(FORTUNE Magazine) – Bill Gross is one of the uniquely creative people I've ever met in my life. --Compaq co-founder Ben Rosen

Bill Gross is a modern-day P.T. Barnum. --Anonymous posting on f***edcompany.com

Bill Gross is both those things. He's also the most amazing entrepreneur I've met in 20 years of covering business, one of the smartest Internet thinkers I've encountered, a stupendously poor manager, and a tragic figure of the Internet bust. But that's getting ahead of the story.

So let's start with a few cut-and-dried facts. Gross, 42, is the founder and CEO of Idealab, a privately held "incubator" in Pasadena, with 100 employees there and in four satellite offices. Gross popularized the incubator concept when he founded Idealab back in 1996, spawning a movement that now includes several dozen imitators, with names like Divine interVentures and eCompanies. A few of the better-known startups to come out of Idealab are eToys, Citysearch, NetZero, CarsDirect, Cooking.com, FreePC, Tickets.com, and GoTo.com.

Now that the market has collapsed, it is common to scorn such dot-coms and their incubator parents as prime examples of dot-com mania. But it's worth remembering that not all that long ago people swooned over companies like eToys, anticipating the day when their brick-and-mortar competitors would plead for mercy. The Net cognoscenti, meanwhile, viewed incubators as one of those wonderful Internet-inspired innovations that would change business forever--"the Holy Grail" of company creation, as one former Gross disciple intoned. As the champion of the incubator, Gross was seen as a true visionary of the new economy, right up there with Jim Clark and John Doerr.

That is the reputation Gross enjoyed last May, when I first started exploring this story. The market had turned--in retrospect, it's clear that the Nasdaq's long tumble started in March--but neither he nor I knew that then. Gross, who had raised $1 billion in private financing in late January, was absolutely certain that the market would come roaring back. In April, Idealab had filed a registration statement with the SEC, announcing its intention to go public. Typically, Gross thought that his IPO would be a lot more than just your run-of-the-mill public offering. The Idealab IPO, he said at the time, might well provide the spark to rekindle the entire market, just as eBay's IPO had revived the flagging bull back in the fall of 1998. And you know what? I believed him.

I believed him because I was dazzled by him. A small, wiry man, Gross had an infectious, boyish enthusiasm that was charming and irresistible. He spoke so rapidly--jumping from topic to topic as if he were hyperlinking--that it was hard to keep up with him, and had so much energy he seemed constantly on the verge of jumping out of his own skin. He bubbled over with irrepressible optimism.

And his brain! That's what really set him apart. You could practically see the ideas bursting out of it, one after another, each more offbeat, more original, more promising than the last. The sheer profusion of ideas--and the way he got excited as he described them--was a large part of his charisma, I later realized: It was his overpowering intellect, more than any personal charm, that drew people to him and caused them to buy into his vision.

And then there was the fact that he seemed to have serious business chops too; this mad scientist had the soul of an entrepreneur. He appeared to have an instinctual sense of how to operate in "Internet time." His model for starting companies seemed flexible, open, and Darwinianly efficient all at once. Old economy execs flocked to his companies, hungry for the opportunity to operate in such a cutting-edge environment (especially when they were sure their options would soon be worth millions). And they were going to work for a man who loved what he did. "When we get an idea and start working on it, and then make a company out of it," he said excitedly--here he paused for a few seconds of dreamy contemplation--"that's just the most awesome thing." He made it sound like good sex. Back then, Gross seemed to embody everything that was cool and new and great about the Internet.

And then I kept reporting the story, and the stock market kept crashing, and Gross' companies started imploding, and he made a series of spectacularly stupid decisions, and he blew through $800 million with nothing to show in return. Which is why now, even though Idealab probably has enough money to survive this tough time--and even though some of its newer companies are quite promising--I understand that Bill Gross doesn't represent just the promise of the Internet. He represents its dark side as well.

