2025 OCP Summit》AMD announces the adoption of Meta open rack standard to build Helios rack-level AI platform

Chip manufacturer AMD publicly displayed a static model of the "Helios" rack-level platform for the first time at the 2025 OCP Global Summit. "Helios" was developed using the new Open Rack Wide (ORW) specification proposed by Meta, extending AMD's open hardware concept from chips to systems and racks, representing a major progress in open and interoperable AI infrastructure.

AMD stated that “Helios” continues AMD’s leadership in the fields of AI and high-performance computing (HPC), laying the foundation for providing open and scalable infrastructure and providing strong momentum for the growing global demand for AI. To meet the needs of GW-scale data centers, the new ORW specification defines an open, double-width rack optimized for the power, cooling, and serviceability needs of next-generation AI systems. By adopting ORW and OCP standards, "Helios" provides the industry with a unified and standardized foundation to support large-scale development and deployment of high-performance AI infrastructure.

Forrest Norrod, executive vice president and general manager of AMD’s Data Center Solutions Group, said that open collaboration is the key to effectively scaling AI. Through "Helios", we transform open standards into practically deployable systems, combining AMD Instinct GPUs, EPYC CPUs and open network architecture to provide the industry with a flexible, high-performance platform built specifically for the next generation of AI workloads.

AMD “Helios” rack-level platform integrates open computing standard architectures such as OCP DC-MHS, UALink and Ultrafast Ethernet Consortium (UEC), and supports open scale-up and scale-out network architectures. The rack features quick-detach liquid cooling for sustained cooling performance, a double-width configuration for improved serviceability, and multi-path flexibility with standards-based Ethernet networking.

As a reference design, "Helios" enables OEMs, ODMs, and hyperscalers to quickly adopt, expand, and customize open AI systems, thereby shortening deployment time, improving interoperability, and supporting the effective expansion of AI and HPC workloads. The “Helios” platform demonstrates AMD’s continued cooperation with the OCP community to help build an open and scalable infrastructure for global AI deployment.