(FORTUNE Magazine) – It is one of those golden, glowing autumn days when, despite all the traffic and hassle of the new Northern California, you understand why people still flock here and never leave. Some 20 miles south of San Francisco the hills are gilded by the setting sun. But down in the bayside flatlands of Redwood Shores, the source of the dappling light would seem to be the Emerald City, the glittering blue-green complex of office towers of the Oracle Corporation. Brilliant, dazzling Emerald City, the would-be center of the e-commerce universe.

Here, inside building 500 on the 11th floor, CEO Larry Ellison, sleek as ever in his $7,000 suit, glides into a spacious office. Paneled in warm, blond wood and with a panoramic view of the giant man-made lake on the company's campus, the office seems part inscrutable sanctuary, part New Age power statement. In other words, vintage Ellison. But it turns out that this isn't really Ellison's office. It's actually the former office of Ray Lane, Oracle's ex-president--and erstwhile heir apparent of Ellison. Given the contentious nature of Lane's departure, it seems a truly bizarre venue for a chat about the company's future. But maybe Ellison is sending a message: My underlings are as replaceable as furniture. No employee departure can harm the prospects of this company.

If there is such a message, it's quickly lost in the "dialogue" that follows. For nearly an hour the 56-year-old Ellison careens from one unexpected topic to the next: from Mother Teresa to Bismarck to the Sacramento Kings' star forward Chris Webber to the late Supreme Court Justice Louis Brandeis to country singer Faith Hill to Al Gore's and George W.'s SAT scores to relational databases. And then, finally, toward the end of his fascinating harangue, Ellison alights upon his glorious epiphany.

This big revelation, this strategy-changing moment, came to him in the fall of 1997 when Oracle was doing $5.7 billion a year in sales--about half of what it does now. Yes, Oracle had been immensely successful. Yes, its database software business was flourishing, in spite of Microsoft's big push to sell a much cheaper database called SQL Server. And yes, traditional database rivals like Sybase and Informix were fading further and further from view. But Ellison knew that the database business was showing signs of maturing, and that his company would be hard-pressed to continue delivering 30% growth rates year after year after year--especially when faced with a slew of upstart competitors. Oracle had always thought its software would be the core of operations in corporate America, but now a host of new rivals--Siebel, PeopleSoft, and SAP among them--were offering programs that threatened to relegate the database to mere plumbing. Some observers viewed the threat as modest, believing that Oracle could simply coexist with these relative neophytes, whose applications were predicated on tapping into databases. But that wasn't Larry Ellison's take. That wasn't his take at all.

The question he faced was simple: Where would Oracle go from here? How could Ellison take his company to the next level--the leadership role he so deeply believed it deserved? And then it hit him, he says. Oracle would create a suite of Internet-based enterprise application software. These enterprise applications would work perfectly with Oracle databases. The resulting combination would be irresistible to corporate IT guys looking to make their lives simpler. It would be an unbeatable combination. And Oracle would wipe out its competitors, wipe 'em all right out, just as it had marginalized Sybase and Informix. "I realized this was the right thing to do for two reasons," recalls Ellison. "Number one, we're the big guys. We have the size and the brainpower to do this. No other software company does. If we're going to do applications, let's do it. Let's really do it. Two, I of course saw how incredibly successful Microsoft's Office suite was. Here we had a chance to do the same thing."

Whoa! Hold on here. The same strategy as Microsoft? Isn't this the same Larry Ellison who hired private detectives to dive into Mister Softee's dumpsters to dig up dirt on the company? ("Unsavory, yucky--kind of. But legal," Ellison says.) Aren't Oracle and Microsoft the bitterest blood rivals in techland? The new economy's Celtics and Lakers, Corn Huskers and Sooners, Hatfields and McCoys?

Exactly. And though Ellison may be flattering Bill Gates by copying him, he's also hoping to thrash him. To that end, Oracle's famously flamboyant chief has committed himself to a radical, bet-the-company strategy to build a colossus of finely integrated business applications that will do for the Net what Microsoft's software did for the old world of personal computers.

