As artificial intelligence (AI) progresses towards the era of intelligent agents, the computational requirements in data centers are undergoing a significant transformation. On October 12, 2026, AMD introduced the Helion rack-level AI computing reference design. Built in accordance with the Open Compute Project (OCP) standards, this solution primarily revolves around CPUs and can be flexibly integrated with GPUs based on specific needs. It presents two distinct configurations: a CPU-centric rack capable of supporting up to 32,768 CPU cores, and an integrated rack that combines CPUs, GPUs, and general-purpose computing capabilities. These configurations are respectively aimed at hyperscale deployments and enterprise-level AI agent applications. AMD intends to contribute the Helion specifications to the OCP community. Moreover, the relevant rack configurations will be collaboratively developed with ecosystem partners, with an anticipated launch in November 2026.