AMD introduced Helion racks with CPU and GPU configurations adjustable for agentic AI workloads. The Helios-based design uses EPYC Venice, with a CPU-only configuration described as supporting up to 32,768 Zen 6 cores. It supports air and liquid cooling and includes token routing to direct requests to appropriate resources. AMD announced partner availability in November and plans to release the design through OCP. Maximum configurations alone do not establish real agent throughput or cost advantages.