ChipsThe story, in brief
IBM says it can fit nearly 100 billion transistors on a chip - why the milestone matters
100 billion transistors on a single chip. IBM just reset the hardware race that powers every AI model on earth.
The KeyNews take
Why it matters
Sub-1nm chip architecture is the foundational infrastructure unlocking next-gen AI model training and inference. This transistor density breakthrough directly impacts compute capacity, cost-per-FLOP, and the feasibility of even larger models — making it a critical inflection point for labs racing to scale.
The key facts
5 to knowIBM achieves sub-1 nanometer chip architecture
Nearly 100 billion transistors on single chip
Marks milestone in race for advanced chip density
Direct infrastructure implication for AI training/inference compute
Published June 25, 2026
Go to the source
ZDNet AIzdnet.com
Publisher excerpt: The race for the first sub-one-nanometer chip is over, and IBM has won.