Brain-inspired neuromorphic computer uses tiny LEDs to cut AI energy use

German researchers from Technische Universität Braunschweig, Leibniz University Hannover, Ostfalia University of Applied Sciences, and PTB are developing a neuromorphic computer called BRIGHT that uses microscopic LEDs instead of traditional silicon transistors to drastically reduce AI energy consumption. Funded with USD 17.6 million by the state of Lower Saxony and the Volkswagen Foundation, the project aims to create a brain-inspired architecture that enables massively parallel signal processing, mimicking neuronal communication while consuming only a fraction of the power used by conventional AI hardware. A working demonstrator has already been built at the LENA research center, with plans to optimize optical connections, LED components, and hybrid chip integration over the next five years. The innovation combines silicon-based CMOS circuits with gallium nitride LED devices in a hybrid system, merging logic/control and efficient light emission technologies. This approach supports neuromorphic computing by implementing neural networks directly in hardware rather than simulating them digitally, offering a more energy-efficient alternative to