Gordon Moore Talks Past, Future
At an Intel Developer Forum keynote event, Gordon Moore, co-founder of Intel in 1968 and now Chairman Emeritus, remembered the beginning of the semiconductor industry and talked about its future.
Alexander E. Braun, Senior Editor -- Semiconductor International, 9/19/2007 7:13:00 AM
At an Intel Developer Forum keynote event, Gordon Moore, co-founder of Intel in 1968 and now Chairman Emeritus, remembered the beginning of the semiconductor industry and talked about its future.
Moore was greeted by a standing ovation and interviewed on stage by Moira Gunn, hostess of the National Public Radio programs, Tech Nation and Bio Tech Nations. Moore, who has a B.S. in chemistry and a Ph.D. in chemistry and physics, began work in semiconductors when he was recruited by William Shockley for the Shockley Semiconductor Laboratory division of Beckman Instruments, and left with Bob Noyce and others to found the Fairchild Semiconductor Corp. (South Portland, Me.).
In 1965, he wrote the seminal article on what would become known as Moore’s Law. “Yield was a problem back then,” he recalled. “We had a transistor yield of maybe 20% on a wafer, and if you put eight transistors on a circuit, people thought you took 0.28 and that would be the yield you would get, which would not make them very inexpensive. The idea that you got IC yields comparable to what you had with individual transistors, was hard to believe, plus they did not think them reliable, because they couldn’t measure the parameters of individual transistors and resistors.” Selling ICs for less than what individual components cost changed the economics of the industry, leading to larger volumes and lower costs.
At Intel, Moore and his colleagues pursued three new technologies. “We had a new twist on bipolar transistors, but it was immediately copied by Texas Instruments and Fairchild. We tried another approach, which was multiple chip assembly — MOS memory chips and bipolar drivers — but it was too complex. We also tried the silicon gate MOS approach, which turned out to be right for us, and had a seven-year monopoly on it, which allowed us to become established.”
Asked if he foresees an end to the law that bears his name, Moore answered in the affirmative: “Any physical quantity that grows exponentially predicts disaster, an ending. You can’t go beyond certain limits. When asked what are microelectronics’ fundamental limitations, Stevens Hawkins answered ‘the speed of light and the atomic nature of matter.’” Moore noted that today we are not too far from this. “Before we had our hafnium breakthrough, we were down to five molecular layers of insulator. You cannot go below this.” He added that he is always amazed by how technologists have pushed these off, and of how, for as long as he remembers, fundamental limits have been two or three generations away. However, Moore believes that in about a decade, the industry will hit a wall.
In reply to questions posted for him on the Internet during the event, Moore said that the computer has enabled a rate of innovation for science similar to that experienced by the semiconductor industry, noting that all sciences now depend on computers for their progress.
Asked if anything amazed him about the industry over his long career, Moore said wafer size. “We’re at 300 mm. I never believed that they’d get that big.” He added that in the early 70s, he did an exponential extrapolation about wafer size, from the 75 mm ones used then to the year 2000. “I extrapolated a size of 58 inches, which goes to prove how foolish extrapolating exponentials can be,” he said.
Looking into the future of computing, Moore believes that the next revolution will come when good language recognition enables humans and computers to interact closer. According to him, this will happen when a computer can understand context. “Whether you mean ‘to,’’too’or ‘two.’ At that point, a whole new era will start.”