The essential condition of Bill Gross' business life is that he has always had more ideas than he knew what to do with. Some were brilliant, and some were loony, but the fact that they poured out of him practically unbidden is what drove him to start Idealab in the first place. A graduate of Caltech, Gross has always been half scientist and half entrepreneur; while still in college he sold stereo systems built around a speaker technology he had patented. As a star at Lotus Development in the 1980s and as the founder of Knowledge Adventure, a successful multimedia company (it was sold to CUC for $100 million in 1997), Gross had to repress most of his ideas for the greater good. After all, most business success depends on a very different cast of mind from his: an ability to focus single-mindedly on one task and bring it to completion.

The arrival of the commercial Internet in the mid-1990s made Gross think that maybe, at long last, he had found a vehicle that might give his brain free rein. He could start a company that would bottle his ideas and turn them into successful Internet businesses. He'd "incubate" companies--yes, that's what he'd do! He'd turn the venture capital concept on its head--instead of raising money and waiting for ideas to walk in the door, he'd generate ideas and then go raise the money from the VCs. He and his crew would work over the business models, recruit the executives, help the companies get their Websites up and running, and create separate equity pools for each company. Then, when the companies were ready, he'd spin them out the door. That this was a radical way of creating companies did not give Gross a moment's pause. Why would it? Trying things other people hadn't thought of before was what he lived for.

Even before he founded Idealab, Gross started his first Internet company, Citysearch. (For more on Citysearch, see "Barry Diller's Dot-Com Nuptials.") It was September 1995--just a month after Netscape went public. Gross originally wanted to create a Website that would post photographs of retail establishments in cities all over the country as a way to aid travelers. But Charles Conn, a McKinsey partner Gross had recruited to be CEO, wanted to go in another direction. "It wasn't really the picture that was central," Conn says now. "It was the idea of getting perishable information about cities. That, I thought, would work." No problem: Gross was perfectly happy to shift the model. This early experience would be repeated over and over again: Gross would start a company having no clue whether its business model would work; in fact, as often as not, he assumed it wouldn't work. But he was completely sanguine that he, and the executives he had recruited, could figure something out that would. He thrived on moving fast, testing what consumers responded to, changing midstream, and, of course, peppering everybody in every company with lots of ideas. As a result, Idealab has always had an undercurrent of chaos. Gross' first hire at Idealab, a young woman named Marcia Goodstein, learned how to become a businesswoman by trying to make order out of the chaos. Today, as Idealab's president and COO--my "alter ego," Gross calls her--she still plays that role.

As Idealab set up its first dozen or so companies, Gross tried to codify the incubation process. He had firm ideas about the need to be generous with equity, about the importance of passion among the top executives of an Idealab company, and about the amount of money Idealab should invest in a company. That amount was amazingly small--just $250,000--but the logic was actually quite sound. Gross believed that if an Idealab company couldn't land venture capital beyond that original sum, then the idea probably wasn't all that good to begin with. Another critical test was whether Idealab employees decided to join a new company; if no one internally was interested in pursuing an idea, then it probably was a lousy idea. Even with these rules, though, life at Idealab was unlike anything else in corporate America. It was more like business jazz, a free-flowing improvisation, with Gross as Charlie Parker, Louis Armstrong, and Duke Ellington rolled into one.

The world was astounded. Reporters marveled at how Idealab operated on "Internet time," going from idea to startup in the blink of an eye. When Idealab startups like eToys and GoTo.com went public in 1999, the stocks skyrocketed. Other self-styled "serial entrepreneurs" made pilgrimages to Idealab's Pasadena headquarters, where they would invariably be wowed by the setup. The whole process seemed so cool that some of these visitors set up their own incubators, using Idealab as the model. Inc. magazine went so far as to claim that Gross' incubator model had the potential to transform the very nature of company creation in America. In truth, that's what Gross believed himself.

Which is not to say that everything ran perfectly. There was a certain ruthlessness built into the Idealab model--not every company was bound to succeed, of course, which meant that people who had been lured from well-paying jobs could find themselves out of work six or nine months later. Since Gross knew far more about how to start a business than how to run one, there was a certain amateur-hour quality to Idealab. And there were tensions between Gross and some of his CEOs, who found his incessant suggestions and e-mails annoying and even counterproductive.