Indeed, one might argue that databases are already the system software of the e-commerce Internet, just as Windows is the key software of the PC realm (though the market share of Oracle's database software, roughly 40%, is less than half of what Microsoft's flagship program controls in PCs). So a suite of Oracle Net software applications, says Ellison, is a natural extension. Making Oracle the one-stop shop for database and application software is the right thing to do for his corporate customers. "It's like if you want to buy a car," says the high-energy CEO in one of his familiar rants. "Would you get an engine from BMW, a chassis from Jaguar, windshield wipers from Ford? No, of course not. Right now with the software that's out there, you need a glue gun--or hire all these consultants to put it together. They call it best of breed. I call it a mess. We want to put an end to that. Sometimes choice is a bad thing. If you say that our strategy is like Microsoft's in that it's giving customers what they want in one seamless package, then yes, it's like Microsoft."

On March 12, 1986, Oracle Corp. went public at $15 a share. It closed the day at $20.75, giving it a market value of $270 million. Twenty-four hours later, on March 13, Microsoft went public at $21. It ended the session up $7, giving it a market cap of $700 million. From that day on--now almost 15 years ago--these two wildly successful companies have become inextricably linked. Linked, perhaps as opposites, in the way north and south poles are.

In the bizarro universe of high-tech titans, the matchup is between California playboy and disheveled Seattle nerd. It's gonad vs. brain. Satyr vs. geek. For the longest time, of course, the geek had the upper hand. Microsoft was substance. Oracle was...a sideshow. Microsoft (until recently) never hit so much as a speed bump. Oracle crashed--and almost burned--not that long ago.

But then a funny thing happened. On Feb. 23 of this year the value of $10,000 invested in Oracle's IPO surpassed the value of $10,000 invested in Microsoft's IPO. The lines on the graph had crossed. Since then, they have continued to diverge. This particular point could be as insignificant as any Wall Street chartist voodoo, of course. But to Ellison, at least, it portends something monumental. He believes that Oracle will soon supplant Microsoft as the world's dominant software company. And if that occurs he believes that another critical thing will happen too: Ellison, who owns 24.2% of Oracle--a mammoth position for a CEO of an S&P 500 company--will supplant Bill Gates as the richest man in the world.

To some who know him well, this goal may matter even more to Ellison than his company's becoming the dominant software player on the Net. But then, knowing the real Ellison isn't easy. When Apple founder and CEO Steve Jobs (whom Ellison repeatedly refers to as his best friend) is asked what he thinks is the driving motivation of his pal, he chuckles, thinks for a moment, and then whispers into the phone: "Rosebud."

Born to an unwed mother who gave him to some relatives to raise, Ellison has never known his biological father and met his biological mother only once. Ellison says his family, Jewish immigrants from Europe, took the name "Ellison" after Ellis Island. He grew up working class in an apartment building in Chicago. His adopted father, Louis, was an accountant who was forever telling Larry that he was a good-for-nothing. According to The Difference Between God and Larry Ellison, by Mike Wilson, Ellison was considered smart but bridled under authority and was not a particularly great student. He attended the University of Illinois and the University of Chicago, quit both, and headed west to work for technology companies.

In 1977 he and programmers Bob Miner and Ed Oates founded a software company. They named their first product Oracle and sold it to various early clients, including the CIA. Ellison wasn't the star programmer--Miner (an extremely likable guy who died of lung cancer in 1994) and Oates were--but he was the venture's driving force. And even then he was a memorable character. One Oracle VP, who joined the company in 1981, remembers Ellison's picking him up at the airport back then. "He showed up in a Mazda RX-7, which didn't really have a back seat, but somehow there is this blonde back there. He drove to his home, and the gal got out and walked off. We went inside the house, and there was another woman inside." Also impressive was Ellison's technical expertise: "He knew chip speeds and drive specs," says the VP.

Identifying relational databases as a great business opportunity is one of several unquestionably brilliant moves Ellison has made over his career. Databases existed back then, but they were static and couldn't interact with each other. A maker of sweaters, for example, might be able to ask for the number of pink sweaters one customer had ordered but not how many pink sweaters all customers had bought. In the 1970s computer scientists at IBM wrote about a new kind of database that was flexible and could do ad hoc queries. It could sort through all the sweaters. IBM management failed to capitalize on the idea, however, and Ellison seized upon it. While it is common lore that IBM essentially handed over the PC system-software business to Microsoft, it is less known that Big Blue also took a pass on what would become Oracle's business.

And so Oracle began writing database software for Digital Equipment Corp.'s VAX minicomputers and mainframes. Customers such as government agencies and manufacturers used Oracle programs to manage their data, such as supplier lists and accounts. From the beginning, Oracle was a sales-driven, win-at-all-costs pirate ship of a company. Ellison drove everyone hard, going through wives and top executives like pairs of boots. He developed a reputation for falling out of love with Oracle execs just before their options vested or their bonuses were due. "For most people there really is only a certain number of years they can take," says one ex-executive. "It's not like he's necessarily a bad guy. He's just very, very intense."

