Microsoft will deploy AMD's new Helios rack AI system on Azure to power frontier model inference, the companies announced Monday, marking the first publicly named customer for AMD's most direct challenge yet to Nvidia's data center dominance.
Helios shipments, including to Microsoft, begin in the second half of 2026. The platform bundles AMD's next-generation Instinct MI455X GPUs, sixth-generation EPYC "Venice" processors, Pensando networking tech, and ROCm software into a single integrated rack system for large-scale training and inference.
"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," CEO Lisa Su said. The deal spans three new Azure offerings covering AI data processing, chip design, and production-scale AI inference. Microsoft will also add two new Azure virtual machine instances running on AMD's Venice CPUs, one for agentic AI and data pipelines and another for semiconductor design.
Microsoft CEO Satya Nadella said the company is "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." The announcement positions Helios as the first rack-scale system to rival Nvidia's Grace Blackwell and Vera Rubin platforms, which have dominated the AI hardware market. Microsoft's adoption signals that hyperscalers are actively seeking alternatives to a single chip supplier.
AMD says eight of the top 10 AI companies already run workloads on its Instinct GPUs, including OpenAI, Cohere, and SpaceXAI. Microsoft, which was the first to adopt AMD's MI300X GPU in 2023, is deepening its AMD relationship as it ramps up internal model development. The company announced seven in-house AI models in June.
AMD shares rose more than 5% in Monday trading on the news.
