David Culler wants to teach Intel Corp some new tricks.
The University of California computer-science professor runs the biggest chipmaker's newest research lab, in the penthouse of a Berkeley bank building overlooking the San Francisco Bay.
Culler and some new Intel scientists are challenging a tradition of secrecy at the company and throughout Silicon Valley. That secrecy often prevents the give-and-take collaboration that might speed development of products that could change how doctors diagnose patients, how firefighters track forest blazes and how consumers handle household tasks.
"The more open it is, the more effective it turns out," Culler said, pointing to advances such as the Internet and Unix programming language, which didn't start in strict corporate labs but have remade how business is done.
Meshing the team-oriented college atmosphere with Intel's insular culture has been tricky. Culler is mocking up an "openness agreement," designed to combat the nondisclosure agreements that companies require to limit discussion of proprietary work.
Culler said he wants people to watch what they say about secret projects, and he's set up enclosed work areas, dubbed "blue offices," for times when confidentiality is necessary. The rest of the lab is mostly open space, with couches set along the windows offering scenic views, to inspire collaboration.
The approach is a departure for Intel, and one the chipmaker needs to master. Intel and rivals guard patents and ideas closely, frequently suing to protect licensing fees and profit. Yet, if Intel wants the first look at work that may yield cutting-edge products in five or 10 years, it must open its research to outsiders and bet that it can bring the products to market faster than rivals.
"The breakthrough isn't something you can contain and hold on to," he said. "You compete on how well you understand it."
Intel has similar labs at Carnegie Mellon University in Pittsburgh and the University of Washington in Seattle. David Tennenhouse, who runs the company's research group, wants ultimately to have six or eight centers, including two outside the US. The Berkeley lab held its formal opening yesterday.
Culler and his counterparts at the other labs are on leave from their schools. After returning to full-time college work when their one- or two-year terms end, they'll spend one day a week as Intel consultants.
Their scientists will develop broad ideas that may or may not eventually lead to Intel products. Tennenhouse foresees gizmos such as tiny devices that can be thrown from helicopters into forest fires to detect hot spots, software to manage storage distributed among many sites and ways to change how people and PCs interact. It will take US$4 million to US$6 million a year to run each center when fully built out, he said.
Historically, Intel's relationships with universities have come through grants and fellowships given to professors and students. The projects Tennenhouse touts started in schools and have grown to the cusp of some breakthrough that may come faster with money and brainpower from Intel.
"This gives Intel a deeper mechanism to transfer the learning from some leading university researchers into their company," said Henry Chesbrough, an assistant professor at the Harvard Business School who studies new R&D styles.
RSS