Microsoft is expanding its strategic partnership with AMD across GPUs, CPUs, networking and software for its Azure cloud platform. The agreement includes a large-scale deployment of AMD’s Helios rack-scale AI infrastructure to support frontier-model inference, Azure AI services and customer applications.
AMD expects to begin shipping Helios systems to Microsoft and other customers during the second half of 2026. The companies did not disclose the number of systems Microsoft plans to deploy or the financial value of the expanded relationship.
AMD Helios combines Instinct MI455X GPUs, 6th Gen EPYC processors, Pensando networking technologies and the ROCm software platform within an integrated rack-scale architecture. The system is designed to provide the computing, memory, networking and software capabilities needed to operate large AI models across data center clusters.
Microsoft plans to use Helios primarily for AI inference, the process through which trained models generate responses, predictions and other outputs. As organizations move more AI applications into production, inference is becoming an increasingly important part of data center computing demand.
The deployment will support Microsoft’s own frontier models and AI products while providing infrastructure to Azure customers. Developers and enterprises will be able to access AMD-powered computing through Azure services, including Azure Foundry Managed Compute.
Azure Foundry Managed Compute is intended to help organizations deploy and scale production AI workloads without directly managing all of the underlying infrastructure. The addition of AMD Helios gives customers another option alongside the other accelerators and computing systems available through Azure.
The partnership also gives frontier-model developers access to AMD infrastructure for both training and serving large-scale models. Training requires substantial computing resources to build and refine a model, while serving refers to operating that model for users after development.
Microsoft will also introduce two Azure virtual machine series powered by AMD’s 6th Gen EPYC processors, code-named Venice. The new systems will expand Azure’s AMD-powered offerings across AI, data processing and engineering workloads.
Azure HDv2 virtual machines will be designed for agentic AI applications and data pipelines. These workloads can involve multiple AI agents, large datasets and interconnected processing steps that require significant CPU performance and memory bandwidth.
Azure HXv2 virtual machines will target semiconductor design and other technically intensive engineering workloads. Chip design applications frequently require large amounts of computing capacity to perform simulations, verification and electronic design automation tasks.
The collaboration also extends to the networking infrastructure connecting Azure’s computing systems. Microsoft is broadening its use of AMD Pensando data processing units across AI backend networks and selected Azure services.
Data processing units offload networking, security and infrastructure tasks that would otherwise consume CPU resources. This allows the primary processors and AI accelerators to devote more of their capacity to customer applications and model workloads.
AMD and Microsoft are also integrating Pensando technologies with Azure Boost. Azure Boost separates virtualization and infrastructure functions from customer workloads to improve networking performance, storage efficiency and connection processing across Microsoft’s cloud fleet.
The expanded partnership represents a full-stack deployment of AMD technology rather than the use of a single processor category. Microsoft will combine AMD accelerators, CPUs, networking products and software to support increasingly large and complex AI systems.
ROCm, AMD’s open software platform for GPU computing, provides tools and libraries that allow developers to run AI and high-performance computing workloads on AMD accelerators. Software compatibility and model support will be important as Microsoft scales Helios across Azure and makes the infrastructure available to customers.
Microsoft said customers are seeking infrastructure optimized for a wide range of AI activities, including training, inference, data preparation, search and reinforcement learning. The expanded AMD portfolio is intended to give those customers additional choice when selecting infrastructure for each stage of AI development and deployment.
For AMD, the agreement provides a major cloud deployment for its next-generation AI platform and strengthens its long-running relationship with Microsoft. The companies expect to continue collaborating on open, high-performance infrastructure as demand for AI computing, networking and data processing grows.
KEY QUOTES:
“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. 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.”
Dr. Lisa Su, Chair and CEO of AMD
“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.”
Satya Nadella, Chairman and CEO of Microsoft
