Technical muscle and a history of innovation made Sun a Silicon Valley standard-bearer. It also blinded famously combative Scott McNealy to the coming Linux wars. Now he's fighting to survive.
| PLUS
| Bill Gates, Entertainment God
| Linus Torvalds: The Peacemaker
Stiff-legged and hunched over, Scott McNealy limps slowly to a wheat-colored couch in the corner of his office. His eyes are bleary, and a wrinkly wattle is forming about his neck. In his semi-exhausted state, McNealy looks almost frail. There was a time not that long ago when the smash-mouthed, overamped CEO of Sun Microsystems would have been shuffling along like this because of a nasty collision during a no-holds-barred intramural hockey game. Instead, the culprit is a long international flight home two days ago following a week-and-a-half swing through Asia. "I'm getting old," groans the 48-year-old McNealy.
Age may be one explanation for his woes, but there are others: his company's stock price, slumping sales, and customer defections, not to mention threats posed by a burgeoning list of competitors. Share prices of nearly every publicly traded Silicon Valley firm have plummeted since 2000, of course, but few have fallen as precipitously as Sun's. As of early May, its stock, down 94 percent from a 2000 high, was trading pretty much exactly where it was in 1996, when the computer maker was in the early arc of its dotcom-fueled rocket ride. In its heyday, Sun sometimes sold as much as $5 billion worth of computer equipment, software, and services in a single quarter, but over the past few years quarterly revenue has dropped to nearly half that. Revenue fell an additional 10 percent in the first three months of 2003, marking the eighth consecutive quarterly decline.
The problem is that Sun specializes in the kind of top-end systems – super-sleek servers running on scores of expensive, high-performance chips yoked together by Solaris, Sun's well-regarded operating system – that have fallen out of favor. Morgan Stanley, Credit Suisse First Boston, Pixar, and the E*Trade Group are among the major companies that have swapped out Sun machines in recent months for so-called Lintel boxes: cheaper servers running the Linux operating system on Intel-compatible microprocessors. Even Google, headed up by former Sun star Eric Schmidt, uses Lintel machines to run one of the world's busiest Web sites. Of the 20 Wall Street analysts who cover Sun, only one rates the company a buy. This month, Wired drops Sun from its list of the 40 most forward-looking companies, a group it has been part of since the list debuted in 1998.
Not long ago, Sun was celebrated as one of the Four Horsemen of the Internet. Companies looking to make their mark on the Web knew exactly which vendors to seek out: Cisco for routers, EMC for storage equipment, Oracle for database software, and Sun for servers – those centralized computers that pump out every page on the Web and every piece of email. And while Cisco, Oracle, and EMC have struggled through the downturn, none of the three has suffered nearly as much as Sun. Like all great rise and fall stories, Sun's saga is one replete with hubris, missed opportunities, and outright mistakes. But the story reduces down to this: McNealy spent the second half of the 1990s monomaniacally obsessed with everything having to do with Microsoft, from its monopoly-like practices to the general unreliability of the Windows operating system. Meanwhile, stalwarts like Hewlett-Packard and IBM began selling servers on par with Sun's most powerful and expensive machines. Dell and Intel, propelled by Linux, started cutting into Sun's core business at the low end. By the time Sun woke up to this new reality, the smart-guy pundits were asking if the company would be the first big casualty of Linux.
Now that Sun recognizes the problem, its response is familiar. The company is busily spinning out new products while simultaneously going on the attack. Sun's executives have a history of turning competitors into enemies, as if the company can remain truly focused only if at war. In the 1980s, IBM and HP were singled out as foes after they muscled into a market that had been Sun's specialty: powerful workstations favored by engineers, scientists, and designers. In the mid-1990s, McNealy declared war on Microsoft. These days, Sun's enemy seems to be conventional wisdom as propagated by Wall Street analysts and the press. More than a few people are declaring Sun yet another once-great technology company destined to become a niche player.
"Given the financial situation that Sun finds itself in," says Jonathan Eunice, an analyst with Illuminata, a Nashua, New Hampshire-based IT research firm, "customers have to be a little bit frightened of partnering up with them." Eunice, who has followed Sun for over a decade, is more bullish on the company than many of his cohorts, but he also argues that, for better or worse, perception matters. "Customers have to wonder how significant a player it'll be in five years," he says.
