AMD, a US chipmaker based in Santa Clara, California, said on July 20 that Microsoft will deploy its Helios rack-scale system on Azure for AI inference.

The deployment expands the companies’ partnership across GPUs, central processing units, networking, and software.

AMD said Helios combines Instinct MI455X GPUs, 6th Gen EPYC “Venice” CPUs, Pensando networking, and ROCm software.

AMD plans to ship Helios to customers, including Microsoft, in the second half of 2026.

Microsoft said Azure will use Helios for inference workloads covering frontier models, Azure AI services, and customer applications.

Helios is a 72-GPU rack-scale AI system, not a standalone chip, designed for large-scale AI training and inference.

Microsoft also plans to add Azure HDv2 and HXv2 virtual machines based on Venice processors.

AMD said the companies are also integrating Azure Boost with AMD technologies as Azure expands its use of Pensando data processing units.

According to Microsoft, Azure Boost offloads virtualization, networking, and storage tasks from host servers to dedicated hardware, freeing up CPU capacity for customer workloads.

🔗 Source: AMD

Recent AMD developments