AMD Helios is an open, scalable rack-scale AI infrastructure designed for frontier AI and sovereign computing. The co-designed rack integrates 72 AMD Instinct MI455X GPUs (CDNA 5 architecture, MI400/"MI450" series), AMD EPYC "Venice" CPUs and AMD Pensando "Vulcano" NICs, connected via UALink.
The solution delivers up to 2.9 exaFLOPS of FP4 and 1.4 exaFLOPS of FP8 compute and 31 TB of HBM4 memory per rack, with 23.3 TB/s per-GPU memory bandwidth and 260 TB/s of scale-up bandwidth. It runs on the open AMD ROCm software stack, with AMD Infinity Fabric providing CPU-GPU connectivity.
Helios uses an open double-wide ORW (Open Rack Wide) rack format for modern AI datacenters. It targets training and inference of trillion-parameter models; volume deployments are expected in the second half of 2026.
Rack-scale AI system
Which group AMD Helios belongs to and how it is built
A component type covering complete rack-scale AI systems: dozens of GPUs/accelerators tied to CPUs and high-speed networking (scale-up/scale-out) within a single rack, designed for frontier-AI training and inference.
A rack-scale design in which GPUs (AI accelerators), CPUs and NICs are co-designed and densely interconnected (e.g. via UALink, Infinity Fabric, Ethernet), typically in an open rack format (e.g. double-wide ORW), to maximize bandwidth, data movement and scalability for frontier AI.
Basic physical properties of AMD Helios — dimensions, weight and materials