Intense, complicated, and a bit off-putting. Ellison's like the kid that came up to you on the playground when you were in fourth grade and said, "Everything I say to you is a lie, including this." Friends and colleagues describe him as insecure, arrogant, brilliant, sleazy, funny, cool, and ruthless. A celebrity billionaire. A rogue. His own CFO calls him a dictator. He says he's engaged (it will be his fourth marriage) to Melanie Craft, an author of Danielle Steel-like fare such as Hard-Hearted Man. He's preparing to race for the America's Cup and loves to talk sailing with Walter Cronkite. A huge hoops fan, he says NBA commissioner David Stern has suggested that he buy an NBA team.

Indeed, gossiping about Larry is one of the most popular spectator sports in Silicon Valley. Here is some of the latest dish about him served up at Buck's, the Valley's famed eatery: Larry's maid stole his Rolex. (Not true.) Larry forged the smog sticker on his McLaren. (Ellison will neither confirm nor deny this one.) Larry stiffed his boatyard. Larry--a huge fan of Bill Clinton's--is going to hire former White House spokesman Mike McCurry to be his personal publicist. (Who knows?)

"He works very hard at his bad-boy image," says Ellison's pal, former Intel CEO Andy Grove. But what makes all the boats and babes and braggadocio bearable--barely--is the fact that Ellison and his company have just been knocking the cover off the ball. In a business that chews companies up and spits 'em out by the quarter, Oracle has made it big.

Here are the numbers: The company did over $10 billion in sales last year. It had over $3 billion in operating profits, which are growing over 40% a year. (You don't even want to talk net income. Last year Oracle logged a nearly $7 billion gain in marketable securities.) Its operating profit margin has marched steadily upward over the past two years, reaching as high as 41% in a recent quarter. Since September 1998 the stock is up more than tenfold. (I don't own ORCL, but I wish I did!) The company now has a $184 billion market cap, making it one of the largest in the world. Beyond that, Oracle--along with Intel, Hewlett-Packard, Cisco, and Sun--is now one of the Silicon Valley elite. Not bad for a company that used to be best known for vaporware and an insaniac CEO.

Analysts describe the new business Ellison is attacking as the "Internet applications business" and estimate it is a $30 billion market, growing by 25% a year. As for Ellison, he's dead set on wrapping his arms around the whole thing, starting with his suite of Internet-based integrated business applications, 11i, introduced this past spring. "I would be surprised if we didn't do it," he said in a recent conversation. A good many observers, of course, think that's crazy.

First of all, his scheme carries huge risk. The market for these software applications is changing at Internet speed and is enormously complex. Consider what these applications actually do: They take whole segments of a company's operations and streamline them--sifting through all the sales, financial, and customer data that a company might have and then distributing them to managers and employees in all sorts of usable packages. And now, thanks to the Internet, a company can tie suppliers and customers into this movable data feast as well. For instance, such programs can help salespeople on the road have access to the corporate data they need and can configure products on the fly for their customers. Still other applications totally reorganize supply-chain management. Not surprisingly, given the complexity of all this, Ellison's company doesn't have nearly the leverage in this world that Microsoft had when it started serving up suites of desktop applications to PC users.

And then, of course, there's the issue of how much the new strategy will benefit Oracle on Wall Street. Some analysts caution that even if the plan does work, its success is already factored into Oracle's pricey stock--right now ORCL has a forward P/E of 32, about the same as Microsoft's. "The debate isn't whether it's a great company," says Robertson Stephens analyst Eric Upin, "but how much Oracle's stock should be worth given the risk factors."

Over the past couple of years Oracle's "apps" sales have accounted for between 5% and 13% of revenues, depending on the quarter. And explosive growth beyond that threshold won't be as easy as Ellison makes it sound. One problem: Companies that are already leaders in enterprise applications--including PeopleSoft in the human resources category (see box); SAP, the giant German maker of enterprise resource software; and Siebel, in customer relationship management software--are migrating to the Web as well. What's more, a whole new crop of enterprise software companies has built entire businesses around the Web from the ground up: Ariba and Commerce One in procurement and exchanges, Broadvision in e-commerce, and i2 in the supply-chain arena. In any event the situation is so in flux that it would be hard for even the savviest Sand Hill Road player to pick a definite winner.

