The prediction was modified to two years or so in 1975. Today, Gordon Moore's sketch and his article explaining his thesis, published in the April 19, 1965, issue of Electronics magazine, embodies a concept that VLSI guru Carver Mead later coined as "Moore's Law."

The topic of progressive silicon scaling originally came up during one of Mead's and Moore's regular after-hours chats at Fairchild, where Mead was consulting and Moore was director of R&D labs. Moore later became a co-founder of Intel Corp. "I was working on the problem of electron tunneling [which limited how small transistors could get] and Gordon asked me what the limits were," said Mead, co-author of a seminal textbook on VLSI circuit design and now the Gordon and Betty Moore professor emeritus at the California Institute of Technology. At the time of his discussion with Moore, circuit line widths under 10 microns were deemed impossible, Mead said. But he looked at the problem and figured out how to scale down to 0.5 micron. "People said we were crazy," Mead said. But Moore believed him.

While Mead was traveling and meeting with members of the scientific community in 1971 to convince them of his findings, Moore would constantly supply him with charts that graphically showed the potential for scaling to help Mead's proselytizing. It was while chatting with a journalist that Mead coined the term Moore's Law. When the journalist subsequently used the term in an article, said Mead, everything changed.

"It was a catchy phrase," said Mead. The fortuitous combination of Mead's phrase, his research and his seminal 1971 paper on electron tunneling made everything that followed possible, said Mead. It was less about technology than about unlocking human potential, he said. "It made people believe in the future, in possibilities; it was all about the human spirit."

Craig Barrett of Intel, then a materials science professor at Stanford University, said, "Back then it was taken more as an observation than a road map for the industry."

The thinking at the time was that "this can't go on forever, but it was going on and people were considering it a standard for what you had to do to be competitive," said Barrett, who joined Intel in 1974 and became its CEO in 1998.

"There have been an endless series of engineering challenges, and that is what has attracted smart engineers who don't understand that they are not supposed to be able to solve these problems," said Barrett.

In his 1965 article in Electronics, Moore suggested that the integration trend "will lead to such wonders as home computers — or at least terminals connected to a central computer." The editors of Electronics found that so amusing they commissioned a cartoon spoofing the concept of a personal computer for consumers.

Yet Moore's observation is as real today as ever, according to Barrett, who sees it holding true for the next 10 to 15 years.

But there will have to be a lot of cross-disciplinary work for that to happen, said Mead, who attributed progress to date to chemical and manufacturing processes, not physics. "We've gotten kind of lazy in letting chemistry do all the work," he said. "Now we have to start thinking and it's going to take longer [to double the transistor count] and it will be harder."

Further scaling will take an approach that combines chemistry, physics and biology, Mead said.

"We're still using the same old creaky stuff [architectures], but biology shows us that there are very different ways to do computing," he said.