At the time, though, this was easy to overlook. In 1998, Gross was even asked to pen an article for the Harvard Business Review. His piece focused on what he had learned about creating companies. One key, he declared, was sticking to the $250,000 limit. Only once had he ignored that limit, he wrote, and that was with a company called Ideamarket. Gross had been so optimistic about the prospects for Ideamarket--which was intended to be a marketplace for intellectual property--that he continued to pump money into it, in dribs and drabs, until the total reached $810,000--when he finally pulled the plug. "I'm convinced," he wrote of that experience, "that too much help early on will ensure failure later. If companies don't know that they can run out of money, they won't be thinking of ways not to run out of money."

In Internet circles, the conventional wisdom holds that Gross' big mistake was in concentrating so heavily on e-commerce companies. There is some truth to this charge, but it's a pat truth and not particularly instructive. After all, many people were concentrating on e-commerce in the late 1990s; one of them, Amazon's Jeff Bezos, got himself named Time magazine's Person of the Year. Let's be honest here: We all once believed in e-commerce--journalists and investors and Net moguls alike--and we all turned out to be wrong. In retrospect, even with eToys now trading at 16 cents a share, Gross' decision to focus on those companies seems pretty forgivable.

No, his real mistake was that he allowed his natural optimism to short-circuit his judgment. True, the Internet mania of 1999 and early 2000 warped a lot of people's judgment--but few were swayed to the extent that Gross was, and to such disastrous effect. His natural self-confidence morphed into a crippling arrogance, and because his personality was so magnetic, people who worked alongside him--people with lots of business experience, people who should have known better--were swayed right along with him. Caught up in the Bill Gross force field, they couldn't say no.

Like most Internet tragedies, this one began with the stock market and its crazy Internet valuations. Gross, alas, didn't believe that the valuations placed on Internet companies were crazy at all; on the contrary, he thought they were a well-earned validation of everything Idealab had been doing. When eToys went public in May of 1999, for instance, it was worth $7.8 billion--and Idealab's original $200,000 stake was suddenly worth $1.5 billion. Idealab's search engine startup, GoTo.com, went public a month later--and was worth $5 billion by November. Idealab owned about 20% of it.

According to Gross' original plan, after an Idealab company went public, the incubator would sell off some shares to replenish its coffers and to reward Idealab's employees and investors. But with the moment of truth at hand, Gross couldn't bring himself to sell the stock. He could cite lots of justifications. For instance, Idealab didn't like the tax consequences of distributing the stock. Besides, by holding on to the stock, Gross could claim that he was standing by his companies. Indeed, he began arguing that his refusal to sell exemplified an important difference between Idealab and a venture capital firm: His incubator didn't cut and run the minute a company went public.

But there was another calculation at work too: If eToys was worth $7.8 billion on its first day as a public company, just imagine how much more it could be worth in the future! In Gross' mind, Idealab companies could only go in one direction: up.

Gross' next bad decision, which came in the same general time frame, was to dramatically expand Idealab. Believing that Idealab had largely perfected the art of company creation, Gross now thought he could juggle far more than the handful of companies he was starting each year.

Eventually Idealab became a pretty big company, with 250 employees and four brand-new offices in Boston, New York, Silicon Valley, and even London. As Idealab expanded, Gross wooed corporate heavyweights, such as Robert Kavner, an early investor who had been a bigwig at AT&T. Kavner joined Idealab as a member of its executive committee and the head of its Silicon Valley office. (Later even more august company signed on: Both Compaq co-founder Ben Rosen and General Electric CEO Jack Welch joined Idealab's board.)

But as Idealab got bigger, Gross began to abandon all the rules he had once established to guide it. Remember the $250,000 limit? Suddenly Gross was no longer worried that "too much help early" would lead to failure later on. His new concern was that he didn't want to share the potential upside of Idealab companies with venture capitalists. So Idealab started investing large sums in its companies. And that passion he demanded in Idealab company executives? Within half an hour of meeting people he was telling them that they were destined to be rich if they joined an Idealab startup. He also threw up a host of me-too companies, on the theory that anything others could do on the Internet, Idealab could do better. "The hubris," says one Gross watcher, "was incredible."

Of all the mistakes Gross made, three stand out as both particularly representative and particularly horrible. Not surprisingly, they all took place early last year, just as Internet mania was peaking.

