The New Player [SUN MICROSYSTEMS NO. 150] Perpetual underdog Sun is now part of the tech elite. To stay there, it must merely hold off Wintel.
By David Kirkpatrick

(FORTUNE Magazine) – Think of Sun Microsystems as the Big-Picture computer company. From its founding 18 years ago, Sun's executives have been thinking Big Thoughts. Even now they're thinking big, about where computing, the Internet--and Sun--will go next. Says CEO Scott McNealy: "The thermostat, light bulb, clock radio--everything should be connected to the Internet." Elaborates chief scientist Bill Joy: "Every single object manufactured will be created with a wireless address and will carry an identifier." Like McNealy, Joy believes that everything will be connected via the Internet--except maybe paper clips. (And even that might be possible, just not practical.)

What in the world does this have to do with Sun's business, which is selling heavy-duty computer systems that can cost as much as $20 million? A lot, as a matter of fact. So far, future think has worked wonders for Sun. For years the company talked about how "the network is the computer," and the rest of the world just shrugged. Even though Sun is the only computer company that doesn't rely at all on industry-standard parts from Microsoft or Intel--it makes its own chips, the SPARC microprocessors, and its own operating system, the Solaris flavor of Unix--every machine the company has ever made was designed to be networked with other machines. When the Internet came along, Sun was ready. Spinning the vision and sticking to it put Sun in a very sweet spot. Now, says Rolf De Vegt, an industry consultant at Renaissance Strategy, "Sun has the strongest sense of direction among the computer hardware companies."

As the Internet's importance has grown, so has Sun's. In the past few years the company has matured from a scrappy underdog into a pillar of the technology industry. These days, when someone thinks about outfitting a Website, Sun usually tops the list. It has become a safe choice, much as IBM used to be. Sun is now in the same league as Microsoft, Intel, Dell, Cisco, Oracle, EMC, and, yes, IBM--one of the signature companies driving global change and wealth creation. It is a key partner of some of those companies (Cisco, Oracle, and, in a way, IBM) and an archcompetitor of others (McNealy and Microsoft CEO Bill Gates have an ancient rivalry, and Intel is mostly a foe--more on that later). While Sun is smaller than most of the other elite, its revenues have been climbing swiftly, lately at a rate surpassing 25% a year. In 1999, sales totaled $13 billion. In the past two years Sun's stock has rocketed up ninefold, bringing its market capitalization to $170 billion.

What happened? Well, let's go back to that grand vision thing. Sun is trying to be the company whose systems run the Internet. That's all. The irrepressible McNealy draws an analogy to Lucent, which provides big switches that run the phone network. "What Lucent is to the telephone, we'll be to everything with a digital or electrical heartbeat. Lucent's switch does dial tone, and ours does Web tone, which includes buy tone, sell tone, bill tone, auction tone, directory tone..." (He could extend this list indefinitely.) "I honestly think this could end up being the largest system-equipment business in the history of anything."

Unless you're a CIO or a Webmaster, you probably aren't aware of how much you rely on Sun. But every time you buy a stock on E*Trade, an old crock on eBay, or a book on Amazon, Sun is in the background. Most of the sites hosted by companies like Exodus use Sun machines. Many new dot-coms take it as a given that they should run their online operations on Sun gear. E-commerce as we know it would be virtually impossible without Oracle databases running on Sun computers. Says Lee Blaylock, CEO of ServiceLane.com, a Dallas startup that refers individuals and businesses to service providers: "Our back end is entirely Sun, because if you want to truly scale the Web, the proven platform is Oracle and Sun, period." Blaylock, by the way, spent seven years working at Dell Computer. Sun knows how to market to these techies. Says Merrill Lynch analyst Steve Milunovich: "Two-thirds of Sun's advantage is marketing. Customers tell us they buy Sun just because they feel they can't go wrong."

The signature technology that placed Sun at the center of this Net universe is Java, its home-brewed programming language. The day Java was introduced in 1995, it wasn't any easier to understand than the "network is the computer" cry had been ten years earlier. In its demo that day, Sun showed how Java could deliver a live stock ticker scrolling across a Web page. Big deal. But what made that ticker possible was truly radical. The Java language sends software code down a Web pipe not just to display text or pictures but to perform tasks. Before Java, you could see a spreadsheet--say, a table of possible mortgage payments--on a Web page; after Java, you could interact with that Web spreadsheet--which is why you can now plug in your own numbers and figure out exactly what you will pay for your new house. A lingua franca for the Internet Age, Java lets programmers write a single Internet application that can work on almost any device. In the past the only way a programmer could be sure her program would have a huge audience was by writing it for Microsoft Windows.

