Details
- AMD and Microsoft announced an expanded strategic partnership covering GPUs, CPUs, networking and software on Azure, centered on deploying the AMD Helios rack-scale AI solution for frontier model inference across Microsoft, Azure AI services and customer workloads.
- The collaboration involves AMD Instinct MI455X GPUs, 6th Gen AMD EPYC "Venice" processors, Pensando DPUs, and ROCm software integrated into Helios, alongside broader Pensando DPU use in Azure's AI backend networking and services.
- Azure will introduce two new VM series powered by 6th Gen EPYC "Venice": HDv2 targeting agentic AI and data pipelines, and HXv2 aimed at semiconductor design and engineering workloads, expanding Azure's AMD EPYC-based portfolio.
- Microsoft and AMD are integrating AMD silicon with Azure Boost to improve networking performance, efficiency and connection processing at cloud scale, building on Microsoft’s existing Pensando DPU deployment.
- AMD will begin shipping Helios to customers, including Microsoft, in the second half of 2026, positioning Helios as an open rack-scale platform for large-scale AI training and inference and adding Azure to a growing list of hyperscale Helios adopters.
Impact
This partnership strengthens AMD’s position in hyperscale AI infrastructure and diversifies Azure’s high-end AI compute beyond incumbent GPU providers, supporting frontier model inference and production AI workloads. Over the next 12–24 months, Helios and EPYC-powered VMs are likely to influence cloud AI economics, give enterprises more architectural choice for agentic and data-intensive workloads, and accelerate investment in rack-scale, networking-optimized AI systems.