McNealy ridicules the prevailing view. "If it was wisdom," he scoffs, "they wouldn't call it conventional wisdom." Forget the stock price and flagging sales, he argues, and focus on Sun's record of innovation. Technology, he insists, will turn the company around. He talks up a next-generation chip potentially 30 times faster than the microprocessors currently running Sun's most powerful machines. He and his team also point to a pair of software initiatives, N1 and Project Orion, that some of the company's biggest brains are hard at work on. "With all we've got up our sleeves, mate," declares Andy Lark, the New Zealand-born marketing chief, "we've got the competition quaking in their boots." That's quintessential Sun: The tech industry is ready to write off the computer maker, but McNealy and company are smugly imagining future glory.
After nearly 20 years as Sun's CEO, Scott McNealy seems genuinely baffled by the notion that his company is doomed. Back in the mid-1980s, when Sun was still a startup, it had neither reputation nor intellectual property, and it faced a murderer's row of competitors. One quarter it even needed to borrow $50 million to make payroll. By comparison, says McNealy, Sun today is robust. And yet, "I look at the skepticism now and it's way higher than it was then," he says, incredulous. He argues that Sun is much stronger because of the past. The enormous profits it was able to generate in the second half of the 1990s mean that the company now has on hand $5.5 billion in cash and convertible securities. Those years also burnished Sun's reputation as an industry innovator. Millions of developers code in Java, the programming language Sun invented that cleverly dealt with the limitations of then-dominant C++ and at the same time allowed disparate computer systems to communicate more easily with one another. That era also showcased Solaris, the company's version of Unix, which proved to be reliable even when harnessing a hundred or more microprocessors simultaneously. Even rivals at IBM and HP, which sell their own proprietary forms of Unix, acknowledge that Solaris is the yardstick by which their operating systems are measured.
Yet during these boom times, Sun made a series of blunders the effects of which are still felt today. Illuminata's Eunice ticks off what he dubs "significant" tech glitches that were revealed once the bubble had burst, including memory problems with Sun's top-selling server that were aggravated by the company's insistence that customers sign a nondisclosure agreement if they wanted a patch. Bugs might be common in the computer industry, but Sun has always placed itself a cut above the rest, and attaching a price of silence to a fix didn't endear it to customers. The company also proved too aggressive in hiring, bringing on thousands of employees in the last quarter of 2000, when Nasdaq was deep into its swan dive. Sun eventually laid off 20 percent of its 43,300-person workforce, long after outfits like Oracle and Intel right-sized themselves for tough times.
Then there was the big boss's preoccupation with Microsoft. Until recently, it was not uncommon for McNealy to squander more than half a speech taunting Bill Gates and company, ridiculing their technology as inferior and their tactics as unsavory. In 1997, Sun filed suit, charging Microsoft with attempting to sabotage Java. (Sun has won some preliminary rounds of this protracted legal battle, which is currently before a federal appeals court.) In contrast, HP and IBM chose to make peace with Gates, since Microsoft represents a critical component in nearly any network.
Sun's costliest mistakes during the late '90s, however, were its failure to embrace Linux and its tardiness in recognizing Intel as a serious threat to its core server business. The server market comprises two general categories, low-end systems that lash together up to eight computer chips and sell for less than $100,000, and the high-end fire-breathers that connect scores of microprocessors and can cost into the millions. It's in the low-end market once dominated by Sun that Linux and Intel pose a threat. Linux, the open source operating system derived from Unix and branded by the well-recognized penguin mascot, has been a favorite of hardcore software developers since the late-1990s. Sun, home to some of the brighter, more creative minds in tech, would have been a natural place for Linux to take hold – and in a way it did. Back then, Eunice would regularly see Sun programmers at Linux user group meetings, but they stopped coming, he says, "because they got the message that the top execs at Sun weren't interested." Solaris was the superior product, and Linux was a threat to the company's bottom line. "We could've owned the Linux story," says one former Sun exec, if not for intransigence at the top. It wasn't until 2002 that McNealy literally donned a penguin suit at a Sun conference to announce that the company would begin offering customers who bought its low-end servers a choice between Solaris and Linux.