Java gave Sun huge buzz, helping the company's servers become the Web hardware of choice. But its real value may just be emerging now, as the language that can stitch together the new world of computing, which is less PC-centric and more about a huge range of devices. Everything from cell phones and pagers to personal digital assistants and even TV sets will be designed to carry Internet data. With Java inside, these machines can handle data and software from all kinds of providers. Microsoft has its own idea for this "interoperable" future--all the devices could run a version of its Windows CE operating system. In such a world Java might be less necessary. But Windows CE is viewed as unwieldy, and its competitors, notably the Palm OS, are entrenched.

Sun already has deals to put Java on wireless products from Nokia, Ericsson, Motorola, Sony, Palm, and the Symbian consortium of phonemakers that is jointly developing a new wireless operating system. Japan's NTT DoCoMo is building Java into its data-enabled cell-phone service, the world's largest. That means DoCoMo suppliers Fujitsu, Matsushita, and NEC will soon be producing Java-enabled phones. (DoCoMo phones will also use Windows CE. For more, see "Gates & Co. Attack Asia.")

Licensing deals such as these could eventually generate considerable revenues for Sun, but that's not why Sun is pushing Java so hard. Says Mike Clary, a top software exec: "The ultimate objective is to make sure we have an environment we can sell servers into." Companies that work with Java at the front end, that have it in their cell phones or set-top boxes, tend to want Sun on the back end too. Most of the data for DoCoMo's Internet phones already come from Sun servers. Last year Sun secured multiyear deals, each totaling at least $500 million, with Lucent and Motorola. They will build Sun equipment into the systems they sell to cellular providers. Sun also signed a big deal to help Vodafone, the world's largest wireless operator, build its global Internet data infrastructure. Says Renaissance's De Vegt: "The telecom carriers are the most picky customers in terms of reliability. And that's where Sun is really the market leader."

Though Sun sells these customers a variety of servers, the centerpiece is its E10000 (known as the Starfire), the biggest and most robust. It has become the trusted platform for doing the most massive Internet tasks, like managing an Oracle database that juggles 100,000 transactions per minute. The price of an E10000 can easily exceed $10 million a pop, and while the machines are not infallible--last year it was worldwide news when eBay, a big Sun customer, had several lengthy outages--they are enormously profitable for the company. Sun is the leading seller of high-end Unix systems, with a 39% market share in the third quarter of 1999, according to International Data Corp.-- a percentage that has been growing.

Microsoft, of course, is not willing to just concede this market to Sun. The software giant is waging a dogged anti-Sun campaign, aiming to convince customers to use its new, beefed-up Windows 2000 operating system instead of Solaris. Microsoft's Website devotes an entire section--called dot-truth.com--to disparaging Sun. A sample: "The hype: Sun claims to be a leader in system reliability and more reliable than Windows. The reality: Major customers, such as Quote.com, are switching from Sun to the Microsoft Windows platform because it offers better reliability."

Microsoft CEO Steve Ballmer is clearly a guy who spends a lot of time thinking about Sun. In an hourlong interview on the subject, he argues passionately that Sun's decline is as inevitable as, well, the sunset. He insists that efficiencies that flow from the vast volumes of the Intel/Microsoft world will overwhelm Sun with comparable solutions for a fraction of the cost. Since Sun's products are produced in much smaller volumes than those of the Wintel duo, it can't get anywhere near their economies of scale. Take the computer revolution's most basic building block, microprocessors: Texas Instruments, which produces the SPARC chips for Sun, turns out about two million a year; Intel, by contrast, last year produced more than 90 million microprocessors. That gives Sun a big cost disadvantage, which, says Ballmer, means that it must concentrate on high-end equipment, for which customers so far have willingly paid its premium prices.

