In 1965, Gordon Moore — co-founder of Intel — observed that the number of transistors on a chip was doubling about every two years. Each generation could fit twice as many components into the same space. The cost per transistor was plummeting at the same rate.

He predicted the trend would continue for at least a decade. He was wildly wrong. It continued for almost 60 years.

The first commercial microprocessor, Intel's 4004 in 1971, had 2,300 transistors. Today's leading chips contain over 100 billion. The transistors themselves have shrunk from the size of a virus to the size of a few atoms. Apple's M-series chips, made by TSMC, use a 3-nanometer process. A nanometer is one-billionth of a meter. A water molecule is about 0.275 nanometers across. The features being etched into silicon are now only about ten times wider than a single atom.

Manufacturing happens in fabrication plants — fabs — that cost over $20 billion to build. The lithography machines that print the patterns onto silicon are made by a single company, ASML in the Netherlands. Each machine costs $200 million and uses extreme ultraviolet light generated by zapping droplets of molten tin with a laser 50,000 times per second.

The trend is slowing. Below 2 nanometers, transistors hit fundamental quantum limits — electrons start tunneling through walls that are too thin to stop them. The next decade will see the end of geometric scaling and the beginning of new strategies: 3D stacking, photonic computing, specialized AI chips.

The single most powerful trend in modern technology is finally approaching its physical limit.