Moore's Law

In 1965 Gordon Moore, then director of R&D at Fairchild Semiconductor, was asked to write a short piece for the thirty-fifth anniversary issue of Electronics magazine. He drew a line through five data points and projected it forward a decade.

What it actually said

The 1965 article, "Cramming more components onto integrated circuits," argued that the number of components per integrated circuit at minimum cost had been doubling roughly every year, and that by 1975 a chip might hold as many as 65,000. The argument was economic rather than physical — the cost per component falls as the count rises.

The prediction proved close to exact: a memory chip released in 1975 carried 65,536 components. That year Moore, by then at Intel, revised the rate to a doubling every two years.

Not a law

Moore never called it a law. The name was applied later by Caltech’s Carver Mead. What it describes is not a force of nature but a commitment: the industry read the projection as a target, poured sixty years of roadmaps into meeting it, and thereby made it true.

The half that was left behind

Five years before Moore wrote it, he had been in the audience at the 1960 ISSCC when Douglas Engelbart delivered Microelectronics and the Art of Similitude. John Markoff reported the connection in The New York Times in 2005.

Moore’s Law is the comfortable half of Engelbart’s claim: the numbers improve on schedule, so keep building. The uncomfortable half — that each new scale retires the old assumptions — was not the part the industry set its clock by.