Sun's belated adoption of Linux was a major mistake for an outfit that has long characterized itself as an industry thought-leader quick to recognize crucial technologies. For instance, Sun was an early advocate of TCP/IP, the protocols that today are to the Internet what zip codes and street addresses are to the postal service. By the time McNealy put on his penguin suit, though, Linux had already caught the imaginations of CIOs inside even the richest corporate behemoths. Potential new customers might call on a Sun salesperson if contemplating a strategy based partly on Java, but few customers will think of Sun if seeking to work with Linux. And most are.
"You recognize the world's changing," says Michael Fay, who works in IBM's server group, "but it's just too easy to keep cashing checks." Just as IBM's attachment to its lucrative mainframe business meant Big Blue was late to enter the server market, Fay says, Sun's success blinded it to Linux.
McNealy's dismissive attitude toward Intel has also hurt the company. Sun's executives tend to scorn Intel's chips as slow, energy inefficient, and generally inferior to their own. Since the mid-1990s, Sun and IBM have used 64-bit microprocessors, which would seem inherently superior to the Intel-designed 32-bit chips installed on cheaper, low-end Lintel or Wintel (running Windows) boxes stamped out by companies like Dell. A 64-bit processor, after all, can gulp down information in bites twice as large as those of a 32-bit chip. Yet boxes running on 32-bit processors can handle most common applications as efficiently as a more expensive Sun server. Users running simulations for scientific study, for example, would find great advantage in 64-bit processors. But nowadays, those seeking a machine to handle simple database searches or Web-page delivery tend to buy servers based on 32-bit chips and mass-produced at a fraction of the cost of those Sun was selling at the start of this decade.
Last year, 88 percent of the servers sold worldwide were rigged with Intel-designed chips, according to research firm IDC, accounting for nearly half of all spending on servers. Meanwhile, the overall server market has shrunk, from $61 billion in 2000 to $44 billion in 2002. And the giant lead Sun once enjoyed over both IBM and HP in high-end systems has evaporated as both competitors have introduced lines that more or less match Sun spec for spec.
Sun finally cried uncle in August, unveiling a server powered by a 32-bit Intel chip that gave users a choice between Linux and an older version of Solaris. Then, in February, the company slashed prices on several of its low-end servers. Sun execs acknowledge that Dell is still better equipped to mass-produce machines at a lower price, but they also claim Sun's offerings have technical advantages and are more energy efficient.
"No one can deny that Sun was slow" to realize its machines were no longer superior, says Jean Bozman, an IDC analyst who specializes in server research. But, she also argues, the company "proved itself very nimble" once it recognized its mistake.
Phil Hester, a former IBM executive who is now the CEO of Newisys, a server-design firm, agrees, to a point. "There are a lot of very smart people at Sun who have proven themselves very good at diagnosing a problem and then figuring out a solution. They'll figure out how to reinvent themselves. How long it takes them to do that, though, and how deep the hole gets in the interim, and how big they are when they come out – that's a different discussion."
Inside the company, it's an article of faith that innovation will chart Sun's course through an uncertain future. In the early 1990s, tech's chattering class declared the company a goner because HP, IBM, and Silicon Graphics all had aggressively entered the workstation market, then Sun's specialty. Innovations like Java returned Sun to the spotlight, helping it beat back the competition. Today, Sun's leadership assumes the company's technical prowess will again propel them past their foes. "We like it," says Greg Papadopoulos, the chief technology officer, "when people underestimate us."
Sun's road map to a successful future relies on faster chips, automation software, and a scheme for its operating system more or less borrowed from the Microsoft playbook. Mark Tolliver, who as chief strategy officer occupies the office next to McNealy's, singles out the company's "throughput computing" microprocessors as the most compelling of the three approaches. When finally ready for market, he says, these new chips will process information at a rate 15 to 30 times greater than the outfit's current leader, the UltraSparc III.
Throughput seems ingenious. One of the major challenges for chip designers is what's called memory latency. That's when a semiconductor sits idly while awaiting bits from a computer's memory needed to complete a given task. "We decided to stop fighting memory latency and learn how to live with it," says David Yen, the executive VP in charge of Sun's microprocessor division. As Yen describes it, a throughput chip constantly initiates new tasks, so it never sits idle. For example, while it awaits the information needed to complete a complex database search, it might launch a second and maybe a third and a fourth database query. Juggling tasks simultaneously rather than serially is known as multithreading, and Sun execs contend it's possible on systems running its machines but not on Wintel or Lintel boxes.
