For Best Results, Decentralize R&D

Intel Corp. is one of the world's most successful high-tech companies, but it doesn't have a centralized research and development laboratory. Why? Because Chairman Gordon E. Moore discovered long ago, at the birth of Silicon Valley, that centralized R&D labs just don't work very well. Good ideas are forever bottled up inside them--or leak to the outside world when scientists leave to start their own companies. So Intel does all its research and development in its operating divisions, ensuring close teamwork between the people who know the market and those who know the technology. The bottom line: Decentralizing R&D pays off.

Industrial giants around the world are catching on to the idea. Dutch electronics maker Philips once funded all its R&D centrally, which insulated its labs from the needs of the business units. Now, under a new American R&D chief, 70% of R&D funding comes from business units. At ABB Asea Brown Boveri, the Swedish-Swiss conglomerate, the figure is closer to 90%. AT&T Bell Laboratories, once an empire unto itself, has decentralized to the point where most of the people wearing Bell Labs badges draw their salaries from AT&T business units.

Central labs still have an important role. They're good for researching base technologies that can be used by many company operations. By pooling their resources, these units can fund long-term work that would be beyond the resources of any one of them alone. But beware: In the name of exploring base technologies, companies and R&D consortiums may simply water the soil for upstart rivals. That kind of altruism is more properly the role of government, which should pick up the slack as companies reduce their commitment to non-commercial research.

As for profit-generating R&D, don't bottle it up. Farsighted business managers know the technology they need better than an isolated lab, and they're not afraid to pay for it. Witness Intel's R&D budget this year: a hefty $900 million.