(FORTUNE Magazine) – Could anyone be surprised that Sun Microsystems CEO Scott McNealy named his son Maverick? The company he leads is a maverick too. To the supremacy of Intel and Microsoft, the rest of the computer world says, "Oh, all right," but Sun says no. Every other major player has begun a process of conciliation with, if not outright surrender to, the so-called Wintel dynasty. Sun offers nothing but alternatives: its own anti-Windows software, its own microprocessors that compete with Intel's, its own ideas on how to equip and run state-of-the-art corporate networks.

Is this a strategy that makes any sense? Well, customers are buying in. Five years ago Sun was a profitable but unremarkable $3.6-billion-a-year maker of machines for engineers. Today it is a dominant player in powerful computers for the Internet and corporate intranets, a booming market that has helped Sun kick its annual sales up to $8.6 billion. Sun has better than quadrupled earnings, from $173 million in 1992 to $762 million in the year ended June 30. And it's pocketing more of each dollar of sales: The net margin during that period is up from 4.8% to 8.9%. No wonder Sun's stock price has almost doubled in the past 12 months.

McNealy has also worked a transformation that may be just as important for the company's future--he's made Sun cool. Jim Moore, a computer competition expert who runs GeoPartners Research in Cambridge, Mass., compares Sun to IBM in its heyday, when customers viewed it as the repository of wisdom and competence: "Sun has suddenly become a thought leader for the whole industry."

With its network-friendly Java programming language, unveiled just two years ago, Sun has thrilled the industry with new possibilities. Java promises to let programmers write software that works equally well on any kind of device attached to a network, flowing from one machine to another. If the language lives up to Sun's claims, it could sap the power Microsoft enjoys with Windows' utter dominance of computing desktops.

Sun grasped the importance of networking well before Microsoft or any other major computer maker, and that insight largely explains both its growth and its current glow. From its vantage as a maker of workstations, beefed-up desktop devices used mainly by engineers, Sun observed that sharing data between computers was crucial to key business tasks. McNealy worked aggressively to transform Sun's product line to capitalize on networking. The company rejiggered its Unix operating system for workstations, called Solaris, to run servers that coordinate work and store data on networks. In 1993 it resuscitated an old corporate slogan, "The network is the computer." Though in earlier years only techies could make sense of it, this time the slogan helped galvanize sales. (McNealy liked it so much he even named his dog Network--further proof that being named by McNealy is a risky thing.)

Then, in 1994, came the explosion of the Internet's World Wide Web. Millions of users came to believe that the network was indeed the computer. Since Sun's Unix rivals, IBM and Hewlett-Packard, had been slower to make the transition, customers turned to Sun. According to the La Jolla, Cal., research firm Computer Intelligence, 26% of all Web servers in use in the U.S. were made by Sun, more than any other company.

Now, according to Daniel Kunstler of J.P. Morgan, $3.4 billion of the company's revenues come from sales of servers and associated hardware. Workstations account for $2.6 billion, services bring in another $1.7 billion, and software and other goodies add $900 million. That's a sweet mix, since servers, which cost anywhere from $14,000 to more than $1 million, carry an estimated 65% gross margin, on average. The typical workstation, on the other hand, sells for about $15,000, and margins average only about 38%.

Sun gets those margins despite the expense of controlling both the operating system and the chip on which it runs, like a pint-sized IBM or a mini-Wintel. It devotes several thousand people to maintaining Solaris, and another thousand to designing its SPARC microprocessors, which are manufactured for Sun machines by companies like Fujitsu. John Doerr, the Kleiner Perkins venture capitalist who funded Sun in its infancy and now sits on its board, proudly calls the company "the last standing, fully integrated computing company adding its own value at the chip, OS, and systems level."

Sun is now the leading provider of Unix-based servers, the reliable machines on which companies typically install the software applications that really run the enterprise, like SAP or PeopleSoft. Solaris is the Unix to beat. Brian Richardson, who analyzes the server market for Meta Group in Stamford, Conn., says Sun has "by far the most momentum" of all the Unix outfits trying to woo big corporate customers. The Java buzz has helped spur sales, even though the language does nothing to make a Sun server better than a competitor's. Says Ed Zander, who heads Sun's computer products division: "I've had doors opened the last year because CIOs want to hear about this Java thing. I go in with a Java pitch and walk out with a server win."

Exploiting Java to sell servers isn't some neat perk--it's a necessity, since the type of computer Sun was built on, the workstation, is going the way of the minicomputer. PCs incorporating cheaper Intel microprocessors are replacing more expensive machines made by Sun and others. Selling PCs that run Microsoft's Windows NT operating system, Compaq, Dell, and others are pretty much taking over the market for workstations priced at less than $10,000. That low-end business accounts for about one-third of Sun's $2.6 billion in workstation revenues.

In theory, NT will move up the scale and supplant Solaris servers. Says Microsoft CEO Bill Gates: "It's inevitable that NT will end up passing by Solaris in the areas where Sun got its sales in the past." Adds Charles Fitzgerald, the rabidly competitive manager who heads Microsoft's own Java efforts: "The PC industry is an amazing economic machine, and Sun has never competed with that machine before. And here we come. In the workstation business you've already seen it turn; the server business is next."

