SignalSpawn
Technology

Microsoft puts AMD Helios on Azure — MI455X racks for frontier inference in H2 2026

Microsoft puts AMD Helios on Azure — MI455X racks for frontier inference in H2 2026

Nvidia still owns the AI rack narrative. Microsoft just wrote AMD a louder chapter. On July 20, Azure announced it will deploy AMD’s Helios rack-scale platform for frontier-model inference and Azure AI services — Instinct MI455X GPUs, 6th-gen EPYC “Venice” CPUs, Pensando networking, ROCm — with customer shipments aimed at the second half of 2026. Meta, OpenAI, and Oracle were already on the Helios dance card; Azure is the hyperscale receipt that moves the stock price.

AMD Helios / Azure AI infrastructure
Helios rack-scale arrives on Azure as ND MI455X v7 for production inference.

Three Azure SKUs, one partnership

ND MI455X v7 is the Helios inference VM line. HDv2 pairs nearly 500 physical Venice cores, 4TB RAM, 32TB local NVMe, and 400Gb Azure Boost networking for data prep, search, reinforcement learning, and agent coordination. HXv2 targets chip design / EDA with 176 Venice cores above 5 GHz, fat caches, up to ~4TB RAM, and 800Gb InfiniBand for MPI-scale sims. Pensando DPUs also absorb more Azure Boost networking/storage work.

“Through our collaboration with AMD, we are expanding the Azure infrastructure portfolio with AMD Helios to give customers the performance, scale, and choice they need.”

— Satya Nadella — Microsoft / AMD announcement (July 2026)

Microsoft blog: Microsoft expands Azure AI and HPC infrastructure with AMD.

Choice is the slogan. Heterogeneous silicon is the strategy. CUDA is still the mountain. Helios on Azure is AMD’s best climbing gear yet.

Author

Cristiano Lima

Date Published

Keep reading

View all