Azure Maia

Azure Maia is Microsoft’s multigenerational, purpose-built AI infrastructure system for running inference workloads across Microsoft products and services. Maia 200 is in production and supports workloads such as Microsoft Copilot, while also being used by Microsoft’s MAI team for model development and evaluation. The system runs OpenAI models, Microsoft MAI models and leading open-source model families.

The next generation, Azure Maia 300, is designed for long-context reasoning, latency-sensitive conversational interactions and sustained multi-step agentic tasks. Microsoft plans to begin Maia 300 production deployments in 2027 and scale the Azure Maia family to multi-gigawatt deployments starting that year. Its design emphasizes memory bandwidth, data movement, networking and power efficiency at cloud scale.

Microsoft reports that Maia 200 delivers up to 1.7x more tokens per watt than leading GPU infrastructure for built-in AI capabilities in Teams, Copilot and Cowork experiences. For Maia 300, Microsoft expects 40% more tokens per megawatt than leading GPU-based systems and up to 7x the rack-scale token throughput of Maia 200 in a full 96-accelerator configuration. These figures are Microsoft-reported results and expectations from the cited announcement.