(FORTUNE Small Business) – As you walk the halls of Nvidia's Silicon Valley headquarters, you pass rooms named after the stadiums and arenas of some of the great sports franchises of our time: the Candlestick Park room, the Boston Garden room, the Yankee Stadium room. And as you read off the names, you start to feel that this company takes success seriously. These suspicions are confirmed when I enter the appropriately named "press box" and meet the man considered by many to be among the most competitive entrepreneurs in America. Dressed in the de rigueur tech uniform of work shirt and jeans, Jen Hsun Huang at first does not seem particularly obsessed with winning. But behind his wire-rim glasses his eyes dance with animation.

Still, it's only when Huang begins his story that you realize this man is driven. So driven that twice he has, against all odds, rescued his startup from what looked like certain death, and then turned it into the world's hottest chip company. In one instance Huang, whose company makes the semiconductors that run the graphics in PCs and videogame players, watched the entire marketplace shift just before he was about to launch his first chip. ("It was a carnage of massive proportions," recalls Huang. "We just hit the wall with a splat.") Only a year or so later he discovered, just before they were to be shipped to customers, that a huge percentage of one of Nvidia's new chips was flawed. After all that, Huang still managed to pull off Lazarus-like acts of management bravado that put his company in overdrive and made it the Intel of the graphics-chip world.

Talk about gazelles (see page 54). According to Peter Labe, a technology analyst at Buckingham Research, Nvidia's sales will hit $1.1 billion this fiscal year, up 53% from fiscal 2000, and earnings will rise 51%, to $151 million. Not bad for an eight-year-old startup. The stock is on a roll too. Since going public in early 1999, Nvidia has watched its shares rise from an initial offering at $6 to a recent price of $74--a performance all the more impressive at a time when chip stocks have been battered.

Why are investors pushing the buy buttons on their PCs faster than a teenager firing in a Lara Croft game? Labe's answer is that Nvidia controls 48% of the desktop computer-graphics chip market and is moving aggressively into laptops. Major PC makers including Dell, Compaq, Gateway, and Toshiba use them to light up their screens with everything from color photos to banner ads to streaming video. And that's just the beginning. Nvidia is also the sole provider of the superfast graphics chip in Microsoft's new videogame console, the X-Box, the alleged "Sony PlayStation2 killer" that is said to generate near-Hollywood-style graphics.

Jen Hsun Huang learned how to deal with adversity from an early age. In 1973, the 9-year-old Huang was living in Thailand with his Chinese parents. Worried about the political situation there, his parents decided to send him to a U.S. boarding school his uncle had found in a magazine ad. Before long Huang was enrolled in the Oneida Baptist Institute in Oneida, Ky. Not only was Huang the first foreigner ever to attend the school, but he believes he was also the first Chinese ever to walk the streets of Oneida. Moreover, it turned out that the institute was a school for troubled kids. Huang's 17-year-old roommate was fresh out of prison and covered with stab wounds. Huang says that during his two years at the school he got beat up a lot and spent afternoons cleaning the latrine. Nonetheless, he looks on those days as a valuable experience. "Now I don't get scared often. I don't worry about going places I haven't gone before," says Huang, "and I can tolerate a lot of discomfort."

In the mid-1970s, Huang reunited with his parents in Oregon, where he picked up an interest in computers during high school. Huang's competitive spirit also began to show; he became a nationally ranked Ping-Pong champ. After graduation he continued his computer studies at Oregon State and Stanford, where Huang picked up a master's in engineering. Huang got hired as a chip designer, first at powerhouse AMD and then at LSI Logic. But after some years the young engineer realized that working for others didn't fit his style, and he began dreaming about starting a graphics-chip company. "I saw," says Huang, "that this technology could open up a world of color, sound, and movement--that it could bring travel, entertainment, and education to life on a computer screen."

Around Christmas of 1992, he had dinner at Denny's with Curtis Priem, one of the top engineers at Sun Microsystems, who was itching for a new opportunity. Agreeing that graphics could be huge, the two decided then and there to launch Nvidia. Priem would provide the chip design know-how, and Huang would build the company. On Feb. 17, 1993, Huang turned 30 and left LSI Logic to form Nvidia. From the very beginning the company walked and talked like a classic Silicon Valley startup. Huang cashed out all his LSI stock options--everything he owned in the world. With his $150,000 stake he planned to launch his business and at the same time support his wife, Laurie, and their two children. His office was in the dining room of Priem's home in Fremont, Calif., where he and his partner, fueled by coffee and doughnuts, got Nvidia up and running. "We just worked until we dropped dead," recalls Huang. "I remember often falling asleep right in the middle of his dining room."

Huang didn't want to make just any graphics chip; he wanted to make one so powerful that it would revolutionize the market. In July of 1993 he persuaded two of Silicon Valley's most prominent venture capital firms, Sutter Hill and Sequoia, to put up $2.5 million in seed money. Huang was on his way. The company's first big break came later that year when Sega licensed Nvidia's chip for its videogame console. What intrigued both the VCs and Sega was Huang's strategy. Back then, graphics chips drank up a lot of expensive computer memory. So Huang designed his first chip, the NV1, to be a memory miser. The advantage: Computer makers could save money on memory and still generate great graphics.

