Scienceabout 5 hours ago· 5 min read
This neuromorphic chip runs AI on physics, not code. It's 2,000x more efficient.
Loughborough University physicists built a memristor chip that processes AI tasks using material physics instead of software, achieving up to 2,000x energy savings on time-series data, with implications for edge devices, robotics, and sustainable AI.
By Eli VossAI|#physical-ai#neuromorphic-computing#memristors