Meanwhile, there's a potentially big pricing problem with the Net apps suite strategy. Again, it's helpful to revisit the Microsoft Office analogy. When the Beast of Redmond began to bundle its desktop computing software for word processing, spreadsheets, and presentations, it was able to price the additional programs so dirt cheap that it knocked the competition (WordPerfect, Lotus) nearly out of the market. Oracle, however, would likely find such a marketing strategy--for example, essentially giving away an entire HR department's software for the price of a single program--to be a dangerous drag on its bottom line.

The aggressiveness of Ellison's strategy scares some. "Oracle hasn't been a leader in this business," points out Upin of Robertson Stephens, "and all of sudden they are talking about $1.3 billion in applications revenue for fiscal 2001 [May year-end]." Nor is Ellison's good buddy Andy Grove optimistic. "So often, when you see companies try to move from one area of dominance to another, they fail," says Grove. "I have my doubts about this one."

You don't have to be a software expert to understand what might be raising Grove's cautionary flags. Imagine taking all of a company's business software applications, ripping them out department by department, and replacing them with one global package. There's no small amount of institutional inertia that prevents companies from simply throwing over one app for another. These applications are to a large degree customized--and even though it will be far easier to make Internet-based apps fit a particular business without a lot of modification, it still is not an easy thing to pull off. Like crossing the desert on foot.

Nor is it easy to design standard apps that work for all the different kinds of customers that Oracle says it's trying to serve. An e-business has very different app needs than a manufacturer. A bank needs different kinds of apps than a trucking company.

Ellison says he's aware of all this. He knows its an enormous job. But he's got the programmers to do it. "No other company could create a suite of applications," he says. "They just aren't big enough."

In fact, Ellison's loyal lieutenants say that size factor--and Oracle's famously aggressive sales pitch--is already working. "I see CEOs come in here and get on their knees and weep after we explain this to them," says Mark Barrenechea, head of Oracle's CRM business. Says Oracle EVP Jay Nussbaum: "I went with Larry to visit Sandy Weill at Citigroup. And after Larry pitched Sandy, Sandy said, 'You don't have to sell me anymore. I'm a numbers guy. This is awesome. I want some of that.' " (Weill confirms the conversation but says he phrased it a little differently.)

Still, rhetoric aside, Wall Street was clearly a little disappointed in Oracle's apps biz for the quarter ended in August. Sure, applications revenue grew more than 42%, to $156 million, but analysts like Goldman's Rick Sherlund were looking for 65% growth. Likewise, though Oracle CFO Jeff Henley predicts the company will have 50 corporate applications customers by the end of the November quarter, the company has just about 27 now, with the recent signing of Agilent. Interestingly, most of Oracle's applications customers thus far have been smaller businesses or divisions--or newer ones that don't have many legacy systems to throw away.

"I think you'll be able to tell if they're on track over the next few quarters," says Morgan Stanley software analyst Chuck Phillips. "Oracle's sales force needs to really focus on applications. And the company could use one key executive besides Ellison who champions this cause."

One likely scenario is that Ellison's applications strategy will initially play hit-and-miss, making it a rough ride for shareholders. As the whole business evolves, though, many are putting their faith in Ellison's ability to reshape these products and make them work somehow, somewhere. That sort of finessing is something Ellison is very good at. "Who knows [what the outcome will be]?" says Jobs. "But at least Larry is moving forward, going somewhere that makes sense. Then if he has to change, he will. You can't get anywhere by standing still."

As with all things Oracle, in the end, it all comes back to Larry. And that may be the way he likes it. "Larry is misunderstood," offers Dave Roux, former head of corporate development at Oracle and now at buyout firm Silver Lake Partners. "Is he a great technologist? No, there are 100 guys in the Valley as good as he is. Is he a great manager? No, but he's been smart enough to get them. What he is is a great leader. His great strength is to make exceptional employees do the impossible."

