Moore's Law
The industry's self-fulfilling prophecy that chips double their transistor count roughly every two years, the metronome that drove fifty years of exponential computing.
In 1965 Gordon Moore, then head of research at Fairchild Semiconductor and soon to co-found Intel, wrote a magazine article with a modest observation and an audacious extrapolation. The number of components crammed onto a single chip had been doubling roughly every year, and he guessed it would keep doing so. That guess, later relaxed to a doubling about every two years, became the metronome of the modern economy. It is called a law, but it is nothing of the sort.
Not a law of physics
Nothing in nature requires transistors to shrink on schedule. Moore's Law is better understood as a self-fulfilling prophecy: an entire industry agreed on how fast progress ought to go, then organized decades of investment to make it happen. Chip designers, equipment makers, and materials firms all planned to the same cadence, each betting that the others would deliver the piece they needed on time. Carver Mead, the Caltech physicist who coined the name around 1975, saw this early. The law worked because everyone believed it and acted accordingly.
The results were staggering. The first integrated circuits held a handful of transistors. A modern processor holds tens of billions, each far smaller than a virus. Because a smaller transistor is also faster and cheaper, every couple of years brought roughly double the computing power at similar cost. Put differently, computing became not just better but exponentially cheaper, which is why it seeped into cars, watches, and greeting cards. Compounded over fifty years, that is the gap between a room-sized machine and the phone in your pocket, which carries more computing power than all of NASA had for the Apollo landings.
Why it is slowing
The doubling cannot run forever, because transistors are approaching the size of individual atoms. Features on today's most advanced chips are measured in single-digit nanometers, close to the point where quantum effects make a switch leaky and unreliable. Each new node now demands exotic physics and staggering money: a single leading-edge fabrication plant costs tens of billions of dollars, and printing those features takes machines so complex that only one company on earth can build them.
That is precisely why chips became a geopolitical prize. When progress was cheap and widely shared, no one fought over it. Now that squeezing out the next node requires a handful of firms, a few countries, and tools only one company can make, the ability to build cutting-edge chips has turned into a lever of national power. Governments that once ignored semiconductors now treat them like oil.
Moore himself was clear-eyed about the ending. Exponentials, he liked to note, always stop. The remarkable thing is not that his law is finally bending after half a century, but that a single sentence in a 1965 trade magazine set the pace of the digital age for so long, and that a whole civilization chose to keep it.