Run agent workloads close to the user, reducing latency for interactive edge-computing agents. Deploy globally distributed applications with built-in GPU support for AI inference at the edge.