New AI Chip Operates Decentrally and Energy-Efficiently
NEWS,
TUM Press Release
|
Researchers at the Technical University of Munich (TUM) have developed an innovative AI chip called “AI Pro” that operates without relying on cloud connectivity, making it especially secure and energy-efficient. The chip is based on a neuromorphic architecture inspired by the human brain, allowing it to process data directly on-site.
Prof. Hussam Amrouch is researching the future chip generations in his Garching laboratory at the Siemens Technology Centre.
Thanks to a method called “hyperdimensional computing,” the chip can recognize patterns and similarities without needing millions of training data sets – a completely new approach in AI development. Local processing of sensitive data not only improves cybersecurity but also drastically reduces energy consumption: the chip uses up to ten times less energy than conventional models. Applications range from smartwatches to autonomous drone navigation.