After two years of development, just when Nvidia was rolling out its new chip, the unexpected happened. The cost of PC memory fell through the floor, erasing in one fell swoop Huang's cost advantage over rival chipmakers. Worse still, Nvidia's chip ran on a proprietary technology. Why would customers like Sega and Dell buy Nvidia's chips when, for the same money, they could buy rival chips that operated on an industry standard and performed just as well? By early 1996, Sega and other customers disappeared, and Nvidia was left with hundreds of thousands of chips it couldn't sell.

Facing such a disaster, many entrepreneurs would have given up. Instead, Huang regrouped. He laid off 70 of 110 people, including all the sales staff; scrapped the NV1 chip; and went back to the VCs for more funding. They supplied it. Says Sutter Hill's Jim Gaither, one of the VCs who stuck by Huang: "Original plans seldom make it all the way. We were betting on this great team of engineers." Huang had learned a hard lesson. "We thought that if you built the best technology, they would come. That's not true. We had to become better versed in reading markets and reading consumer demands." The key, Huang figured, was to stick to the industry standard (which was what customers wanted) and produce a chip better than everyone else's. But that wouldn't be easy. Says Chris Malachowsky, VP of hardware engineering and one of Nvidia's founders: "Making the switch was a tough strategic decision to make culturally, because the company had prided itself on being different, on having a unique technology. Now it was being forced to play catch-up."

Many of Huang's talented engineers rankled at the idea of becoming a "me too" chipmaker and threatened to leave the company. Throughout those tough weeks, Huang and his leadership team kept explaining to employees the rationale for the strategy shift. He gave speeches stressing that they all must look at themselves not as a product but as a company. Says Malachowsky: "We kept hitting upon the idea that we were all betting on the talent of everyone in the company, not on a particular technology."

Their campaign worked, key employees stayed on, and the result a year later was an entirely new graphics chip, the Riva 128, which complied with industry standards but was about 400% faster than anything else on the market. PC manufacturers like Dell and Gateway began to line up for the new chip, and by 1997, Huang had hundreds of thousands of them in production.

Once again, it was make-or-break time for Nvidia. It was down to its last $2 million or $3 million in the bank, and it had $28 million tied up in the production of the Riva 128. The company had to ship the chips to customers on time or else close its doors.

Trouble was, the Taiwanese supplier making the chips for Nvidia produced an extraordinarily high number of flawed units (about 30% of the chips were bad, compared with an industry standard of about 5%), and no one could figure out the flaw in the design. At that poor level of quality, there was no way Huang could ship these chips to customers like Dell.

Once again, rather than concede defeat, Huang pressed on. In a tense meeting, Malachowsky suggested, "Why not test every chip by hand?" Huang's first thought was, "That by itself would kill us off." But then he jumped on the idea. What did he have to lose? Starting immediately, he, Chris, and every one of Nvidia's employees would work day and night in headquarters to literally test each and every one of the hundreds of thousands of chips before they were shipped to customers. Each chip had to be tediously inserted in a PC, tested, removed, and shipped. It took about five minutes to test each one--thousands of hours of man power. It was a daunting task, but as Huang puts it, "We figured if we could do it, if we could test every chip, we could save the company."

They did, and the Riva 128 became a big hit. Says Huang: "There were a lot of things we learned about ourselves--that we were willing to go to almost any length to make sure we were successful." The chip-testing story is a legend at Nvidia. To this day employees tell of how everyone from the CEO to the mailroom guys chipped in to save the company. Says Huang: "All the life threatening, all the challenging, all the extraordinary endeavors become lore and help you galvanize the company." By encouraging his employees to repeat this tale of heroism, Huang is really trying to get them to celebrate the character of the people inside the company. "A lot of companies think they have a fundamental technological advantage," says Huang, "and that's just bunk. The company that sustains itself must have something else; it's the people. It must be cultural. It must be some basic way that we do things."

Huang also believes he wouldn't have survived if he hadn't created a culture where intellectual honesty was key, where people were willing to identify and correct mistakes well before they became fatal. As Malachowsky explains, "Most accidents are a result of a chain of events. You've got to cut the chain off each time and stop the hurt, stop the mistake. The key is to be willing to evaluate where you are and make sure the chain doesn't end up in a crash." Or as Huang puts it more succinctly, "The recovery only happens once you say, 'I'm bleeding.'"

Today Nvidia is running full out to complete its new graphics chip for Microsoft's X-Box, which the company says is the fastest ever built for a videogame console. Is the X-Box potentially another life-threatening situation for Nvidia? Says Huang: "I don't think so. It's an extraordinary opportunity for us. It's a big bet; it's a large chip. We're under a very tight deadline and under a powerful microscope. We'll pull it off."

Besides, Huang sees his new generation of graphics chips as doing much more than driving videogames. He plans to leverage the architecture of his new graphics chip to build a family of products for PCs and Internet appliances. In nongeek English that means that Nvidia plans to provide the graphics horsepower for everything from handheld computers to cell phones, and perhaps even televisions and home theaters.

But should disaster strike again, the battle-scarred Huang is unlikely to be at a loss for what to do.