AMD and HPE Expand Collaboration to Advance Open Rack-Scale AI Infrastructure

AMD has expanded its long-term collaboration with HPE to advance the next generation of open, scalable AI infrastructure with the introduction of the AMD “Helios” rack-scale architecture. HPE will be among the first system providers to adopt the new design, which integrates AMD EPYC™ CPUs, AMD Instinct™ GPUs, AMD Pensando™ networking, and the ROCm™ open software stack into a unified platform engineered for high-performance AI and HPC environments.

“Helios” is designed to simplify deployment of large-scale AI clusters by offering an open, standards-based architecture built on OCP Open Rack Wide specifications. The platform delivers up to 2.9 exaFLOPS of FP4 performance per rack, with next-generation AMD EPYC “Venice” processors, Instinct MI455X accelerators, and Pensando Vulcano NICs. A purpose-built HPE Juniper Networking scale-up switch—developed with Broadcom and supporting the Ultra Accelerator Link over Ethernet (UALoE) standard—enables seamless, high-bandwidth connectivity across the rack, accelerating deployment for cloud and enterprise customers.

AMD and HPE’s collaboration will also power Europe’s next generation of HPC and AI with Herder, a new supercomputer for the High-Performance Computing Center Stuttgart (HLRS). Built on the HPE Cray Supercomputing GX5000 platform and featuring AMD Instinct MI430X GPUs and next-generation EPYC “Venice” CPUs, Herder will support advanced simulation, machine learning, and hybrid HPC/AI workflows for scientific research and industrial innovation. The system is scheduled for delivery in the second half of 2027.

HPE will bring the AMD “Helios” AI Rack-Scale Architecture to customers globally in 2026.