To Roux's point, Ellison has become a master at rallying the troops. At Oracle's recent user group conference, the Larry faithful line up outside an auditorium for hours before his speech. When the doors open, 8,000 programmers race into the hall as if they're in some kind of frontier land grab. The pyrotechnics are cutting edge. Pulsing, futuristic sounds pound through the room. Abstract, flashing techno-images, like something out of The Matrix, blink across giant screens. And then Larry and his entourage march in. It's almost a rock star thing, with the buzz of the crowd and dozens of cameras flashing. Though surrounded by high-level executives, such as his enigmatic chief of staff, Safra Catz (the top woman executive at Oracle), marketing chief Mark Jarvis, and others--all wearing black--Ellison stands out. He's the tall one, wearing a gray suit this time, with the de rigueur black mock turtleneck. The whole thing is like a group of Romulan ministers surrounding their Romulan-in-chief. That's one metaphor. Sighs one employee watching: "Sometimes I feel like I work for Mephistopheles."

No question about it, the Ellison persona is powerful stuff. While some senior Oracle employees seem to get a kick out of him, others fear him (as in, "What kind of mood is he in today?"). He does not tolerate the uninformed. Says Oracle's IT chief Gary Roberts: "Larry will say, 'If I know more about your business than you do, we have a problem.' " Like Churchill, whom he professes great admiration for, Ellison likes to keep the message simple and repeat it often.

He is also a big believer in being a contrarian just for the sake of being one. "You can't win without being completely different," he says. "When everyone else says we are crazy, I say, 'Gee, we must really be on to something.' The louder they say it, the more excited I get."

It's true in his personal life too. He says his Japanese-style estate in tony Woodside will take almost a decade to complete. For years now it's looked like a gravel pit or a permanent construction site, pleasing the neighbors no end. Ellison isn't part of the club. Any club. "He will never be invited to Sun Valley," says a knowledgeable insider regarding Herb Allen's vaunted CEO confab. "Never." Ellison laughs in response. Or the Business Roundtable either, presumably.

But of all his personality quirks, the one that worries the Street most is Ellison's surprising ability to disengage from the company ship. Indeed, he has long had a history of popping in and out of a day-to-day leadership role, depending on how well the company is performing and how involved he is in his myriad outside interests.

Lately he has been very engaged. As it has become clear that the Internet is revolutionizing business, Ellison has charged back into Oracle, delving into parts of the company in which he had previously expressed little interest: "The way businesses worked changed," he says, getting up a head of steam. "Suddenly computers are central to the way we do marketing. Computers are central to the way we do selling. Computers are central to the way we do support. Hold it, hold it! I'm good at this computer stuff. I can help. I'm suddenly competent. An e-business--oh, I can run one of those. I'm good at that. A conventional business, I'm not very good at that."

Ellison has boiled down all this change to what he considers a couple of salient points about business: First, the Internet provides executives with incredible access to information. Second, you can take huge chunks of cost out of your business using that information. Third, Oracle's software can help executives do this. And finally, what better place to try this out than at Oracle itself?

And so Ellison has set out to make Oracle an e-business. A brutal evaluation of personnel has been instituted, with Ellison very much at the helm. "Larry asks, 'If you aren't making or selling Oracle software, explain to me again what you are doing at this company,' " notes one employee. Amazingly, even as Oracle's revenue grew almost 15% over last fiscal year, its head count actually fell 5.6%, from 43,800 to 41,320. The parsing involves hardware too. (For a detailed account of Ellison's cost-saving program, see the November 2000 issue of eCompany Now.) Oracle recently went from having 97 e-mail servers to having two, and from 140 global pricing models to just one. And so on. "When we started this effort last year, I thought we could save $500 million," says Ellison. "We ended up saving $1 billion, and I'm convinced we can save $1 billion, maybe $2 billion more."

So where is Microsoft in all this? Well, it's around. The software behemoth to the north has a product called SQL Server, a database product that has made some inroads. "SQL Server?" taunts Oracle executive vice president Gary Bloom. "Yawn. Yawn. Yawn." No, the folks at Redwood Shores don't have much fear of the folks up in Redmond anymore. Not for years, really--ever since Netscape became Oracle's "heat shield," to use Ellison's phrase. "Microsoft looked at Netscape and got terrified because they thought the browser would make Windows moot. They spent all of their efforts trying to beat Netscape, and they broke the law too," Ellison says in his most serious voice. "In the end they won that battle, but they ended up losing the war because the real important battle wasn't fought on the desktop. It was fought on the server. They ignored Oracle. Maybe they thought they'd take care of us later after they dealt with Netscape. Now, my cat could write a browser, but the Oracle database? Very complicated. They'll never catch us now."

