Gunning for an elephant in Silicon Valley

If you don't envy Andreas Bechtolsheim, you may not know enough about him. Even by the standards of Silicon Valley hotshots, his combination of talent and luck is rare. In 1982 the brainy young engineer co-founded Sun Microsystems and helped usher in a new era of computing by dreaming up a workstation that was cheaper and more versatile than the minicomputers popular with companies at the time. Then in 1998, at a friend's house in Palo Alto, Bechtolsheim had the good fortune to meet two young Stanford University grad students working on a new search engine. He wrote the pair a check for $100,000 on the spot, and Larry Page and Sergey Brin proceeded to turn his initial stake into more than $1 billion in Google stock.
Now the 54-year-old has an audacious new goal: taking on Cisco Systems. Although Cisco is the dominant provider of the networking equipment at the heart of the Internet, Bechtolsheim thinks his tiny Arista Networks has an edge in the fastest-growing part of the market: blazing-fast gear for the data centers where more and more computing is done. If he's right, Arista could undercut Cisco in a critical business, complicating the giant's efforts to hit CEO John Chambers' aggressive growth goals. "We're just a fly in the ointment to Cisco today," says Bechtolsheim. "But it's very hard for large companies to change their fundamental business models."
He claims an edge in cost and performance. Arista's switches, computers that direct traffic around the Net, are less than half the price of comparable Cisco products. Lane Patterson, chief technology officer of data center company Equinix, says Arista's equipment can be one fifth the cost, "a phenomenally compelling price." Its gear also runs on cutting-edge software that analysts say has advantages over Cisco's 20-year-old software, IOS, which is as prevalent in networking as Microsoft's Windows is in personal computing. Arista's software has the potential to run more kinds of programs and won't crash if one gets a bug.
Cisco isn't showing any sign of vulnerability. Its hold on the networking market is so complete it's branching out into dozens of new markets, from video to wireless technology. Competition is increasingly from tech giants, particularly Hewlett-Packard. Cisco declined to comment on Arista but points out that Cisco has 1,000 customers for its data-center products, while Arista claims 130.
Still, Arista is getting traction. The company has a broad clientele only two years after introducing its first product. A third of its customers are big Wall Street firms looking for faster computing systems and speedier execution of trades. Bechtolsheim also recruited a heavyweight to lead the company, Jayshree Ullal, former architect of Cisco's own data-center effort. "[Arista] is doing very well," says Pacific Crest Securities analyst Brent Bracelin. "Clearly there's a strong appetite for alternatives to Cisco."


Selling his startups
Bechtolsheim has long been ahead of his time. He grew up on a farm in Germany and earned $100,000 at age 16 by writing software for a local tech firm. He came to the U.S. on a Fulbright Scholarship but dropped out of a doctoral program at Stanford University to help found Sun. After leaving Sun in 1995, he kept spotting opportunities to do what tech leaders were doing, only cheaper or better. He sold one startup to Cisco for $220 million and another to Sun for $91 million. "Andy is to corporate computing what Steve Jobs is to consumer computing," says Sun co-founder and Chairman Scott G. McNealy.
Bechtolsheim sees opportunity in networking in part because of Cisco's dominance. The company specializes in reliable gear and strong customer support, charging premium prices that help garner 65% margins. That's a winning formula with Cisco's bread-and-butter customers—conservative corporate buyers most concerned with keeping their networks up and running. But for companies building big data centers—Wall Street firms and Net giants such as Google—slashing a few milliseconds off a stock trade or search query is more valuable than kid-glove support. "We are totally convinced we have a home run on our hands," says Bechtolsheim.
That is, unless he decides to sell out again. Already, there's talk IBM or HP might try to buy Arista. Bechtolsheim insists that's not his plan. Arista is much further along than his previous startups, and he's cranking up his sales effort for a possible initial public offering. While he admires Cisco, Bechtolsheim says tech industry history is full of fallen giants. "It's strange how none of the big companies were the winners in the next era," he says.

Burrows is a senior writer for BusinessWeek, based in Silicon Valley.