SANTA CLARA, Calif., October 2 — The observation that has powered the digital age for six decades is running into the hard limits of physics, and the industry built on it is reinventing itself in real time.
The prediction that held
In 1965, Gordon Moore — then running research at Fairchild Semiconductor, later co-founding Intel — wrote in Electronics Magazine that the number of components on a chip would climb from about 60 to 60,000 by 1975. He was essentially right. The Intel 1103 DRAM arrived in 1970 with 1,024 bits.
The 4004 microprocessor followed in 1971 with 2,300 transistors. By 1978 the 8086 packed 29,000.
The 80386 hit 275,000 in 1985. The Pentium reached 3.1 million in 1993. The Pentium 4 pushed to 42 million in 2000.
The Core 2 Duo crossed 291 million in 2006. Today’s leading chips count transistors in the tens of billions.
Why it worked for so long
The industry kept the pace through a parade of manufacturing breakthroughs: photolithography that prints ever-finer circuits, wafers that grew from two inches to 300 millimeters, and the switch to CMOS technology that made chips both faster and cooler. Each step bought another doubling. But transistors are now measured in atoms.
TSMC and Samsung reached 5-nanometer commercial production around 2020. At that scale, electrons start to tunnel through barriers they shouldn’t cross, leakage currents climb, and heat becomes a design crisis rather than a nuisance.
The old playbook — just shrink everything — has hit a quantum ceiling. Chipmakers have pivoted. Multi-core processors spread work across many slower cores instead of one blazing-fast one.
Specialized accelerators — GPUs for graphics and AI, TPUs for tensor math — handle specific workloads far more efficiently than general-purpose cores. And 3D stacking lets designers build upward, layering logic on memory to shorten data paths.
TSMC’s 3-nanometer process entered volume production in 2022; Samsung’s 3-nanometer gate-all-around technology followed. A single 3-nanometer wafer from TSMC now costs over $20,000, a price tag that reflects the staggering complexity of atomic-scale fabrication.
What comes next
The next generation of transistors will likely abandon the planar finFET for gate-all-around FETs and nanosheet designs that wrap the channel on all sides. Materials beyond silicon — gallium nitride and silicon carbide — are already moving into power electronics. Meanwhile, research into quantum computing and neuromorphic chips aims to sidestep classical scaling limits entirely.
The practical upshot: the exponential curve Moore drew isn’t ending, but it is bending into new shapes. Watch for the first commercial gate-all-around chips and the slow migration of quantum accelerators from lab to cloud — that’s where the next doubling lives.