Do Ellison and Gates talk? "We used to years ago," says Ellison, who claims he cut off the conversations because Gates was poaching his ideas. [A spokesman for Microsoft declined comment.] "Bill did call me up a few years ago and invited me up to Redmond," Ellison continues. "He wanted me to speak to his people. He said, 'Come on, we have a lot of smart people up here.' I said, 'I know. That's why I don't want to go up. Why should I go up there and talk to them?' "

It's an early October morning at the Emerald City, and Jeff Henley is sputtering mad. Some say the 55-year-old Oracle CFO has his foot out the door--he's been at the company for almost a decade and has $500 million of ORCL options and stock--but at the moment he looks plenty engaged. "Those !@#$%& idiots," he says, grimacing as he rushes by. "I'm going to issue a press release!" It's no wonder Henley is pissed. Oracle stock is in a free fall because of a presentation Henley made to analysts the day before, in which he said...nothing. Nothing new at all. Just reiterated what he'd said when Oracle announced its first-quarter earnings a few weeks before: "Don't worry, application sales are on target. Database sales are great, but don't expect them to keep growing 30%. The pipeline is full. If you loved [fiscal] 2000, you're going to love 2001." And yet immediately after his presentation at San Francisco's Palace Hotel, the stock begins to tank, ending the day down more than 12%. At that point Henley is surprised but not particularly upset. After all, there were rumors that Merrill Lynch downgraded the stock (not true), and this is earnings warning season. A little confusion here--no big deal.

Then the next morning, the stock hits the fan. On the front page of the San Francisco Chronicle's business section is the screaming headline ORACLE STOCK DIVES AFTER SALES WARNING. CNBC's Squawk Box reports that Henley's presentation looked "canned" and that he had "cautious body language." (This later amuses Ellison because Henley has just lost 20 pounds. "Maybe we need to fatten Jeff back up," he chuckles.) At the market open Oracle stock drops another 12% right off the bat. The general perception is that Intel and Apple had already warned about their business going forward, and that the previous day's analyst conference had been Oracle's turn. Oracle board members call the company asking what the hell is going on. And it's all in Henley's lap. And so at 3 P.M. that afternoon, Henley issues a press release: "There is no change in the outlook for Oracle's financial results.... We believe this is going to be a terrific year." The stock firms up a bit, but the damage is done.

In this market, if it wasn't the sales-warning-that-never-happened, something else would have pushed Oracle's stock down, as almost all tech stocks are well off their highs these days. But still, Henley is in a singular position. It ain't easy being Larry Ellison's CFO. "Larry will say one thing, and then I'll notch it down a bit," Henley explains. "I think the analysts are used to that." Like when Ellison says that Oracle will have 40% profit margins forever. Henley later tells the gang it's a stretch target. For reasons like that--Ellison's aggressiveness, to say nothing of his unconventional strategizing--Oracle stock has never been for the faint of heart. Even after the stock's incredible run over the past two years, lots of Wall Streeters have a traumatic history with Oracle. "The feeling is that this is a stock you can ride, but if you don't watch it, at some point Larry will burn you," says one trader. Witness the big air pocket in the fall of 1997, when the company began to move to Web-based applications. Sales dropped, the Street got scared, and the stock caved 40%.

Another area of concern for Wall Street is management. Who's running Oracle besides Ellison? Who's going to succeed Ellison? What happens if Larry gets hit by a bus? (Or crashes his plane? Or sinks his 78-foot yacht? Or drowns while bodysurfing?) For much of the past decade those fears were somewhat assuaged by the presence of Ray Lane. But Oracle's president departed suddenly in June, raising these issues anew. Lane came to Oracle in the wake of the company's near-death experience in 1990--a crisis born out of Oracle's cowboy sales culture. What happened is that the superincentivized and loosely supervised sales force was routinely exaggerating or simply falsifying numbers. (It was around this time that an Oracle sales chief instituted the infamous "Go for the gold" program, in which sales reps were actually paid in gold!) Write-offs and losses piled up into the tens of millions of dollars. The company faced all sorts of litigation. The SEC later fined Oracle for its behavior, and the board reportedly even discussed removing Ellison. On Oct. 31, 1990, ORCL bottomed out at (split adjusted) 13 cents a share.