The first involves that $1 billion that Idealab raised in January 2000. Gross had no trouble finding sophisticated institutional investors, including T. Rowe Price, Dell Computer, and BancBoston Capital. In return for the investment they got 13% of Idealab, which valued the company at around $9 billion. But the money came with a catch. The investors insisted that Gross agree to a "ratchet." That is, if Idealab went public at a valuation of less than $9 billion--or even if it raised additional private financing--the investors would be given a bigger stake in Idealab to keep them whole. Indeed, if Idealab's valuation dropped low enough, they could conceivably wind up owning the majority of the company.

It boggles the mind to think that someone as smart as Bill Gross could agree to something so stupid. What could he have been thinking? But it's obvious what he was thinking. He was thinking that there was no way in the world that Idealab could ever go public at a valuation of less than $9 billion. But today, with Idealab's planned IPO a distant memory and its valuation presumably in the tank, Gross cannot tap the capital markets without turning over his incubator to the investors--a situation that will not change unless he can somehow negotiate the ratchet away.

That wasn't the worst of it. No, the worst was that once he had the $1 billion, Gross began going through it like water. That was mistake No. 2. To be fair, the money had been raised partly to ensure that Idealab not run afoul of SEC rules and find itself regulated as a mutual fund. To win this nod from the SEC, Idealab had to boost its stake in a number of its highflying companies--which, given their valuations at the time, was bound to be expensive.

But still! Idealab plowed $200 million into CarsDirect, its unprofitable auto-buying site. It engaged in a stock swap to up its stake in GoTo.com, paying an average of $80 a share. There went another $220 million. It pumped $60 million into a me-too company called Homepage.com, a personal-home-page company. (Recently renamed Frontera, it is now in the Web-hosting business.) Another $10 million went to an Idealab entertainment company called Z.com. Idealab even began investing in companies it hadn't started, acting as a pure VC. For instance, it put $16 million into Scout Electromedia, which was trying to make a little wireless device with regularly updated city information. By the time the spending spree was over, Idealab had gone through $800 million. In eight months!

And the third mistake? That had to do with one of Idealab's more high-profile startups, Eve.com. Eve was another of those me-too companies Gross launched in 1999, a high-end beauty site that joined a raft of such sites being launched all over Silicon Valley. As e-commerce sites go, Eve was a hit--it quickly became No. 1 in its category. The fact that it had no clear path to profitability--well, nobody worried about things like that in 1999, least of all Bill Gross. Which is why, when LVMH made an offer to buy Eve, for around $100 million, Gross once again took leave of his senses. Instead of letting Eve accept the offer--which would have been a huge win for Idealab--Gross instead made a counteroffer: Idealab would buy 80% of Eve for $110 million. Pleading with Eve's founding executives, Mariam Naficy and Varsha Rao, to sell the company to Idealab instead, he made his nuttiest deal yet: He guaranteed--yes, guaranteed!--that the stock options held by Eve employees would be worth a minimum of $50 million within 18 months. As the largest option holders, Naficy and Rao were being told, in effect, that they would each get $17.5 million within a year and a half, no matter what happened to Eve. Not surprisingly, they accepted Gross' offer.

Can you guess what happened next? Of course you can. By the time the deal was sealed, it was May, and it was already clear that Eve wasn't worth anywhere near what Idealab had paid for it. (In April, Estee Lauder announced it would buy an Eve competitor for a reported $20 million.) Worse, when Gross focused in on Eve's business, he finally realized it didn't work. He brought in some retail turnaround specialists and promised to use some of the proceeds from Idealab's IPO to sustain the company.

The next five months illustrated the downside of the Bill Gross approach. He was a whirlwind of ideas, none of which came close to saving Eve. For example, he wanted to turn Eve into a giant e-commerce portal, and he stuck two of his smaller e-commerce companies into Eve--a failing furniture site and a low-end jewelry site. The turnaround specialists, meanwhile, wanted to make Eve the online equivalent of the first floor of a big department store. But such a repositioning would be expensive, and as Gross gradually came to the realization that Idealab's IPO was not going to take place, he shut off the money spigot. He went to Naficy and Rao and asked them to put some of their guaranteed $17.5 million back into the company. The negotiations went nowhere.

