Microsoft to deploy AMD Helios rackscale solution to support inference workloads and Azure services

Microsoft will deploy AMD's Helios rackscale solution to power model inference and Azure AI services, part of an expanded partnership spanning GPUs, CPUs, networking, and software. The double-wide Helios rack, featuring AMD's Zen 6 Epyc CPUs, MI455X GPUs, and Pensando Vulcano NICs, delivers 2.9 exaflops of FP4 performance per rack and will begin shipping to Microsoft in the second half of 2026. Microsoft will also add two new Azure VM series powered by AMD's sixth-generation Epyc CPUs and expand its deployment of Pensando DPUs.

Microsoft has announced it will deploy AMD’s Helios rackscale solution to power model inference both internally and for its AI customers, with the hardware also set to support the company’s Azure AI services. In a blog post https://blogs.microsoft.com/blog/2026/07/20/microsoft-expands-azure-ai-and-hpc-infrastructure-with-amd/ detailing the planned deployment, Microsoft said it forms part of an expanded partnership with AMD which will span the chip company’s graphics processing units GPUs , central processing units CPUs , networking, and software on Microsoft Azure, with Helios sitting “at the center of the expansion.” The financial terms of the deal or its expected capacity have not been shared. Alongside the Helios deployment, the agreement will see Microsoft also add two virtual machine VM series, Azure HDv2 for agentic AI and data pipelines, and Azure HXv2 for semiconductor design, to Azure. Both will be powered by AMD's sixth-generation ‘Venice’ Epyc CPUs. Microsoft will also expand its deployment of Pensando data processing units DPUs , integrating the hardware into its Azure Boost system to improve networking performance, efficiency, and connection processing at cloud scale. First unveiled in June 2025, AMD’s double-wide Helios AI rack https://www.datacenterdynamics.com/en/news/amd-launches-instinct-mi350-gpus-unveils-double-wide-helios-ai-rack-scale-system/ will feature the chip company’s Zen 6 Epyc “Venice” CPUs, MI455X GPUs, and Pensando Vulcano network interface controllers NICs , delivering 2.9 exaflops of FP4 4-bit floating-point precision performance per rack. AMD will begin shipping the rackscale solution to customers, including Microsoft, in the second half of 2026. “AMD and Microsoft have spent years building high-performance infrastructure together, and today we're extending that partnership across the full stack of AMD AI solutions on Azure,” said AMD CEO Dr. Lisa Su. “Microsoft's new AMD deployments mark an important milestone as we deliver leadership compute solutions to Azure customers and scale the next generation of AI infrastructure together.” Microsoft CEO, Satya Nadella, added: “Customers are looking for AI infrastructure that is optimized for a wide range of workloads, from training and inference to data preparation, search, and reinforcement learning. 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 to build and run the next generation of AI applications.”