gAIn in the Press: heise online on Neuromorphic Chips and the Energy Problem of AI

Can brain-inspired hardware cut the energy bill of data centers?

A feature article by iX/heise online, “Mehr KI-Energieeffizienz in Rechenzentren dank neuromorpher Chips,” examines how neuromorphic hardware could reduce the energy consumption of AI by processing information event-based, in parallel and with memory and computation fused – following the model of the human brain, which delivers roughly an exaflop on about 20 watts.

The article features the gAIn project and three of its members: our gAIn PIs Prof. Holger Boche (TU Munich) and Prof. Gitta Kutyniok (LMU Munich) and Pit Hofmann, junior research group leader for alternative computing and communication approaches at the Deutsche Telekom Chair of Communication Networks, TU Dresden.

At gAIn we investigate the mathematical foundations and hardware–software approaches required for reliable and resource-efficient next-generation AI computing. It is good to see these ideas – and the Bavaria–Saxony collaboration behind them – reach a broad technical readership.

Article (German): https://www.heise.de/hintergrund/Mehr-KI-Energieeffizienz-in-Rechenzentren-dank-neuromorpher-Chips-11361245.html