How Intel Puts Innovation Inside

As the TeraHertz project gathered momentum, Intel executives did something seemingly illogical but extremely shrewd. They took some of their freshest hires -- generally newly minted PhDs -- and assigned them to some of the toughest projects. "They have this great advantage of not knowing what's impossible or what's too hard to be worth trying," El-Mansy explains. "So they try it anyway. And sometimes it works."

Sure enough, a shy bravado characterizes the newest arrivals at Hillsboro. This past September, Zhila Alizadeh-Cope joined Intel after getting her PhD in electrical engineering from the University of Central Florida. She speaks in a near whisper and seldom makes eye contact. But when she talks about her specialty -- chip-making material known as silicon-carbide thin films -- it's easy to see why Intel wanted to hire her. "Our group published six or seven papers when I was in graduate school," she says. "It's a hot field. And I don't know of that many people at Intel who have worked in this area."

Another Intel decision involved putting its Hillsboro researchers next to a full chip-development factory and encouraging them to test new ideas quickly in clean-room production. "I don't want people to argue about equations. I want them to build prototypes and find out what really works," says Marcyk. "We call that a 'gross reality check.' A prototype probably won't work perfectly the first time. But you'll be able to tell if you're getting at least a squiggle. If you are, maybe a big window of opportunity will be there. If you're getting nothing, maybe it's time to move on."

Granted, linking research and production costs more than leaving researchers to their blackboards and buying them more sticks of chalk every week or two. But when the time comes to begin serious manufacturing, "we can ramp up to full production volumes almost overnight," says El-Mansy. "There's no one else in the industry that can do that."

Finally, Intel's top executives resisted the temptation to micromanage the TeraHertz project. At quarterly review meetings, they made sure that researchers were tackling the right problems. And when they heard about serious bottlenecks, they put extra people with the right skills to work on those particular problems. Otherwise, Intel's top brass left day-to-day decision making in the hands of the engineers and their immediate managers.

In one clever twist, Marcyk gave engineers wide latitude to design almost any clean-room experiment that they wanted -- but rationed the number of silicon wafers that they could use in a month. Wafers amount to several hundred dollars apiece, but cost savings was only a small part of his justification. Mostly, he wanted engineers to design the smartest experiments that they could. Making raw silicon a scarce resource helped ensure that.

Looking ahead, Marcyk says, it's likely that Intel engineers will tinker further with the TeraHertz design. Some features may go into mass production as early as 2005; others could take longer. "Moving away from silicon dioxide is a very risky step for us," he says. "But if we don't do it, progress on Moore's Law is going to stop. The PC industry is going to grind to a halt. If we're going to make more progress, this decade will have to be the decade of new materials. It's my job to make sure that Gelsinger's meltdown scenario doesn't happen."

George Anders is Fast Company's Silicon Valley bureau chief.

Update: Anders no longer works with Fast Company.