(FORTUNE Magazine) – Mr. Bezos, can you hold that bottle of water on your head so I can make sure I'm in focus?" "Mr. Bezos, when you jump, can you spread your arms and legs in the air?" "Mr. Bezos, while you're in the air can you turn your body 90 degrees and land in a sitting position?"

Jeff Bezos is having his picture taken jumping on a giant trampoline, and, remarkably, he looks like a man in his element. Most CEOs wouldn't pose for a photo like this in a million years. Risk life and limb to look undignified? Not a chance. But Bezos, who is 39, embraces the experience with the enthusiasm of a 10-year-old in an arcade. He cracks jokes about his receding hairline, he chats up the photographer about his equipment, he even helps the photo assistants rearrange the shot. "We ought to try this on the vomit comet," he says mid-jump, referring to the plane astronauts use to simulate weightlessness. "That would be really cool." He protests only once, after 30 minutes of jumping, when the water bottle he keeps placing on his head for the photographer springs a leak. "Next time we do this I think I'm going to need a stunt double," he says. When it's all over, he pulls out his own camera and orders everyone onto the trampoline for a group shot.

Anyone who knows Bezos understands that it doesn't take much to get the founder and CEO of Amazon.com to start acting up. In good times and bad, Bezos has always been a man of exaggerated gestures, whether he is dressing up like a four-star chef to promote Amazon's kitchen store, climbing atop a conference table on all fours to signal his interest in a business presentation, or just unleashing his famous braying honk of a laugh. During the Internet bubble his oversized personality made him seem fun and inspiring, and he shrewdly used that to make himself and his company one of the most talked about business stories in a generation. When Amazon's stock price fell and its losses continued to mount, he endured whispers that his behavior made him look clueless.

Today only three questions about Bezos's behavior are relevant: Is he as goofy as Bill Gates? Is he as goofy as Michael Dell? And is he as goofy as the late Sam Walton? Such comparisons would have been laughable when conventional wisdom had Amazon joining the dot-com trash heap. Not anymore. Amazon has started to thrive. While most of American business is still sputtering, Amazon's revenues, at $4 billion, are growing by more than 20% a year. Marketing, inventory, and warehouse operating costs, once so high they made old-fashioned retailers look efficient, are now so low that only Dell's and very few others' are better. Amazon's operating profit margin, at 5% in the all-important fourth quarter, beat that of most retailers, and approached Wal-Mart's 6%. And Amazon is generating so much cash--$135 million last year, rising to an estimated $300 million this year--that it just paid off 12% of its $2.3 billion debt. At a recent $30, Amazon's share price is at a two-and-a-half-year high, making it one of the top stocks over the last five years, even taking into account its rise and fall during the bubble. It has outperformed Dell, Cisco, Microsoft, Wal-Mart, and GE, to name a few.

And while Amazon still hasn't turned a profit yet, it's headed toward profitability so fast that investors have stopped worrying about whether it will ever make money. Now they simply debate how much. Profitability has taken awhile because the company borrowed heavily during the bubble to finance its growth, and interest-related expenses still suck away a lot of what flows to the bottom line. Its relatively new electronics, tools, and kitchen business is losing money too, though at a rapidly shrinking rate. But with costs falling and revenues and operating profit rising each quarter, it's expected to make roughly $200 million this year. Further out, Legg Mason's Lisa Rapuano--who along with Bill Miller has been Amazon's largest outside shareholder for three years--now believes that net income will approach $800 million on $8 billion in revenues by 2007. If she's right, that would make Amazon one of the most profitable corporations in the country.

Even Warren Buffett, who's never been a big backer of technology businesses, is now an Amazon and Bezos fan. He doesn't own the stock, saying he can't yet figure out how to value it. But Buffett owns $459 million worth of Amazon's bonds, making him one of Amazon's biggest debt holders. "I don't know if Amazon is going to weigh 150 pounds or 300 pounds, but the one thing I do know is that they are not anorexic," he says. "I've been using a computer for eight or ten years now and I still really pay for only three things on the Internet: the Wall Street Journal, online bridge, and books from Amazon.com. That they are one of only three companies online that have gotten money out of my pocket tells me they are doing something right."

Put simply, Bezos is steadily proving himself to be one of the great CEOs of his generation. Sure, guys like Jerry Yang and David Filo, who started Yahoo and still work there, and Pierre Omidyar, who started eBay and then split, are rich. But when the history of those companies is written, much of the glory will go to the professional managers who came in to make the firms thrive. Bezos has not only demonstrated the vision to get his venture off the ground but he's proved that he can lead Amazon through good times and bad, and that he has the management chops to oversee a company that's grown from a handful of people to nearly 8,000.

