Moore's Law Was a Sales Pitch That Came True

On April 19, 1965, Gordon Moore — then director of R&D at Fairchild Semiconductor — published an article in Electronics magazine titled "Cramming more components onto integrated circuits." In it, he plotted five data points and noticed that the number of transistors on a chip had been roughly doubling every year. He extrapolated forward to 1975 and predicted the trend would hold.

It did. Moore later revised the doubling period to two years, and the industry kept up for nearly five decades. A 1971 Intel 4004 had about 2,300 transistors. A 2024 Apple M4 chip has roughly 28 billion. The compound effect is why a phone in your pocket has more compute than a 1990s supercomputer.

The trick is that this wasn't a law in the way Newton wrote laws. There was no underlying physical principle requiring chips to double. Moore was describing an economic and engineering trajectory, and the trajectory held because the entire industry — fabs, designers, lithography vendors, customers — coordinated their five-year roadmaps around it. Hitting Moore's pace was the price of admission. Missing it meant being out-competed.

The slowdown started around 2005, when single-core clock speeds plateaued near 3 GHz. Heat became the wall. The industry shifted to multi-core designs, then to specialized chips for AI workloads. Transistor density still grows, but slower, and the cost per transistor has stopped falling.

The law was a self-fulfilling prophecy that ran for half a century. Few prophecies do that well.