Brain-like chip cuts AI energy use 70%

University of Cambridge researchers created a hafnium oxide memristor that mimics brain synapses. The device reduces AI energy consumption by up to 70% by combining memory and processing in one place. Findings were published in Science Advances.