Rumors were flying around Silicon Valley in November 2005 that something intriguing was happening in one of the underground parking garages at
But when a cutting-edge data center in a shipping container was announced in October 2006, the company behind it wasn't Google. It was
Sun was the Google of an earlier era. It scored one of Silicon Valley's great successes back in the 1980s and 1990s by engineering high-performance computers for the most technologically sophisticated clients, from top scientists and engineers in academia to quants on Wall Street and leading automakers—not to mention telecom highfliers and Internet startups. Then came the 2001 dotcom crash. Many of Sun's customers were crushed, and others switched to cheaper equipment powered by improved chips from AMD and Intel. Many who'd been paying up for Sun's Solaris software flocked to Linux for free. Sun's stock plunged from $60 a share to less than $3, and the company lost money nearly every quarter for a half-decade. Outside critics and even former executives predicted that it wouldn't survive.
Now, suddenly, the outlook has brightened.
Orchestrating this new dawn is a ponytailed 41-year-old freethinking CEO who uses his own widely read public blog as a management and marketing tool, who talks about working with "co-opitors" instead of railing at competitors, and who has persuaded 18,000 of Sun's 34,000 employees to work from home—not just as a money-saving tactic (Sun has been able to close two Silicon Valley campuses, trimming $70 million in expenses last year), but also to slim the company's environmental footprint. Jonathan Schwartz, who took over as chief executive from the combative and colorful Scott McNealy a year ago, even shares his personal office with Sun's chief financial officer, Mike Lehman. (Colleagues describe the roommates as the "odd couple," with Schwartz cast as Oscar.)
Schwartz is not a newcomer; he has been at Sun since the company bought his software startup in 1996. But he has brought a new style since replacing McNealy in April 2006. "McNealy had speechwriters to help him bash everyone," says Laura McLellan, an analyst at the Gartner Group. "The first week that Schwartz was CEO, he got on the phone and talked to the CEOs of
Project Blackbox may be Schwartz's biggest initiative to date, and the one that best reflects his vision for the new Sun. The idea of a data center in a shipping container did not originate with Sun or with Google—the concept is usually attributed to Brewster Kahle, a Thinking Machines veteran and founder of the Internet Archive. But it is Schwartz who nurtured and introduced the Blackbox in a bid to give Sun a new signature product and marketing vehicle. As part of his push for environmentalism, he hired back Sun alumnus David Douglas as vice president of "eco-responsibility" and put him in charge of developing the Blackbox. One goal of the project: to produce more computing power using less energy. In today's big data centers, it costs almost as much to power up the computers and cool the buildings from the resulting heat as it does to purchase the equipment. The Blackbox promises 20% greater energy efficiency than a typical corporate data center, due in part to its innovative cooling system.
This spring, Douglas has been shepherding the Blackbox on a cross-country tour to show off what's inside: a plug-and-play data center that a customer can hook up to a power line, a data line, and a cold-water line for cooling, and have running in minutes after it comes off the truck. The 250 servers are mounted on racks atop big spring-coil shock absorbers (in case the whole container is dropped by a forklift operator, which has happened), the advanced cooling system is fully engineered, the software is loaded and ready. If equipped with $3,000 servers from the lower end of Sun's Niagara line, Douglas says, a Blackbox would rank as the 170th-fastest supercomputer in the world—for a price in the low $1 million range, a fraction of what building a data center from scratch would cost.
"You can't find out if you're right until you take the risk," says Papadopoulos, Sun's chief technology officer.
Sun is closemouthed about prelaunch buyers. Gartner's McLellan says that she and her colleagues are "almost entirely convinced" that Sun has been building Blackboxes for Google. (Google, notoriously secretive about its technology, won't comment.) Silicon Valley is buzzing that Microsoft may be a customer—or be building something similar. James Hamilton, a technologist on Microsoft's Windows Live team, said this spring that portable data centers in shipping containers were "an idea whose time has come" and could create "a competitive advantage."
