AMD launches helion rack solution for agent-based AI workloads

AMD has unveiled the Helion rack solution, designed for agent-based AI computing workloads. The rack adheres to the OCP design standards and is built on the open platform concept of AMD's Helios rack solution. It allows partners and customers to flexibly combine diverse CPU and GPU resources according to their needs. The pure CPU version, based on the sixth-generation EPYC processor, can scale up to 32,768 cores. AMD plans to contribute the Helion specification to OCP, enabling the community to further develop and adopt it. The company is collaborating with ecosystem partners to develop the Helion rack, set to launch officially in November 2026. AMD aims to provide a scalable and flexible infrastructure for AI workloads, emphasizing the importance of open standards and collaboration. The Helion rack is positioned as a key component in the evolution of AI computing, offering both performance and scalability. By contributing to OCP, AMD is fostering a broader ecosystem for AI innovation. The release is expected to have a significant impact on data centers and AI deployment strategies in the coming months.