There's power in a glorious failure.

Failure is just part of the culture of innovation. Accept it and become stronger. The infamous Pentium flaw in 1994 was devastating, and we went through all the stages of grief -- denial, anger, acceptance. It was incredibly painful to the company and to me, personally. But we managed to become better as a result. It marked a real transition. I'm a different person today. I've beefed up the way we validate our technology before it gets out the door. We went from having a product-engineered orientation to a consumer orientation. We broke down barriers inside the company because we were all involved in an emergency.

Everyone was working on the problem -- engineering, marketing, administration -- we all became customer-service reps, answering phone calls and solving consumers' problems about replacement. We all recognized that the problem threatened the image of Intel. We had real teamwork and came through the crisis together. Now we know that we can respond to any crisis 10 times faster than before. That's a sign of entrepreneurship.

Think in epic images, but be ready to sweat the small stuff.

I have a vision. I see the world with 1 billion personal computers operating. Not only that -- I also see them all connected. My job is to define Intel's role in this vision. We know we will have to be positioned to expand our volume for new applications, so my vision has turned to manufacturing. Now I'm thinking in real time about our capabilities and what we need to do to expand them. Much of my work is in real-time problem solving.

For instance, about eight years ago, when we were looking at a new architecture for the P6 processor (later named the Pentium II Processor), we knew that we had a great high-end-market product. We didn't know how to apply it to a lower, cheaper desktop -- until someone came up with the idea of making our cache memory more flexible. I was against removing cache from the chip, and after a great argument, we decided to separate the processor and the cache chips. In one sense, this meant downgrading the chip's performance by lessening the cache for desktops. Some of our people argued that we didn't need extremely powerful performance for desktops, and that we could manufacture the desktops faster by installing the smaller cache. For me, that was an aha! moment. We solved a big problem and were able to satisfy both markets: We used a smaller cache for the desktop market and a much larger cache for the server market, where high performance is critical.

Teamwork is innovation.

In most cases, a team leader should have some traditional experience. At the same time, the trick is to appoint leaders to tasks that go beyond their previous experiences. When the leader strives, the team will grow.

When you put a team together, think of it as a solutions-oriented process. Gather people with expertise in architecture, engineering, manufacturing. Also think about the team as a combination of personalities: a mix of wildness and conservativeness, overlayed with people who have a strong sense of reality. Don't just stand back and see what happens -- this is an actual team, not a virtual one. Give it an exercise to shake itself out. Make it a real project -- not a difficult one, but one to practice on.

A small initial success can go a long way toward building team confidence. You can also use the exercise to discover who doesn't belong on the team. When these organizational issues are resolved, the team is ready to take on a bigger task. This is very tangible stuff, but it all has to do with encouraging innovation. It works the same way, no matter how big or how small the company, whether the company is a startup or an established business. In the end, innovation is about getting to the marketplace.

Innovation accelerates with time -- or else.

In the early days, when we finished developing the 286 chip, our engineers looked up, saw there was nothing else to look forward to, and they left. We didn't do a good job of slotting them onto the next project. That was a big mistake. So next time, when we were finishing the 386 chip, we got many engineers started on the 486, thus accelerating the next generation. This matter of timing is critical. Most people aren't looking to finish a project and go sit on a beach. They're looking for new challenges.

In this sense, innovation can never stop. Sometimes it's important for people to move on to another project even before they complete the one they're working on. The important point is that people know that they work in an atmosphere that accepts and encourages change from within -- at the same time that it encourages adaptation to changing markets.

What's Fast

Where do good ideas come from? "It's a contradiction to say your brain can schedule innovations,'' Albert Yu says. "Sometimes we have no clue. Innovators exercise their subconscious. Sometimes late at night, my subconscious takes over and says, 'We've got to make this happen.' And then it strikes me. I jump up and down, and I holler: 'I've got it! I've got it!' Then I rest up and get ready to go to work.''

Pat Dillon (pdwolf@aol.com) has written several feature stories for Fast Company. His latest book, "Lost at Sea: An American Tragedy," was recently published by the Dial Press.