Just as Sun Microsystems' sales slump and investors grow impatient with the company's turnaround plans, Andy Bechtolsheim announces he's leaving—again

As though the market forces battering Sun Microsystems weren't enough, the troubled computer company is absorbing another punch. Andy Bechtolsheim, a co-founder who returned to the company four years ago as a technical savior, is leaving again for the startup world.

Bechtolsheim, who designed Sun's (JAVA) original workstation in the early 1980s after founding the company with Chairman Scott McNealy and others, returned in 2004 after a nine-year hiatus and has worked on lower-priced computers, featuring chips from Advanced Micro Devices (AMD), that Sun hopes will revive growth.

On Oct. 23, Bechtolsheim is leaving Sun for the second time to become chairman and chief development officer at Arista Networks, a startup he has funded that sells a powerful network switch for data centers, and whose customer list includes Google (GOOG). Arista's CEO is Jayshree Ullal, who left Cisco Systems (CSCO) in May after 15 years at the company, and who had worked on some of the company's key products (BusinessWeek.com, 1/28/08).

The departure of Sun's star engineer, who had spearheaded development of a new line of more affordable products the company is pushing, comes as Sun's sales slump and investors grow impatient with the company's turnaround plans. On Oct. 20, Sun said it expects a steep first-quarter loss, sending its shares into a further tailspin. Shares of Sun closed on Oct. 22 down 5¢, or 1%, at 4.72. The shares have lost 18% in the past two days and are down 74% this year. "They just keep taking body blows," says Gordon Haff, an analyst at market researcher Illuminata. The company formally issues quarterly results on Oct. 30.

A Startup with Promise

Arista, a 50-employee company started several years ago with $50 million in funding by Bechtolsheim and chief scientist David Cheriton, sells a line of switches that can transfer data at extremely high speeds from servers that run Web sites and technical computing centers, at a fraction of the cost of comparable products from Cisco and other vendors, according to its executives. Other customers include Lawrence Livermore National Laboratory and digital media streaming company BitGravity. The market for its main product—10-gigabit Ethernet switches—could grow to $5 billion to $10 billion within five years, from about $500 million today, according to Ullal.

As companies serve massive amounts of data through the Internet, the switches that manage the flow of digital traffic have become a bottleneck. "Why should the network cost as much as the server?" says Bechtolsheim, in an interview with Ullal in a bare-bones office in Silicon Valley. "It makes no sense."

By contrast, Sun's server business is under pressure. Margins on the large computers that run corporate data networks are being squeezed amid widespread adoption of standard processors from Intel (INTC) and AMD, and the popularity of the open-source Linux operating system. "The server industry for the most part now is reselling Intel and AMD CPUs," Bechtolsheim says. "You wouldn't start another server company today."

For Sun, the bad news is piling up as CEO Jonathan Schwartz executes an uneven turnaround. Sun said it would report a loss of 2¢ to 12¢ in its first quarter, compared with Wall Street analysts' expectations of a penny-per-share loss. Sun also said revenues would decline to between $2.95 billion and $3.05 billion, short of analysts' expectations.

Open-Source Strategy

The company has suffered from falling sales (BusinessWeek.com, 5/2/08) of computers based on its proprietary Sparc chip and has struggled to sell enough low-end machines based on AMD chips to make up the difference. Schwartz's strategy of releasing much of Sun's software under open-source licenses has also failed to produce enough excitement in the market.

On the financial side, a reverse stock split approved in November hasn't bolstered Sun's share price; the company now trades at a market value of just $3.5 billion. On Oct. 22, Fitch Ratings downgraded its rating on $550 million of Sun's debt due to the company's "deteriorating financial performance."

"They've been in a state of restructuring for years," says Jayson Noland, a senior analyst at Robert W. Baird.

Now some analysts are saying that a sale or replacement of Schwartz, who took over as CEO from McNealy two years ago, could be in the offing. "The probability of a transformational event—divestiture, sale, or management change—is rising," said Citigroup (C) analyst Richard Gardner, who has a buy rating on Sun, in an Oct. 20 note to clients.

Takeover Time?

In August several large investment banks were pitching a Sun takeover to private equity firms, according to bankers and analysts. "It still does pump out a lot of free cash flow," says one banker, who asked not to be named. But, he added, a technology company that could acquire Sun's patents and accounts could be a more likely buyer if Sun gets sold. "I don't think a private equity buyer is going to back Schwartz" and his open-source strategy, he says. A Sun insider says the company is not thinking of a management change and has no plans to sell the company or hive off parts.

For Bechtolsheim, who funded Arista from his personal fortune, the job change marks a return to the startup world at a time when venture capital is flowing to alternative energy and the Web, rather than information technology. "There hasn't been a real focused, reliable startup in this space in 10 years," says Ullal, who worked with Bechtolsheim at Cisco after the networking giant bought one of his previous startups, Granite Systems, in 1996. Bechtolsheim returned to Sun when it bought another company he formed, called Kealia. Bechtolsheim will remain with Sun in a limited, part-time capacity, the company says.

Now Sun will soldier on largely without the help of the designer of key products it's counting on to replace its out-of-favor Sparc computers, at a time when it's trying to reassure customers and investors of its relevance. Says analyst Haff: "They don't need another piece of negative news."