Bezos's fun-loving exterior has made it hard to imagine him as a successful corporate executive, even to some of his biggest backers. But anyone who has worked for Bezos isn't the least bit surprised that he is coming out on top. While Bezos the celebrity has sold himself as a puppy dog, Bezos the boss has never been the least bit cuddly. The laugh is genuine. So is his love of space and Star Trek; his dog Kamala is named after an obscure Star Trek character. But working at Amazon is like seeing Disney World from the underground tunnels: The head keeps smiling but the work is grueling, serious, and demanding. Bezos's weekly management meetings, for example, are four-hour marathons that resemble oral doctorate exams. Executives make presentations about new products, technologies, pricing strategies, or cost-control measures, and Bezos, with the skill of a first-class prosecutor, asks questions until he is satisfied he has explored every angle. People don't leave Amazon just disillusioned or frustrated the way they leave most jobs. They leave exhausted. "When people characterize Jeff as goofy, I laugh," says David Risher, who left Amazon a year ago to go into teaching after working as one of Bezos's top executives for six years.

Call it the Amazon way: Project an image of fun, but inside, hire smart, drive fast, and above all, bet on the numbers. How's customer service doing? Bezos isn't interested in a qualitative answer. He wants to know average customer contacts per order, average time per contact, the breakdown of e-mail vs. telephone contacts, and the total cost to the company of each. Jeff Wilke, who runs customer service and Amazon's warehouse and distribution operations, says he looks at about 300 charts a week for his division alone. The boss makes no apologies for his love of data. "With most decisions, you can do the math and figure out the right answer, and math-based decisions always trump opinion and judgment," he says. "The trouble with most corporations is that they make judgment-based decisions when data-based decisions could be made." Says public relations director Bill Curry: "I've seen Jeff end discussions by saying, 'We don't need to debate this, because this is data we can get.'"

Once the math is in, Bezos wants things done his way. Every press release that quotes him must cross his desk. If Bezos can't get an answer from one of his senior executives, he'll micromanage his way four levels into their hierarchy. "I've not seen an effective manager or leader who can't spend some fraction of time down in the trenches," he says. "If they don't do that, they get out of touch with reality, and their whole thought and management process becomes abstract and disconnected." This isn't the place for people who value their autonomy. If Bezos demands something get done fast--and employees say he always wants projects done in half the time it seems reasonable to complete them--he'll simply commandeer the people he needs. Says a former employee: "We had a 110-page contract with Toys "R" Us in 2001, and he had a point of view on everything--the economics, the service-level agreement, on price parity, even on Amazon's right to get hot toys because it was giving up its own toy business to do the deal."

There is a cost associated with this management style: Bezos has trouble keeping top people. Amazon's total turnover is about average for e-commerce companies, at 15% a year, but the senior ranks seem to turn over faster than that. Amazon says the turnover comes with torrid growth. Even so, an Amazonian estimates that as many as 20 of the company's top 50 executives have left in the past two years. Last year, three of Bezos's direct reports announced plans to leave in the space of nine months: Risher split for the University of Washington Business School; CFO Warren Jenson took the CFO slot at videogame designer Electronic Arts; and Mark Britto, the senior vice president in charge of business development, became CEO of online help desk Keen.com. "At the end of the day there is a sole-proprietorship aspect to the business," says a former Amazonian. "Everyone knows it's Jeff's company, and that if you work there you are hired help. You can be paid well and be fulfilled professionally for a while, but you know that no matter how long you stay, you're not going to become CEO."

The turnover doesn't seem to bother Bezos all that much. He's a relentless recruiter. When Jenson left, for example, Bezos brought in GE lighting CFO Thomas Szkutak. Even though Bezos is a hard-driving boss, he hires like-minded intelligent people and then plays to their desire to learn. Last month he and chief scientist Andreas Weigend invited MIT physicist Neil Gershenfeld to lecture to 400 Amazonians about the falling price of manufacturing. Gershenfeld daydreamed about how someday Amazon might offer consumers the ability not only to buy anything on its site but also to custom-build and design any product they choose. Monthly lectures like this do little to help move the latest hot book or cellphone. But Bezos uses them to keep employees thinking big. "People here like to invent, and as a result other people who like to invent are attracted here," he says. "And people who don't like to invent are uncomfortable here. So it's self-reinforcing."

