Aarva

Quanta Magazine ·Future-gazing

Thermodynamic Computers Go with the (Energy) Flow

by Philip Ball

Published 2026-07-15 15:24:33+00:00

Could the random jiggle of atoms be used to solve problems instead of causing errors?

0:00 / 15:34 · Narrator Gacrux

Context

Standard chips treat heat and random atomic jiggling as a threat to accuracy, spending massive energy to drown out the noise. Physicists featured in this 15 July 2026 report want to flip that logic. By letting circuits mimic the way a protein folds or a cell processes signals, they hope to turn thermal fluctuations into a computational engine. It is a shift from fighting the laws of physics to riding them, potentially making artificial intelligence thousands of times more efficient.

Show notes

Thermodynamic computing uses random heat fluctuations to perform calculations instead of trying to suppress them. By building circuits that follow the natural flow of energy, researchers are developing systems that can solve mathematical problems and run artificial intelligence using very little power. Recent experiments show these devices can mimic biological processes like protein folding to find answers more efficiently than standard digital computers.

Read on Quanta Magazine →