By the end, there was recrimination on both sides, with each accusing the other of bad faith. In mid-October--with Idealab still on the hook for the options guarantee, and poised to lose $100 million in all--Gross shut Eve down. During the conference call to break the news, an Idealab executive told Eve employees that Idealab would help them find positions at other Idealab startups. The only response was the sound of derisive laughter.

By last fall Gross had developed a new refrain: We've learned our lesson. Looking backward for once in his life, Gross conceded that mistakes had been made. Yes, Idealab had run through the $800 million, and yes, Gross had seen his own reputation go from Internet hero to Internet goat. But almost immediately he decided that what had happened during the last year of Net mania had been good for Idealab--a giant learning experience that would make it a smarter, "more powerful" company. He would put his new "learnings," as he liked to call them, in place--and this time, he vowed, he wouldn't stray.

Thus the new, sobered-up Idealab was going to return to its roots. It would no longer play venture capitalist but would go back to the discipline of the $250,000 limit. It would insist that executives joining Idealab companies be motivated by passion rather than the desire for instant wealth. There would be no more me-too companies. New companies had to have the potential for high margins and a proprietary edge of one sort or another. In other words, no more e-commerce.

During the course of several lengthy interviews, Gross openly discussed what Idealab had done wrong in one situation or another: "We were acting like VCs without the skill or the interest," he told me one day after Scout Electromedia had abruptly shut down--taking Idealab's $16 million with it. But only once did I hear him even glancingly concede his most obvious mistake: the hubris that had enveloped him during the boom. That one time came in a speech Gross made to his staff a few days after the IPO was withdrawn in October. Most of the speech was devoted to the learnings and how they were going to put Idealab back on top. But at one point he admitted that Idealab had gotten greedy and had believed its own press. "We thought we could do no wrong," he said sadly.

In the next minute, though, he was back to his old self. The company had just had a big brainstorming session and had come up with "several multibillion-dollar ideas," he proclaimed, making it sound, once again, as if he could conjure these at will. GoTo.com, whose stock had dropped from $80 to $10, "could be a $100 stock if they show profitability." Z.com, which was close to running out of money, had just cut its staff in half to lower its burn rate. "Now it has a chance to get to profitability," Gross said.

Indeed, the more I saw of Gross, the more I was forced to wonder whether he had learned anything at all. Although he was pushing Idealab's companies to preserve cash, cut burn rates, and find profitable business models, he didn't seem to be applying the same rigor to Idealab itself. Its burn rate was $6.5 million a month--the same as during the height of the boom. It still had 250 employees. More startling to me was that during this same fall time frame, each of Idealab's four branch offices was moving into beautiful--and clearly expensive--new quarters.

Gross dismissed my concern about the new offices with a figurative flick of his wrist. "They only cost $1 million a month," he shrugged nonchalantly. "If each office comes up with just one $50 million company, they will justify their existence for the year. And I view a $50 million company as a complete failure."

His insouciance was startling, especially when combined with the fact that he still seemed so willing to launch companies without knowing whether their business models would work. When you looked at Idealab companies that had failed, wasn't their lack of a working business model a key factor? After all, look at Eve.com; if Gross had understood the beauty business in the first place, wouldn't he have had the sense to steer clear?

But Gross had his own set of counterexamples. Almost every Idealab company, starting with Citysearch, was operating with a business model that was different from the original conceit. Partly this was because the Internet was still so new that it was impossible to know what was going to work until you put something up and saw how people reacted. But it was also because of the way Gross' mind worked. For him, starting up companies was a voyage of discovery, in which half the point was to improvise some cool new way of making something work. (The other half, of course, was to create companies that would succeed in the marketplace.)

The most recent example was a company called Blastoff, which had been formed to send a rocket to the moon. (I'm not kidding.) It was obviously a loony idea, completely unfundable, as things turned out. But in the course of trying to figure out how to monetize a moon landing, some Blastoff engineers devised a way to use peer-to-peer technology to stream a little television picture to the corner of a computer screen. Gross swung into action; before long Idealab had a new company called DesktopTV, which was attempting to put a little television screen on computers--and stream in networks with greater speed and clarity than anyone had done before. Though it was early in the game, it looked as if DesktopTV had a real chance to succeed. And if Gross hadn't started Blastoff, you see, he would never have stumbled upon DesktopTV.

