Cramming More Components onto Integrated Circuits
One graph noticed that chips double in complexity on a clock — and set the pace of the digital age.
In 1965 a chemist drew a line through a few dots and accidentally wrote the timetable the whole computer age would run on.
The big idea
An integrated circuit packs many tiny components — transistors and the like — onto one sliver of silicon. Gordon Moore noticed that the number you could economically fit onto a chip had been doubling about once a year, and bet the trend would keep going.
Doubling sounds modest until you let it run. Double, double, double — after ten doublings you have a thousand times as many; after twenty, a million times. That compounding is why the computer in your pocket has more raw power than the room-sized machines of Moore's day, and why it cost so little.
How it came about
Moore was running research at Fairchild Semiconductor, one of the young companies inventing the silicon chip, when a trade magazine asked him to predict the field's next decade. He had only a few years of data, but plotted on the right kind of graph the points lined up beautifully, and he extrapolated boldly: about 65,000 components on a chip by 1975, when the best of 1965 held a few dozen.
He was close to right. Three years later he co-founded Intel and spent the next decades helping make his own prediction come true. The trend got his name — "Moore's Law," a phrase coined by his friend Carver Mead — and in 1975 Moore himself adjusted it, saying the doubling now took about two years rather than one.
Why it mattered
Moore's Law became less a forecast than a shared deadline. Chip makers, designers and equipment suppliers around the world treated the curve as a target and organized decades of research and billions in investment to keep up with it. That coordinated, relentless pace is what drove the cost of computing down and down — turning a luxury for governments and corporations into something cheap enough to put in phones, cars, watches and toys.
A way to picture it
Picture a single grain of rice on the first square of a chessboard, two on the next, four on the next, doubling all the way. By the last square the pile is astronomical — far more rice than exists. Moore spotted that chips were marching across that chessboard about one square a year. We are now deep into the squares where each step adds staggering numbers, which is exactly why progress in computing has felt explosive rather than steady.
Where it sits
This little article is the quiet engine under most of the Library's modern technology entries. The cheap, abundant computing it predicted is what made the internet (see Berners-Lee), and later the data-hungry neural networks of AlexNet and the Transformer, practical at all. Whether the curve can continue is now one of the central questions of the technology world — physics and economics are both pushing back — but the expectation of exponential progress it created has outlived its original numbers.
The complexity for minimum component costs has increased at a rate of roughly a factor of two per year.
Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years.
The future of integrated electronics is the future of electronics itself.
Integrated circuits will lead to such wonders as home computers — or at least terminals connected to a central computer — automatic controls for automobiles, and personal portable communications equipment.