A Fallen Tech Highflier Sifts Through Bubble Memories
Whether it’s because the wounds are not yet closed, the resentments not faded or the period not yet sufficiently distant in time, a truly valuable insider’s history of the Internet bubble has yet to be written.
“It’s still a little early to have perspective on something that happened in the last 10 years,” Eric Brewer was telling a couple of hundred connoisseurs of high-tech wreckage one evening not long ago.
“This is a first attempt to understand what the Internet bubble meant. Here’s what it looked like,” he added, projecting on an overhead screen the familiar Matterhorn-like outline of an Internet stock price trajectory. In this case, the path was that of Inktomi Corp., the search engine and network technology company he co-founded as a Berkeley computer science professor in 1996 and helped run until its 2003 acquisition by Yahoo Inc.
In his talk at the Computer History Museum in Mountain View, Calif., and in a later interview, Brewer -- at 37 an ex- billionaire (on paper) -- covered a few aspects of the technology crash that have all but dropped out of public memory.
One is that the crash encompassed not only dot-coms but also telecommunications companies. The crisis in the latter sector, in fact, is what brought down Inktomi, which sold network services to companies like WorldCom Inc. without realizing that they had borrowed so much money they were nearly insolvent. “The question we never asked was: Do you trust the money from your customers?” Brewer says.
Another is how the run-up in stock prices distorted the business judgment of even sensible managements. As Inktomi’s market value soared toward its peak of $25 billion in 1998, it sought desperately to make itself worthy of the number.
“We knew what revenues we needed to justify the market cap long term,” Brewer says. That led to wasteful investments in an online shopping service and wireless networking, “because the businesses we understood couldn’t fill that gap.”
Inktomi was never a household name -- by design. Unlike Yahoo and Google Inc., which aimed to build consumer loyalty by flogging their brands, Inktomi aimed to sell search capability to other websites -- to be “the arms dealer to the Internet,” as Brewer says. By forswearing the chance to compete with its customers, the company figured it could sell to everyone.
In part, this strategy reflected Inktomi’s basic technology. The company grew not from a clever algorithm for ranking search results but from Brewer’s novel ideas about supercomputing, his PhD work. The traditional supercomputer was a single huge, muscular machine. Brewer thought of breaking down the same package of computing power into a network of smaller computer workstations, all working in parallel.
As the expanding World Wide Web began to require supercomputer-scale resources, Brewer and his Berkeley team saw their architecture as a perfect fit. It could be easily scaled up -- instead of adding capacity a supercomputer at a time, networkers could add workstations to their computing hives as needed. If any component failed, the other machines would simply route data around it -- mirroring the Internet’s ability to bypass its own traffic jams.
Brewer recognized that the heaviest demand for data servers would come from search engines. After building a prototype with 1.2 million Web pages, surpassing the then-leading search portals Yahoo, Lycos and Infoseek in size and speed, Brewer and his graduate student Paul Gauthier founded Inktomi in February 1996.
The arms dealer strategy worked for a while. Yahoo, Microsoft Corp.’s MSN and America Online hosted huge consumer portals, but their search engines were rebranded versions purchased from Inktomi. This role fueled Inktomi’s successful IPO in 1998. The company took on the cultural aura of dot-com highfliers. Its devoted workers let off steam with indoor water-gun fights. Employment rose from 200 to 1,100. The stock split twice.
But cracks soon appeared in its strategy. Its clients were exceedingly fickle about the need for search engines. Sometimes they regarded search as central to attracting users, sometimes only peripheral. Generally they were loath to pay much for it.
Inktomi began to rely more on selling network services, such as its “Traffic Server” product, which allowed Internet service providers and portals to cache popular Web pages off the main network, so they could be fed to users fast and without clogging the Internet backbone. When the 1998 Starr Report generated online demand of 300,000 copies a minute without crashing the Net, the principle was proved.
By 2000, Inktomi was earning 60% of its revenue from caching. A year later, thanks to the telecom crash, that revenue dropped by half and never recovered. The first layoff of 250 employees came in April 2001; two further rounds followed that year, and more the next. The Inktomi acquired by Yahoo employed 140, down nearly 90% from its peak.
But by then the online world was back in the thrall of search, driven by the epiphany that advertisers would pay cash for prominent placement on a page of search results. (This is the source of most of Google’s income.) Inktomi had developed a targeted ad server as early as 1996. But it couldn’t sell the system to clients, and lacking its own portal it couldn’t exploit the ad market itself.
Brewer returned to the Berkeley faculty after the Yahoo merger. Although securities law prohibited him from selling Inktomi stock on the way up -- or down -- he acknowledges salvaging $10 million to $100 million. His momentary status as a paper billionaire did bequeath him, however, a taste for seeking big solutions for big problems. (“You start to think of what you can fix with a billion dollars.”) His current interest is in developing technology for the Third World, such as economical communications kiosks to give rural residents access to job banks, government services and agricultural market prices.
There’s also the gratification of having developed a technology, and not a frivolous one, that lives on. “It’s a very nice feeling,” Brewer says, “to have had an effect on everyone.”
Golden State appears every Monday and Thursday. You
can reach Michael Hiltzik at golden.state@latimes.com.
