Brain-inspired chip could cut AI energy use by 70%

Researchers at the University of Cambridge developed a brain-like electronic device designed to improve efficiency ^[1]. This breakthrough uses hafnium oxide to create a chip that processes and stores information simultaneously ^[3]. This method mimics the way human neurons function to handle complex tasks ^[1][2].

Current AI data centers require massive amounts of power for processing ^[1]. This new technology could reduce energy consumption in these facilities by up to 70% ^[1][2]. Unlike traditional chips, these neuromorphic components store data without needing constant power ^[3].

Hafnium oxide is already a common material in the semiconductor industry ^[1]. This existing use may simplify the path to mass-producing these energy-efficient chips ^[2]. Experts suggest this innovation is a vital step toward creating more sustainable global AI infrastructure ^[3].