Physics World Weekly Podcast

Reversible computing could help solve AI’s looming energy crisis

Jun 26, 2025
Hannah Earley, a mathematician and physicist and co-founder of Vaire Computing, dives into the revolutionary world of reversible computing. She explains how this technology could dramatically slash energy consumption in AI applications. The discussion covers the evolution from theoretical concepts to tangible hardware, including innovative prototypes currently under development. Earley also touches on how specialized logic gates, like Toffoli and Fredkin, can enhance overall computing efficiency while paving the way for a sustainable future.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

Energy Efficiency of Reversible Computing

  • Reversible computing can drastically reduce energy use by avoiding information erasure heat costs.
  • It leverages physical reversibility to approach thermodynamic energy limits in computation.
INSIGHT

How Reversible CMOS Saves Energy

  • Reversible CMOS computing recycles signal energy instead of dissipating it as heat.
  • It recovers charge via reservoirs like inductors, greatly reducing power wastage compared to conventional logic.
INSIGHT

Adiabatic Operations Enable Efficiency

  • Adiabatic operations in reversible computing use trapezoidal signal waves to slow switching and limit energy loss.
  • Slowing switching relative to intrinsic transistor speeds lets reversible chips run in GHz with major energy savings.
Get the Snipd Podcast app to discover more snips from this episode
Get the app