UK physicists’ brain-inspired chip could make AI systems 2,000 times more energy efficient

Researchers at Loughborough University built a neuromorphic chip that processes time-varying data in hardware.

The device uses a niobium-oxide thin-film memristor with random nanopores to form a physical reservoir.

Tests showed it can perform XOR, pixelated number recognition, and Lorenz-63 time-series prediction.

The team reported energy usage up to 2,000 times lower than some software-based methods for certain tasks.