Moore's law
In plain wordsGordon Moore's prediction that chipmakers could pack twice as many transistors onto a chip roughly every two years — a pace that for decades made computers steadily faster and cheaper, and has slowed but not stopped.
The long-running doubling, about every two years, of how many transistors can be put on an integrated circuit at the lowest cost per component. Originally a business forecast that the semiconductor industry adopted as a planning target, it held approximately from the 1970s into the 2010s, driven by lithographic shrinking (process nodesprocess node, n.: The label for a generation of chip-making technology, like '3 nanometer'. Smaller numbers mean newer, denser chips, but the number no longer measures any actual part of the chip.) and, until about 2005, accompanied by Dennard scalingDennard scaling, n.: The rule that as transistors shrank they also used proportionally less power, so chips got faster without getting hotter. It stopped working in the mid-2000s, ending the era of ever-faster single processors., which made each transistor faster and more energy-efficient as well. Since then cost per transistor has fallen more slowly, and leading-edge gains come increasingly from advanced packagingadvanced packaging, n.: Ways of joining several chips very closely inside one package — side by side on a silicon base or stacked on top of each other — so they act almost like one bigger chip., chipletschiplet, n.: One of several smaller chips that are made separately and then joined inside a single package to work as one processor, which is cheaper than making one giant chip and lets makers reuse parts., specialized acceleratorsAI accelerator, n.: Any chip built or tuned to speed up AI — graphics chips, Google's TPUs, phone 'neural engines' and many startup designs. Such chips do the heavy math of training and running models. and lower-precision arithmetic rather than from shrinking alone. AI training computetraining compute, n.: The total number of arithmetic operations performed to train a model. It is the standard yardstick for the size of an AI training effort, and some laws use it to decide which models face extra rules. has grown much faster than Moore's law — on the order of 4–5× per year for frontier runs in the 2010s and 2020s — through larger budgets and more chips, not just better chips.
The cluster got bigger, not just the chips: our compute grew far faster than Moore's law.
Origin
Gordon Moore, then at Fairchild Semiconductor, observed in 'Cramming more components onto integrated circuits' (Electronics, 19 April 1965) that component counts were doubling yearly; in 1975 he revised the pace to about every two years. Carver Mead popularized the name 'Moore's law' in the 1970s. The often-quoted 18-month figure comes from Intel's David House.
Usage note
Whether Moore's law is 'dead' is disputed and partly definitional: Nvidia's Jensen Huang declared it so in 2022, while Intel's leaders said it continued at a slower pace. The name is also used loosely for any exponential improvement in computing.
Related
- Dennard scalingDennard scaling, n.: The rule that as transistors shrank they also used proportionally less power, so chips got faster without getting hotter. It stopped working in the mid-2000s, ending the era of ever-faster single processors.
- process nodeprocess node, n.: The label for a generation of chip-making technology, like '3 nanometer'. Smaller numbers mean newer, denser chips, but the number no longer measures any actual part of the chip.
- Huang's lawHuang's law, n.: A claim, named after Nvidia's chief executive, that AI chip performance improves much faster than Moore's law because chips, software and number formats all improve together.
- law of accelerating returnslaw of accelerating returns, n.: Ray Kurzweil's claim that technology improves at an ever-faster rate, because each advance helps build the next, so progress speeds up exponentially rather than steadily.
- chipletchiplet, n.: One of several smaller chips that are made separately and then joined inside a single package to work as one processor, which is cheaper than making one giant chip and lets makers reuse parts.
- training computetraining compute, n.: The total number of arithmetic operations performed to train a model. It is the standard yardstick for the size of an AI training effort, and some laws use it to decide which models face extra rules.
- semiconductor foundrysemiconductor foundry, n.: A factory company that manufactures chips designed by other companies. Taiwan's TSMC, the largest, makes nearly all of the world's most advanced AI chips.
Sources
- Moore, G. E. (1965). Cramming more components onto integrated circuits. Electronics 38(8), 19 April 1965.
Mentioned in
computecompute, n.: The raw computing power used to train and run AI, usually measured by how many specialized chips are used and for how long. With data and algorithms, it is a main driver of AI progress., algorithmic progressalgorithmic progress, n.: Improvements in AI methods — not faster chips — that let a model reach the same ability with less computing power. Researchers estimate this has cut the computing needed by half every eight to sixteen months., ASMLASML, prop. n.: A Dutch maker of the giant machines that print circuit patterns onto chips. It alone can build the extreme-ultraviolet models on which today's most advanced chips, including AI chips, depend., process nodeprocess node, n.: The label for a generation of chip-making technology, like '3 nanometer'. Smaller numbers mean newer, denser chips, but the number no longer measures any actual part of the chip., advanced packagingadvanced packaging, n.: Ways of joining several chips very closely inside one package — side by side on a silicon base or stacked on top of each other — so they act almost like one bigger chip., chipletchiplet, n.: One of several smaller chips that are made separately and then joined inside a single package to work as one processor, which is cheaper than making one giant chip and lets makers reuse parts., Dennard scalingDennard scaling, n.: The rule that as transistors shrank they also used proportionally less power, so chips got faster without getting hotter. It stopped working in the mid-2000s, ending the era of ever-faster single processors., extreme ultraviolet lithographyextreme ultraviolet lithography, n.: The chip-printing technique used for the most advanced chips, which draws circuit patterns with ultra-short-wavelength light. A single Dutch company, ASML, makes all the machines, each costing well over $100 million., Huang's lawHuang's law, n.: A claim, named after Nvidia's chief executive, that AI chip performance improves much faster than Moore's law because chips, software and number formats all improve together., law of accelerating returnslaw of accelerating returns, n.: Ray Kurzweil's claim that technology improves at an ever-faster rate, because each advance helps build the next, so progress speeds up exponentially rather than steadily.