Ballmer says that as Wintel systems get steadily better and more powerful, Sun will be trapped in a high-end niche. "They're heading for the rafters," Ballmer says. "They keep climbing farther, farther, farther up the tower. Thank goodness the tower still has some room, but someday they will find there's a ceiling up there." McNealy's response: "That's like saying Lucent is trying to move to the high-end switch market to escape the tin-can-and-string market. I'm not running from Microsoft. I'm chasing the big customer." Microsoft has been using this logic to predict Sun's demise for years. Meanwhile, Sun's growth rate has accelerated.

All the experts seem to agree that eventually Wintel systems could seriously challenge Sun. In a couple of years, Microsoft's Windows 2000 may be widely accepted as robust and reliable for heavy-duty back-end computing, which would hurt Solaris. Also, by then Intel should be producing in volume chips that process data 64 bits at a time, as Sun's SPARC chips do now. (Intel's current chips are slower, processing only 32 bits at a time.) And the low price that Intel could put on its own chips could make Sun's SPARC-based boxes seem more expensive than ever--too expensive, in fact. Some testing data already show that Windows 2000 on Intel chips significantly outperforms Sun on speed and price performance. But perceptions take time to change, and even Ballmer admits that Windows is still seen as crash-prone (though he insists it has always performed better than it was reputed to). What's more, Sun's many corporate customers don't want to go to the trouble of moving their mission-critical applications from Solaris to another platform, no matter how cheap.

Reassuring those customers that it is ready for any kind of future may have been the primary motivation behind a partnership Sun chose to enter into with Intel, in which the two companies were supposedly cooperating to make Solaris work on Intel's 64-bit architecture. But the partnership appears to be collapsing. Intel senior vice president Paul Otellini says Sun's effort was never in good faith: "Sun has not delivered on their promises. Solaris on Intel was always subordinate to Solaris on SPARC, even though the Intel architecture outperforms SPARC. Finally we said we understand their game." Sun officials profess confusion and surprise. Says Sun President Ed Zander: "I'm very disappointed in Intel. In the last month or two they've just reneged completely." He says Solaris performs much better than Windows on Intel chips, and that Sun will continue to work on the project regardless.

Zander may be posturing. He says, "We are all working on this, from the top down." But Otellini notes that no one at Intel has heard from either McNealy or Zander on the matter. And Sun certainly has an incentive to keep its customers wedded to its current systems. It's a high-wire act.

There's one other threat out there, and it's got as much momentum as anything in technology. That's Linux, the Unix operating system being collectively developed by volunteer programmers all over the world in the so-called open source movement. So far Linux, which is mostly run on Intel chips, can't rival Solaris for high-end applications. In fact many people believe that in the short term, Linux poses more of a threat to Microsoft's Windows 2000. But Linux devotees are sure that it will also challenge Solaris. Says Cliff Miller, CEO of TurboLinux, a small company devoted to enhancing Linux's high-end performance: "They're in for a run for their money two to three years out." More likely, Sun will find a way to defuse the threat by bringing Solaris and Linux closer together. There's no reason that Sun couldn't do well selling boxes with SPARC chips running Linux.

Of course, joining the tech elite hasn't created only enemies for Sun. The company has also acquired powerful allies, many of whom have reasons to battle Microsoft and/or Intel. For instance, America Online, a Microsoft adversary, contributed the Internet applications it acquired when it bought Netscape to a company called iPlanet, jointly owned with Sun. IBM, though it competes bitterly with Sun in computers, has been a big proponent and co-developer of Java. And Sun is almost joined at the hip with database king Oracle.

Still, Sun's stock almost certainly has to slow down, given that it trades at 110 times this year's expected earnings. CFO Mike Lehman says Sun will get $200 million in near-term annual cost savings as it dot-coms more of its own operations in sales, distribution, and customer service. But as sales approach $20 billion annually, it will be tough to keep growing rapidly. Merrill's Milunovich has a $118 price target on Sun's stock, which currently trades at $100.

Indeed, that Scott McNealy has positioned Sun at the center of the Internet may already be reflected in the company's stock price. But if the past is any guide, this is just when Sun may start making hay from its new Big Picture. By the time we start connecting light bulbs to the Internet, Sun will be selling the equipment to make it happen. And inside the company they'll already be hammering out a vision for what happens next.

FEEDBACK: dkirkpatrick@fortunemail.com