Apr 30 2008
Ever since the first transistors were carved into a wafer of silicon, chip makers have striven to cram more of the tiny switches into a given area and to get them to work faster. That obsession with performance is now giving way to a new goal--efficient manufacturing--and the result will be a fundamental restructuring of the entire electronics industry.
Four consultants, three engineers, and a professor of management, having worked with the staff of a chip fabrication plant, or fab, belonging to a major semiconductor company, have proved that semiconductor manufacturing can be made much more cost-efficient. In an article in the May issue of IEEE Spectrum magazine, they write that in just seven months the fab was able to reduce the manufacturing cost per wafer by 12 percent and the cycle time--the time it takes to turn a blank silicon wafer into a finished wafer--by 67 percent. All these gains were attained without spending a single extra dollar on changing the design, adding new equipment to the fab, or developing new technology. The authors are Clayton M. Christensen, a professor of management at the Harvard Business School and author of the bestselling book The Innovator's Dilemma; Steven King and Matt Verlinden, engineers turned management consultants; and Woodward Yang, a professor of electrical engineering at the Harvard School of Engineering.
They accomplished this feat by applying methods developed by the Toyota Corporation, long famous for its unrivaled manufacturing efficiency. All work processes should be measured, no extra inventory should be allowed to pile up between the various steps in the manufacturing process, any problems that develop should be noted and solved on the spot by the workers themselves, using scientific standards of inquiry, and the solutions should be used to develop more efficient ways of organizing work. But it's easier said than done: thousands of executives and economists have trooped through Toyota's plants in order to learn from its success, but few have been able to replicate it in their own organizations.
The authors note that the methods work best when applied not to fabs making cutting-edge products, with the smallest transistors and the fastest computation speeds, but rather to those using slightly outdated technology. Yet this will pose no real problem, they argue, since today's fastest chips offer far more power than most customers can effectively use. In a world in which chips are faster than needed, consumers will tend to emphasize other features that help them in their businesses.
One such feature that will supplant sheer performance is the speed to market--the time it takes to get a chip from the drawing board into customers' hands. Many modern electronic devices, such as the fashionable new cellphones and MP3 players, now have design cycles measured in months. Today's hottest product is tomorrow's also-ran. In such an environment, what matters is getting a chip designed and out the door fast, before customer preferences have a chance to change.
The effect of quick, continuing improvement in production efficiency on the auto industry has been obvious. Companies like Toyota that learned to wring more production out of their plants have prospered, while those that did not have gone to the wall. The same may well happen in the semiconductor industry.
Posted April 30th,2008