So in returning to its roots, Idealab was going to continue to rely on Bill Gross to come up with ideas that he could then turn into new companies. In good times or bad, that was never going to change--and it gave people at Idealab faith that the company had a future, despite everything that had happened. I realized this one day when I asked Marcia Goodstein whether it bothered her that Gross was starting so many new companies. She looked at me like I'd lost my mind. "I say thank God he is still coming up with companies," she replied. "That's what makes me understand that we are going to survive all of this."

And indeed, if you too turn your attention to the future--which is where Bill Gross wants you to look, after all--you'll see that she has a point. If Idealab were only the sum of its e-commerce companies, its situation would be truly dire. But it now has a handful of new startups with great potential; and in virtually every case they are the result of some fiendishly clever Bill Gross idea--and his willingness to try things other people have never imagined.

For instance, he devised a way to use clickstream data from ISPs to peek in on just about any Website and see what consumers do when they get there. He has turned this idea into a company called Compete.com, which hopes to sell real-time data about Websites to interested competitors. He's got the .tv Corp., which is in the domain name business--it acquired the rights to the domain .tv from the tiny country of Tuvalu and is marketing .tv as a hip alternative to .com. He's got Airwave, which is trying to create a wireless network, using radio-wave technology, in places like coffee shops and gas stations, so that customers can be connected to the Web just by walking in the door and turning on their laptop or handheld device. He's got Faster.com, which uses peer-to-peer technology to allow people to cache their favorite Websites directly on their hard drives, giving them access much more quickly. To make this scheme work Gross even plans to "sell" bits of people's hard drives (with the users' permission) for caching purposes. Who else would think of something like that?

Most of these companies have attracted venture capital, even in this difficult climate. They've had success recruiting executives, even though the pay at Idealab companies remains heavily weighted toward stock options. They've all gained some "traction" in the marketplace. In other words, they've got real prospects.

Yet every time I found myself getting caught up in Gross' optimism about one or another of these companies, I'd suddenly be caught short. I'd be reminded of the awful mistakes he'd made in the very recent past. There was just no getting around it: He seemed perfectly capable of making the same mistakes all over again. His powerful personality contained both strains--the propensity to come up with truly brilliant ideas and the propensity to blow it. Ultimately, that is what made it so difficult to come to terms with Gross. "Sometimes with Bill, you smack your forehead and say 'Wow!" says one Idealab staffer. "And sometimes you smack your forehead and say 'Yikes!'" With Gross, was it more important to look backward or to look forward? In all the time I worked on this story, I never came up with an adequate answer to that question.

Finally, last month, I had one last conversation with Gross and Goodstein. They told me that Idealab had, at long last, cut its monthly burn rate, from $6.5 million to $2.5 million. It had slashed its staff from 250 to about 100; though most had latched on with various Idealab companies, some had been laid off, and that had been hard. Perhaps most important, Idealab was going to stop funding the branch offices. They were in the process of lining up strategic partners and preparing to become self-sufficient. With these moves, Goodstein said, "We'll have enough money to last for many, many years."

I'm glad to hear it. There is much about Bill Gross that I admire. Idealab's new companies are grappling with some of the Internet's gnarliest problems, and I want to see how they play out. And I want to see what wild idea Gross will dream up next.

Actually, I can tell you that now. Within a few weeks, Gross says, Idealab is going to roll out a new company that he plans to call New.net. He's figured out a way to get around the government restriction on domains--in fact, Idealab is attempting to patent its method for doing just that. "We are launching with 12 domains," he says, among them .store, .firm, .kids, and .chat. Using New.net, people will be able to buy URLs with these domains, and as a result, the .com monopoly will be destroyed once and for all. "It's going to be unbelievably huge," he exults. "It's going to revolutionize the Internet."

It sounds great, doesn't it? But then, with Bill Gross, it always sounds great.

REPORTER ASSOCIATE Ellen Florian

FEEDBACK: jnocera@fortunemail.com