Bezos's obsession with innovation and disregard for hierarchy isn't just CEO pablum. One of Amazon's most prestigious in-house awards is called a Just Do It--the winners are employees who do something they think will help Amazon without getting their boss's permission. It has to be well thought through, but it doesn't have to succeed. There are risks to rewarding such behavior, Bezos agrees, but "the cure--to encourage people to always ask for permission--is worse than the disease."

In the late 1990s, when competition for good employees was intense, Amazon was notorious for scrutinizing customer service reps' college transcripts and SAT scores before offering them jobs. Some see the practice as a waste of time and a bit elitist. Who really cares if a customer service rep has read Foucault's Pendulum?

But Bezos's view from the beginning was that he was creating something new and revolutionary. The company's key to success, therefore, would be its ability to innovate. And, says Bezos, the smartest people, regardless of their job, are always the best innovators. Bezos is famous for asking candidates brain teasers like "How many windows are in the city of San Francisco?" or "How many trees are in New York's Central Park?" At most companies, references are afterthoughts. At Amazon they sometimes matter more than in-person interviews. And after eight years, the company still uses the same list of 23 questions Bezos created to do reference checks when he started the company. Two examples: "What situation would you not put this person in?" or "Is this person one of the best people you've worked with?"

The Amazon way wasn't born at Amazon. Since Bezos was young, he's been obsessed with brainpower and numbers. At 12 years old he was the subject of a book documenting junior high school life called Turning On Bright Minds. After stints living in Albuquerque, Houston, and Miami, Bezos went to Princeton, where he graduated summa cum laude in electrical engineering and computer science. He turned down offers from Bell Labs and Intel, taking a position at a startup run by two Columbia University professors that was building a cutting-edge telecommunications network for Wall Street firms. He eventually left the company but stayed on Wall Street, first selling software to pension fund clients for Bankers Trust, then spending five years working for David Shaw's D.E. Shaw hedge fund. Shaw, a computer scientist by training, was a pioneer in using computers to hunt for mispriced securities. By the time Bezos was 28 he had become the firm's youngest senior vice president ever.

The story of how he started Amazon is now legendary. While working at Shaw in 1994 he read a study that predicted the Internet would explode in popularity. He figured it wouldn't be long before people would be making money selling over the web. After researching a host of items that could sell online, he settled on books. Almost every book was already catalogued electronically, yet no physical bookstore could carry them all. He based Amazon in Seattle because it had a big pool of software engineers and was less than 400 miles from Roseburg, Ore., home to the largest book distribution warehouse in the country.

The beauty of the model, Bezos thought, was that it would give customers access to a giant selection yet he wouldn't have to go through the time, expense, and hassle of opening stores and warehouses and dealing with inventory. It didn't work out that way. Bezos quickly discovered that the only way to make sure customers get a good experience and that Amazon gets inventory at good prices was to operate his own warehouses so he could control the transaction from start to finish.

Building warehouses was a gutsy decision. At about $50 million apiece, they were expensive to set up and even more expensive to operate. In order to stay in business Amazon had to issue more than $2 billion in bonds. Suddenly it looked as if Bezos hadn't built a web company at all. To critics on Wall Street, Amazon appeared no different from L.L. Bean or Eddie Bauer; it just had a nicer website.

Visit one of Amazon's six warehouses today, and it becomes quite clear why Bezos believed he would prove Wall Street wrong. They are models of GE-like efficiency. The Fernley, Nev., site sits about 35 miles east of Reno and hundreds of miles from just about anything else. It doesn't look like much at first. Just three million books, CDs, toys, and housewares in a building a quarter-mile long by 200 yards wide. But here's where the Bezos commitment to numbers and technology pays off: The place is completely computerized. Amazon's warehouses are so high tech that they require as many lines of code to run as Amazon's website does. Computers start the process by sending signals to workers' wireless receivers, telling them what items to pick off the shelves; then they crunch everything, from which item gets plucked first to whether the weight is right for sending.

Along the way the computers generate reams of data on everything from misboxed items to chute backup times--and managers are expected to study the information as if it were the Talmud. Bezos typically visits one warehouse for a week in the fourth quarter. For the workers, it's like their wacky uncle dropping in. For executives like Tim Collins, who manages the Fernley warehouse, it's more like a visit from the IRS. Bezos arrives firing a barrage of inquiries about picking algorithms, line speed, and productivity, and doesn't stop asking until he gets answers that satisfy him.

