OneQode partners with AMD to deploy AI infrastructure using Instinct MI355X GPUs and future AMD Helios rack-scale solutions

OneQode announced a collaboration with AMD to deploy AMD Instinct GPUs for global AI infrastructure expansion, starting with AMD Instinct MI355X GPUs and later incorporating the AMD Helios platform based on an open design submitted by Meta at OCP 2025.

Each MI355X GPU includes 288GB of HBM3E memory and delivers 8TB/s of bandwidth. The planned Helios AI rack uses AMD Instinct MI450 Series GPUs, each offering up to 432GB of HBM4 memory and 19.6TB/s memory bandwidth, with a full rack capable of up to 72 MI450 GPUs delivering 1.4 exaFLOPS FP8 and 2.9 exaFLOPS FP4 performance.

The infrastructure is designed to support AI workloads such as frontier model training, enterprise deployments, and sovereign AI scenarios for governments, research bodies, and AI-first organisations, leveraging the open AMD ROCm software stack for a vendor-neutral platform.