Microsoft unveils Maia 300 AI chip, plans 2027 rollout to production workloads

Microsoft announced its third-generation Maia AI chip, Maia 300, positioning it as the primary AI inference system for its products and services. The company says Maia 300 improves cost and efficiency for running AI workloads in the cloud while supporting newer reasoning models and agentic tasks.

Maia 300 is scheduled to begin rolling out to production AI workloads in 2027, according to Microsoft’s cloud executive Scott Guthrie. Microsoft plans to scale its Azure Maia family to multi-gigawatt deployments starting in 2027, aiming for Maia to become the primary inference system for its AI services.

Microsoft claims Maia 300 delivers 40% more tokens per megawatt than leading GPU-based systems. In a full 96-accelerator configuration, the company says Maia 300 achieves up to 7x the rack-scale token throughput of its predecessor, Maia 200. Maia 200 is already in production across Microsoft’s AI services, including Copilot, GitHub Copilot, Microsoft Foundry, and internal AI models.

Maia 300 uses a chiplet architecture built on TSMC’s N2P process node, separating compute, memory controller, and I/O functions to increase silicon area dedicated to compute performance. Microsoft states this design allows each function to use the technology best suited to its role.

The chip is part of Microsoft’s broader strategy to co-design hardware and AI models, with Microsoft AI CEO Mustafa Suleyman emphasizing the need to own and co-design the platform at scale. Suleyman noted that running Microsoft’s AI models on Maia is a critical focus for the team as it aims to compete with leading labs like Google DeepMind, OpenAI, and Anthropic.