In response, the managers sweat every last drop of productivity out of the warehouses. For example, by redesigning a bottleneck where workers transfer orders arriving in green plastic bins to a conveyor belt that automatically drops them into the appropriate chutes, Amazon has been able to increase the capacity of the Fernley warehouse by 40%. Today, Amazon's warehouses can handle three times the volume they could in 1999, and in the past three years the cost of operating them has fallen from nearly 20% of Amazon's revenues to less than 10% percent. The company doesn't believe it will even have to think about building a new warehouse for another year.

The warehouses are so efficient that Amazon turns over its inventory 20 times a year. Virtually every other retailer is under 15. Indeed, one of the fastest-growing and most profitable parts of Amazon's business today is its use of its warehouses, and sometimes its entire back end, to run the e-commerce business of other retailers, such as Toys "R" Us and Target.

All of this helps explain Bezos's larger point, one he's been making since he started Amazon but that people are only now starting to believe. "In the physical world it's the old saw: location, location, location," he says. "The three most important things for us are technology, technology, technology." Amazon spent big on software development, but now its platform requires little additional investment. Thanks largely to its conversions to the free Linux operating system, technology and content expenses are down 20% in the past two years. "There just aren't other companies that let a consumer order two out of what are millions of products in a warehouse and then quickly and efficiently, at low cost, get those two things into a single box," Bezos says.

Bezos has outfoxed other retailers, too, by welcoming competitors instead of fighting them. Alongside its own wares, Amazon now sells other retailers' products, as well as used items. They are all on the same web page. This sounds suicidal, and when Bezos first proposed the idea in early 2001, that's what most of his staff thought. Amazon was a retailer still in the early stages of developing relationships with suppliers. Now it was going to tell these suppliers that they would be competing on the same page against sellers of new and used items? "He saw eBay as an emerging threat, and he saw this as an opportunity to overtake them," says an ex-Amazonian. "He bet big and put hundreds of his best people on it. But it caused a lot of discomfort." The decision caused such a stir in the book-publishing community that the Authors Guild formally stepped in. They wanted used books sold on a different page from newly published ones.

Selling partners' used and new goods next to Amazon's own has become a cornerstone of its offerings. Amazon can do this because its warehouse operations are so efficient. Amazon earns about the same profit margins selling on commission as it does selling retail. In addition, the company doesn't have to advertise that its prices are lower, because consumers themselves can now compare prices from Amazon and other vendors. Sure this saves money, but it also breeds loyalty, Bezos says. "Giving people the choice to buy new and used side by side is good for customers," he says. "Give them the choice. They're not going to hurt themselves with that choice. The data we have tell us that customers who buy used books from us go on to buy more new books than they have ever bought before. They may not want to plunk down $25 for a brand-new author they've never tried. This lets them experiment."

The other benefit of Amazon's so-called marketplace strategy is that the revenue is almost pure profit. Amazon earns a commission instead of a markup for third-party transactions and incurs no inventory or warehousing costs. Almost 20% of Amazon's unit volume is now sold through others. Another dividend that Bezos counted on: Indirectly sold goods slow the need to add warehouse capacity.

Management by the numbers hasn't always translated into success. Bezos was a big believer in the dot-com economy and bought equity chunks of now defunct or struggling companies like Kozmo.com and Pets.com. The investments cost Amazon close to $350 million between 2000 and 2002. The bubble also prompted Bezos to overexpand. Within a year of adding six warehouses and taking the total to eight, Amazon closed two, laid off nearly 1,500 people, and took a $400 million hit in restructuring charges.

But Bezos was able to recover and make Amazon grow after the bubble. And it is since then that he has proved himself more than just a visionary. In the tech world, few founders have been able to manage their companies into the promised land. For every Tom Watson, Bill Hewlett, Bill Gates, or Michael Dell, there have been plenty more Sean Fannings.

The smart money is just now beginning to consider Jeff Bezos as possibly on par with the first group. Just ask Warren Buffett. He compares Bezos to Fred Smith, the founder of Federal Express. "Here's a guy who took something that is right in front of us--selling books--and put it together with new technology to create in just a couple of years one of the biggest brands in the world. With FedEx everyone knew about the mail and airplanes, but no one had put it together." After 30 years, Smith is still the CEO of FedEx, and his company is the world's largest overnight carrier. Nine years into his tenure, Bezos is bouncing away.

FEEDBACK fvogelstein@fortunemail.com