"Multithreading isn't something you can cook up in a day," says Sun cofounder and chief scientist Bill Joy, who worked closely with Yen and his team during development of the throughput idea. "Basically, you get traffic jams inside the operating system if it hasn't all been engineered from the start to avoid bottlenecks." Sun also plans to improve chip performance by placing several processors on a single chip, an innovation its rivals are also hard at work on.
Tech analysts remain cautious about throughput. For one thing, says IDC's Bozman, Sun chips have run slower than those designed by rivals since at least the mid-1990s, so in part it's playing catch-up. For another, they're still a long way from release. "At this point, all I've heard is a lot of words," Bozman says.
Moreover, Intel already offers a simplified version of throughput, dubbed hyperthreading. Its chips can process two distinct tasks simultaneously by tricking the operating system into thinking it's working with two processors rather than one. Karl Freund, the VP of marketing for IBM's server division, questions the utility of throughput. "So you can handle 16 threads at once," Freund says. "It's not clear to us what kinds of applications can take advantage of that kind of multithreading." In response, Sun execs only shake their heads at the shortsightedness of their foes. "It's preposterous to think that we're not going to keep innovating to take advantage of new technologies," says CTO Papadopoulos.
Sun loyalists pin their hopes on two other big initiatives: N1 and Project Orion. Both fall under the purview of Jonathan Schwartz, the 38-year-old, ponytail-wearing executive VP who oversees Sun's 4,500-person software development group. Schwartz is part of a new team that's taken charge since the exodus of Sun's top brass in 2002 after McNealy demanded a five-year commitment to see the company through hard times.
The idea behind Project Orion is to fold into Solaris a long list of products that Sun now sells separately, such as software that allows a customer to offer instant messaging or to create an easy-to-manage portal. Unlike Microsoft Windows, which includes any number of bundled applications (a browser, a media player, games), Orion allows Solaris users to replace Sun's offering with a competitor's at the click of a button. As Schwartz explains it, Orion is largely about combating Linux, which he casts as a fine solution for linking low-end servers – assuming you have the personnel to integrate a mess of products from various vendors. He estimates that Sun will charge somewhere between $100 and $200 per license for the Orion stack, a fraction of what third-party vendors would charge for a similar array of products. McNealy jokingly refers to Orion as Sun's version of the Ginsu knife.
N1 may be even more important to Sun's future. The software is designed to let administrators more efficiently allocate the resources of large and complex systems at the touch of a button – even automatically. For example, if traffic on a company's server in a Topeka facility suddenly spikes, N1 will immediately rejigger the hub routers to send the spillover to its underused servers in Wichita. And, Sun promises, N1 will be compatible with software other than its own. "With N1," McNealy says, "the one application, one server model is over."
For now, Orion is nothing but a work in progress. And following through on it will be risky for Sun. Bundling a passel of expensive products into an operating system at bargain prices might generate more server sales, but it's also likely to cut into Sun's software revenue at a time when analysts are clamoring for the outfit to earn more in that category. Observers have long rapped Sun for the lack of cash generated from its software. There are some 100 million Java-enabled cell phones worldwide, but the amount Sun earns from licensing the software – somewhere in the tens of millions, according to strategist Tolliver – barely helps its bottom line.
N1 is also more an idea on paper than an actual product, shipping only on Sun's cheapest servers. IBM, HP, and Microsoft are touting similar products. McNealy claims that Sun has at least a one-year jump on the competition, but representatives from IBM and HP argue that they have a one-year lead over Sun. The truth is, it's still early on in what promises to be the next big battle in the server wars.
Sun thinks it has one other great strength that will carry it through tough times: the fighting spirit woven into its DNA and exemplified by its CEO. "Scott wakes up every morning thinking of ways to kill the competition," says Ed Zander, Sun's former president and chief operating officer. It's routine for large corporations to assign a team to study its competitors, but Sun is one of the few that actually employ a full-time chief competitive officer.
That zeal should serve Sun well as it takes part in what people there are inclined to call the "ground wars" – the often nasty competition for deep-pocketed customers. "The ground wars for the high end have been even more fierce than we had initially anticipated," says Shahin Khan, Sun's chief competitive officer until April, when he took over a unit working on machines geared for intensive scientific study. "The down economy has made everyone a bit more desperate."
