Microsoft rolls out Azure Maia 200 across the US, Europe and Australia

Microsoft is to scale its Azure Maia system to multi-gigawatt deployments from 2027. The Maia 200 AI accelerator is being deployed across the US, Europe and Australia, with Maia 300 expected in 2027.

The Azure Maia programme has become a “foundational element” of Microsoft’s AI infrastructure strategy, according to Scott Guthrie, executive vice president of cloud and AI at Microsoft. The Maia family of custom AI chips is designed to power AI inference workloads across Microsoft’s cloud services and applications. This expansion will make Maia the primary inference system for Microsoft products and services.

Maia 200 is serving production workloads, including Microsoft Copilot, and running OpenAI models, Microsoft’s own AI models and leading open-source model families. Microsoft said Maia 200 is supporting the Microsoft AI team as it develops next-generation models, with GitHub Copilot and Microsoft Foundry expected to run on the platform by the end of 2026. Microsoft is using insights gained from running AI services at global scale to co-design hardware and software to optimise real-world use cases for enterprise customers across multiple sectors. According to Microsoft, Maia 200 delivers up to 1.7 times more tokens per watt for AI features in Teams, Copilot and Microsoft 365 Copilot Chat than infrastructure powered by leading graphics processing units (GPUs).

Maia 300 is the third generation of the Azure Maia programme. It is being designed for emerging AI workloads, such as long-context reasoning, agentic AI systems and memory-intensive large language models. According to Microsoft, Maia 300 is expected to deliver a40 per cent more tokens per megawatt than leading GPU-based systems. A full deployment is expected to achieve up to seven times the rack-scale token throughput of its predecessor, Maia 200. A Maia 300 rack can combine 96 accelerators with 27.6TB of high-bandwidth memory, enabling data centres to support more simultaneous AI workloads and interactions within a 167-kilowatt power envelope.

“Our focus is on developing responsible, affordable and scalable platforms that will help everyone amplify their intelligence to drive breakthrough outcomes that matter for them,” said Guthrie in a blog post. “Microsoft is unique because we are able to optimise from silicon to systems to agent on an open, heterogeneous and model diverse platform our customers can count on for their AI transformation, today and in the years ahead. This platform is at the heart of Frontier Transformation, and Azure Maia is a foundational component of our approach to expand access, accelerate innovation and help customers capture more value from AI.”