Moore's law

ops cultureprogramming

The observation that the number of transistors on a chip roughly doubles every couple of years.

Gordon Moore noted the trend in 1965, and it held for decades as a self-fulfilling roadmap the whole industry planned around. It is slowing as physics intrudes, but it remains the reference point for talking about the pace of computing progress.

Moore's law was an observation, not a law of physics: Gordon Moore noticed in 1965 that the number of transistors on an integrated circuit was doubling roughly every year, later revised to every two, and the industry then spent sixty years treating the observation as a schedule to be met. That inversion is the interesting part, because it became a self-fulfilling roadmap that coordinated an entire supply chain.

What it never said is that computers get twice as fast. Transistor density and performance decoupled around 2005, when clock speeds stopped rising because power density made further increases impractical, and the industry pivoted to more cores rather than faster ones. That is why single-threaded performance improvements have been modest for two decades while transistor counts kept climbing.

Its current state is genuinely contested rather than settled. Density gains continue but slower and far more expensively, node names stopped corresponding to physical dimensions, and the frontier has moved to packaging, specialization and stacking rather than shrinking. Whether that counts as the law continuing or ending is a definitional argument, which is usually a sign that the original framing has stopped being the useful question.

Also known as: Moore's law

← All glossary entries