The once mild-mannered HP has signs posted in its server division offices ordering: Flatten, pummel the competition. A VP for starched-shirt IBM casts Sun as "desperate" for making overhyped announcements about upcoming products long before their release so that customers will wait for the launch. All three companies have armed their salespeople with slide presentations chockablock with dubious and spurious assertions – the corporate equivalent of going negative in a political campaign. Sun's version is titled "Slides That IBM and HP-Compaq Do Not Want You to See."
To fight the ground wars, every Sun executive, whether longtime vet or young blood, has sharpened his version of what James Gosling, the Sun programmer who invented Java, calls "Scott's rant." John Gage, who has been with Sun almost since its start, gives his rendition over dinner one night in Palo Alto. Jonathan Schwartz offers a variation in his office days later. Both men focus more on why Sun should regain its rightful place among the tech world's elite rather than why it will.
Gage, who bears the title of chief science officer, is a cerebral presence inside Sun. A dapper man with droopy blue eyes and a well-trimmed gray beard, he talks at great length about the risk disgruntled employees pose to computer networks and the constant threat of server attacks from the outside. He ticks off the names of businesses recently disabled by temporary system failures and contrasts that with Sun systems, so reliable and secure that during the US attack on Iraq, the systems that enabled the generals to communicate with commanders in the field were running atop a customized version of Solaris. And yet, mystifying to Gage, big corporations still turn to outfits other than his own when making large system purchases. "To me, any CIO who makes their organization vulnerable to these trivial events should be fired," he says.
Schwartz, too, has a hard time understanding why a customer would do business with any company other than Sun. For him, it boils down to Sun's Apple-like insistence that it sell complete systems. Wall Street analysts may look askance at this strategy; after all, while Apple's technology is highly regarded, the company is a niche player with a 3 percent market share. But Schwartz believes they fail to recognize the "magical things" a company can do when it can tweak its own microprocessors to optimize the operating system it's been selling for 20 years. Sun will help customers implement Linux on low-end servers if that's their wish, Schwartz says, but only begrudgingly, because at Sun they believe in Sun, period.
So the same hardheaded pride that got Sun into trouble with Linux is alive and well. Nowhere is this clearer than in the office of Scott McNealy, where he keeps what he calls a "decapitated penguin" on a shelf (it's the head from last year's costume). He admits to tactical errors "too numerous to discuss," and then singles out just one: Sun was late jumping into the market for a cheaper set of offerings on the lower rungs of the server world.
Yet when talk turns to Linux, it's as if McNealy can't help himself: He knows he should be courting the world's Linux devotees, but instead he pokes fun at them. He points out that Red Hat, the leading purveyor of Linux systems, announced revenue of $24 million for its last quarter of 2002. "I don't know where this multibillion-dollar Linux business is," he says. "There's IBM math going on here somewhere."
It's a gratuitous dig that reveals what is perhaps Sun's greatest weakness: an obstinate CEO whose most basic instincts spur him to fight rather than accommodate a new landscape. Of course no one is earning billions packaging an operating system a key selling point of which is that it's free. It isn't a matter of the riches Linux will generate but the billions it's siphoning off a server market that shows no sign of rebounding anytime soon. Instead, Sun's top people are inclined to snicker over Solaris' superiority to Linux even as it – and, to a lesser extent, Windows – increasingly appeals to the world's best-endowed corporate customers. It's all so unfair, and at the same time it must feel entirely comfortable to McNealy: Sun is an outsider again, at war with the world.
The Eclipse of Sun The rise of cheap servers running Linux has overshadowed Solaris, Sun's high-end version of Unix. McNealy and company have yet to recover.
Unix vs. Linux Over three years, Unix's market share declined, Windows' crept up, and Linux's soared, as measured by sales of servers running those systems. Still, Unix holds 42 percent of the market, Windows 27 percent, and Linux just 5 percent.
Server Prices Even after a recent price cut, Sun servers cost more than comparable products from competitors Dell and IBM.
Sun Fire V480 Four 900-MHz UltraSparc III Cu processors,16 Gbytes RAM, Solaris 8: $46,995
IBM eServer pSeries 630 Model 6C4 2 x 2-way 1.2-GHz Power4+ processor, 8 Gbytes RAM, SuSE Linux Enterprise Server 8: $35,944
Dell PowerEdge 6650 Four 1.5-GHz Intel Xeon processors, 16 Gbytes RAM, Red Hat Linux 8 Professional: $24,421