Somehow Ellison managed to turn the company around. (The stock is up some 26,000% since then!) Oracle released its Oracle 7 product, vastly superior to the bug-infested version 6. ("A dog could sell 7" is how one Oracle exec put it.) The company's migration from specializing in writing programs for minicomputers to making software for Unix-based servers--another one of Ellison's critical calls--began to pay dividends. And just as significantly, Ellison hired some experienced managers, including Henley and Ray Lane from Booz Allen. Today there is a fair amount of revisionism going on at Oracle that diminishes Lane's contributions, but especially in the early and mid-1990s, there is no question that he helped pull Oracle up and take it to the next level. "He professionalized the sales and service organizations," says one ex-Oracle exec. "He sold Oracle to CEOs. And he stood up to Larry. There were just a few who did that."

But Ellison went from perceiving Lane as a great asset to the organization to seeing him as a hindrance. The roots of this change go back to 1995, when Ellison, who was in Paris at the time, grandly announced the death of the PC. Largely ridiculed then, Ellison's notion hardly sounds far-fetched today. "I came to realize that computing power doesn't have to reside on the desktop," Ellison recalls during an interview, "that putting the power onto the PC was distributing complexity. Computing should be like water or electric power. You don't have a nuclear power plant at people's houses. Computing should be centralized."

What made this vision possible, of course, was the Internet. But instead of gradually moving the company (and customers) to the new model, Ellison wanted a revolution. In 1997 he told his troops they would make no more client/server-based software. This caused an uproar in the sales staff. Lane called Ellison into a meeting with a dozen or so other top Oracle execs and told him in no uncertain terms that they had to keep doing client/server. "I looked around the room," recalls Ellison, "and said, 'Gee, I'm sorry. I didn't know you all felt that way. Okay, we'll still make client/server.' " Then Ellison pauses for effect: "But I was lying." Hogwash, says Lane. "I simply told Larry that we had to make better application software. Ours was terrible then."

By 1998, sources say, Ellison was looking to force Lane out of the company. It wasn't done according to anyone's HR handbook. "Larry took responsibility away from Ray hoping that he would just leave," says one source. Lane didn't.

Then, last spring, Ellison brought in Safra Catz, Oracle's former banker at DLJ. Catz is a mystery to many Oracle watchers. "She's very smart, and Larry really depends on her," says one source. "When you're around the two of them you feel like the third person in the hot tub." Some suggest that Ellison pitted Catz against Lane. In any event, it's pretty evident that they didn't get along. It's also fair to say that Catz's arrival did nothing to delay Lane's departure. Finally, Lane took his $700 million-plus package of Oracle stock and options and left, soon thereafter joining VC firm Kleiner Perkins. Ellison took Lane's title and divided his remaining responsibilities among Bloom, Henley, and Catz.

"Oracle is sort of like the Red Sox," says one insider. "It's always been, 'Where's the pitching?' It's Larry and the seven dwarfs. At other companies they move execs around to give them different experiences. They don't do that at Oracle."

That's not to say there's no management depth at Oracle. "I could run this company," says Gary Bloom, who has operational expertise, though the word is that he needs more seasoning in the leadership department. Ellison also points to Safra Catz as a possible future CEO, though insiders scoff at that notion, saying it's simply Ellison's way of building her up. "Hey, this is the team I had over the past two years," says Ellison, apparently forgetting about Lane, "and they did pretty well, right? What's not to like?"

So how long will Ellison stick around? "Larry is fully engaged right now," says his buddy and fellow iconoclast Jobs. "He isn't going anywhere." Indeed, last year Ellison decided to forgo his salary for a monster 40-million-share option package with a strike price of $7, worth well over $1 billion. (Not surprisingly, Ellison doesn't believe options should be expensed.) And Ellison himself suggests that he'll stick around for another three or four years ("the duration," as he calls it), at which point he believes the great application war will be over. In the meantime, he says, a successor will emerge, "and there are several candidates." But one insider says the Oracle board is "petrified" about succession. "There is no one in sight, and they've been pounding away on this topic for years."

The fact is, though, it will be extremely difficult for anyone to succeed Larry Ellison. Ellison, for all his foibles, built Oracle brick by brick. It's his baby. He is the only tech company CEO who has launched a business in the era of mainframes and taken it to client/server and then to the Internet. At some point he'll have to let go. "After this is all over I'm going to run Quark, this Israeli biotech company I own," says the next richest man in the world, who, as it happens, is deeply interested in the science of longevity. Makes sense. Anyone with this much ambition might have to live forever.

REPORTER ASSOCIATES Julia Boorstin, Jessica Sung

FEEDBACK: aserwer@fortunemail.com