Fitzgerald may be overexcited. McNealy has done a great job ensuring that Sun's servers are insulated against the threat. While companies are happy to have inexpensive Windows NT servers handle their simpler networking tasks, they still rely on Unix for their most critical applications. The reason: Solaris servers crash a lot less than NT servers. Solaris servers are also "scalable," geek-speak for the ability to add capacity and speed as a customer's computing needs grow. Today's top NT server can get only four microprocessors working effectively in unison; Sun's can yoke together 64, resulting in a machine that's about six times more powerful. Says Richardson: "It will be three to five years before NT is considered a reliable enterprise platform."

McNealy, of course, trumpets such news at every opportunity. He is famous for his Microsoft snubs. His barbs are outlandish and rude, but they get Sun noticed. Says Tom Henkel, a server analyst at the Gartner Group in Stamford, Conn.: "With his Gates bashing, McNealy has single-handedly--or single-mouthedly--raised Sun's visibility." By attacking Gates as a megalomaniacal monopolist, McNealy is also tapping into a latent customer yearning--for a company to help keep Microsoft honest. Many CIOs worry about their dependence on Microsoft. They like to see a competitor apply pressure. Many feel that if Microsoft had real competition in operating systems, for example, it couldn't get away with delays in promised releases, such as its recent announcement that it will deliver Windows 98 months late.

By bashing Gates and playing the Java buzz for all it's worth, McNealy has done everything he can to ensure that Sun's top-notch machines are entrenched in corporate networks around the world. Says securities analyst Kunstler of J.P. Morgan: "Never ever underestimate Microsoft's ability to react, but in the meantime, every time we take a breath, Sun installs another server somewhere, and that's one more piece Microsoft has to take away if they want to kill them."

Even a business led by Javaman is subject to economic reality, of course. Eventually the economies of scale Wintel enjoys may catch up with Sun, rendering its hardware overpriced and obsolete and crippling its ability to develop state-of-the-art software. That's the argument Gates makes in pooh-poohing Java's importance: "To duplicate all the value in Windows takes an incredible R&D budget that won't be funded by overpriced servers in the years ahead."

Sun's resources do pale before Microsoft's--while Gates can tap a corporate cash hoard of $9 billion, McNealy's Sun has just $1 billion or so. But Sun doesn't have to develop Java by itself. It has organized its own "amazing economic machine" of companies banking on Java, including, among others, Netscape and Oracle. Sun's biggest ally is IBM, which sees Java as the glue that can finally link its many lines of computers seamlessly. IBM has 2,500 programmers working to improve Java. Sun has 1,500.

All that teamwork is no guarantee that Java will lead to meaningful revenues for Sun. As an industry standard, Java could generate licensing income, though so far such payments are minuscule. McNealy is trying to get Java to work on an enormous range of computing devices. Sun is already starting to sell stripped-down network computers called Javastations, boxes designed to get all their applications and data off the Internet. McNealy offers design and engineering assistance to anyone who wants to build a Java-based device--anything from smart cards to the "Java" ring that McNealy showed off at a recent conference in New York. After illustrating Java's versatility by opening a door with his Java ring, McNealy told the audience why he would push Java into every possible device. "I call it Microsoft's Achilles' leg," he said. "You can't imagine running NT on a smart card." Replies Microsoft's Fitzgerald: "It's absurd to claim that Java, scaling 'from ring to supercomputer,' will be competitive against solutions optimized for their particular segment. Different systems provide different capabilities."

The devices and the licensing revenues may or may not materialize. What is true is that the longer McNealy keeps Java alive as an alternative to Microsoft, the more time he has to persuade corporate buyers to purchase his expensive machines.

He also gets time to work on his relationship with the other half of Wintel--to try to open a rift between Intel and Microsoft. Sun offers a version of its Solaris operating system that runs on Intel's chips. Computer makers that once developed their own versions of Unix are starting to support Solaris. In August, NCR became the first company to move from its version of Unix on Intel to Solaris. Richardson of Meta Group thinks companies like Data General, Sequent, and Unisys could also switch soon.

The success of Solaris on Intel is a hedge against the possibility that Sun won't always be able to afford the R&D to keep its microprocessors competitive with Intel's. McNealy, of course, dismisses the idea that Sun might abandon Sparc. "There's no way," he says. "That would be like Pepsi getting out of the cola business just because there's Coke. Sparc is going to be the alternative to Intel." Asked whether he considers Intel friend or foe, McNealy answers, "Yes," before going on. "It's a customer, a supplier, a competitor, a channel, all kinds of things. You can't find a simple answer. Sun and Intel will be working more closely together. A lot of software is being written for Solaris and Java, and Intel wants to make sure they can provide a platform that runs that software. You notice they don't go around badmouthing Java."

Of course, any help that Intel gives Sun in extending the life of Solaris is an affront to Microsoft, which hopes NT will replace Solaris everywhere. But Intel CEO Andy Grove wants as many kinds of software on his chips as possible. If he can play footsie with McNealy without unduly antagonizing Gates, he expands Intel's market.

Despite the publicity McNealy has reaped from Java, despite his success in making a name for Sun in the corporate market, despite the fact that he has quadrupled profits, McNealy still has scads of skeptics. Some worry that Sun is spreading itself too thin, losing the focus that has made its Unix business the world's most successful. Some moan that Java will prove to be "just another programming language," as Bill Gates puts it. Others believe that the economics of this industry inevitably favor high-volume PC makers, and that Solaris will give way to NT just as surely as the sun gives way to the moon. They may very well be right. But for over a decade, Scott McNealy has confounded critics at every crucial junction. In retrospect, it's clear that McNealy's kid should have been named Maverick Jr. His father's the original.