agreed, and we implemented Ted's idea. That was the original microprocessor.

What got me excited about it was that Ted saw that it was useful for a lot more than a calculator. I specifically remember him saying we could do elevator controls, traffic lights--the industrial market. But because Busicom had paid for the design, it had the rights to the chip. We delivered the first ones in February 1971, and Busicom came around two or three months later, saying it needed a lower price to compete. The only way we can give you a lower price, we said, is if we can have higher volume, and we negotiated the rights for all noncalculator applications in exchange. Then we could go out and really sell the thing. Busicom was very thinly financed. By late summer it was running out of money. So by giving it back its development cost, which I think was $60,000, we got back the complete rights to the microprocessor.

After the fact, we've been criticized for not developing the first personal computer--after all, we had the chips, but we didn't really see that it was going to go anyplace. The engineer who suggested the idea to me--the only application he could think of was having housewives' recipes on it. Even when Steve Jobs came around with the early Apples, it wasn't obvious. You miss a lot of opportunities. The one we were much closer to than the PC was the engineering workstation. For several years we actually did more business on what we called development systems than we did on the microprocessors themselves. These were essentially engineering computers for writing the software, debugging the software and the hardware. We were trying to accelerate the use of microprocessors, and this strategy proved successful by the late '70s, about when we got the design win for what turned out to be the IBM PC.

In a fast-moving industry you get a lot of chances to make mistakes. We've had a few but have generally been able to turn them into serendipitous situations. For example, in 1985 we had not done a very good job on dynamic memory for a while. We had less market share than I liked, and some products were late. The memory industry was losing money by the bucketful: In 1985 the Japanese lost $2 billion, and U.S. companies lost $1 billion. For us to get back to the kind of market share we thought was necessary, something like 20%, we faced making a several-hundred-million-dollar capital investment for which it wasn't obvious we'd get a return. It would have been more difficult to decide to stay in the memory business than it was to get out. The real question was why we didn't do it five years earlier, but it takes a while to know how bad things are.

But that decision turned us into a powerhouse in microprocessors. We had a big memory design group. We refocused it on microprocessors, giving the development team the job of moving from one generation of technology to the next: a fairly extensive redesign of the product we had at the time--must've been the 386. We focused on technology aimed at microprocessors before anybody else, got ahead, and have stayed ahead ever since.

I think we've been able to win because of good people. Noyce and I were the old men: He was 40 and I was 39 when we started. There were no older people in the semiconductor industry in those days. We hired a young group of people with high potential, and we got a broad sampling from the industry. We didn't want to take more people out of Fairchild than was absolutely necessary.

One guy we did take with us was Andy Grove. When I told Andy, who was assistant director of the labs, I was leaving Fairchild, he said, "I want to come too!" I first met Andy back when he was getting his Ph.D. at Berkeley in chemical engineering and was interviewing around. We hit it off. I wrote his thesis advisor for a recommendation, and I got my letter back with the comment on it "This is a truly exceptional individual. Whoever hires him will be very lucky." That was the entire recommendation! He became the leader of the first project I had him work on, even though he was joining it as a fresh member, by his personality and contributions. In setting Intel up, Andy was in charge of getting manufacturing going, and he got interested in how organizations work. I claim he got over his Ph.D. He lost all interest in the technical details and became much more interested in management.

Perhaps even more than for founding Intel, I'm known for an article I wrote in Electronics magazine in 1965. In it I observed that the number of transistors on a semiconductor chip had been doubling every year. In the mid-1970s I revised it to every 24 months. I used to cringe when someone called it Moore's law. I guess I was kind of embarrassed by it. I didn't think it deserved the status of a law. I've gotten used to it by now. Gosh, I read it everyplace I turn.

It changed over time from an observation to something that drives the industry, particularly Intel. With Moore's law, people could predict where things were going to be in the future. In the early days of the semiconductor industry there were a whole bunch of different paths; it seemed as if they were random directions. But when you're making memory at 1K, then 4K, then 16K, and so on, you could plot that and see what you had to do to catch up. In the early 1990s, when I headed up the semiconductor industry association's technology committee, we produced the first semiconductor roadmap. Essentially it said, "This is where it's going," and the participants realized they had to be on that curve or ahead of it, or they were falling behind technologically. Since then it's driven the industry very strongly.

You make a few really good moves. And we've been phenomenally lucky in the way the processor business for the PC has gone. Gee, it's pushing, what, 20 years now?"