6G workloads demand more than general purpose processors. While accelerators offer great performance, they are vendor-specific and static
Most AI tools were designed for datacenters, not networks. As a result, AI integration proposes a big challenge for the evolution to 6G.
Since networks are shared amongst users, keeping sensitive workloads secure becomes progressively harder.
SHARC develops an AI-native RAN architecture that is energy efficient and accelerator friendly, while ensuring execution security.