“For a month, I took a lot of drugs to sleep," says Greg Papadopoulos. Sun's chief technology officer is recalling the days after he convinced his colleagues to scrap a half-billion-dollar investment Sun had made on a new silicon chip. At a 2002 strategy meeting in McNealy's office, as obits were being written for the dotcom era and the company's stock price continued to tumble, Papadopoulos argued that the hundreds of millions of dollars spent developing the chip should be chalked up as yesterday's mistake and instead Sun should pony up new money for a radical new chip design. His idea was to divide a chip into eight independently operating sections, called "cores," each of which could handle four separate computational threads at a time. It would be like having 32 different brains working at once on the same piece of silicon, and, he contended, it could be much faster and more energy efficient. It was a gamble. "But you can't find out if you're right until you take the risk," Papadopoulos says. He got the green light (and the sleepless nights), and over the next three years, Sun invested millions in his bold idea. Even as the company bled money, management poured about 15% of revenues into R&D, roughly the same percentage as Microsoft and Intel.
The chip that ultimately emerged has helped Sun win back its reputation as a leader in technology. It's at the heart of the company's Niagara servers, introduced in December 2005, which power the Blackbox. Aside from speed and bragging rights—no one else had yet marketed a chip with four cores, let alone eight—the Niagaras also offered energy savings. A Niagara chip runs on only 70 watts, less than many common household lightbulbs, and one-third as much as a typical industry chip. California's
Schwartz is trying to reinvent Sun as a new kind of computer company. When he took over, he ignored Wall Street analysts' calls for massive layoffs. Sun's resurgence has to come from creativity, he insists, not just cost cutting. "We're not in business to save money," he says. "We're in business to ship innovations."
Whether Project Blackbox will be more than a niche product is an open question. Jim Burton, an analyst at Ideas International, an information-technology research firm, says that "Sun has carefully studied the problems that keep IT management up at night, then created a product to address those problems. The Blackbox is a simple and elegant solution being introduced at exactly the right time." Others are more skeptical. David Mathog, who runs a computing facility at Caltech and is precisely the sort of person Sun hopes to target, raised some practical issues on an IT discussion group: Would it meet building codes? Could staff move around inside?
"There is enough room for people to comfortably work," responds Sun spokesman Shawn Dainas, adding, "we've had an overwhelmingly positive response from customers who have gone to see the Blackbox in the U.S. tour."
The Blackbox, at a minimum, is a very smart way of capturing people's attention and imagination—and it might even help sell more servers. "While the idea behind Project Blackbox is ingenious," blogged software developer Scott Yang, "it is no more than a strategy for Sun to sell more boxes." The Sun Web site claims that 250 Niagara servers in a Blackbox will support four times as many users at five times the energy efficiency of similarly configured servers from, say, Dell.
"The single highest-impact blog I wrote in the past year was when I apologized to a customer who had a hard time trying to buy from Sun," says CEO Schwartz.
Schwartz rejects the conventional wisdom that computers have become commodities, a necessary cost rather than a potent weapon. Instead, he proposes that more and more companies will be like Google, which has an insatiable appetite for computing and uses it for strategic advantage. And the Blackbox is symbolic of that vision. "This is not just a new kind of package," says legendary computer designer Danny Hillis of Applied Minds, who consulted on the project. "It's a new idea of what a computer company's product should be."
Wall Street's confidence in Schwartz's approach wavered momentarily in early April after Bernstein Research analyst Tony Sacconaghi downgraded the stock, noting that its 50% pop had made it expensive compared to rivals IBM, HP, and Dell, all of which have lower price-sales ratios. But the dip was brief, and both Merrill Lynch analyst Richard Farmer and
Of course, the Street is notoriously fickle and will swing back and forth based on rumor and sentiment—until the numbers come in. And that's where Sun and Schwartz and the Blackbox have to show their muscle. Schwartz is doing what he can: He's wooing rapidly growing startups like the hot social-networking site Twitter with discounts of up to 70%. He'll happily sell a Blackbox loaded with rivals' servers, if that's what a customer wants. And he's mending fences. "The single highest-impact blog I wrote in the past year was when I apologized to a customer who had a hard time trying to buy from Sun," he says. "We need to authentically engage with the community if we want them to trust us, or they won't buy from us."