Microsoft Unveils Maia 300 AI Chip as It Moves Away From Nvidia GPUs

Microsoft Introduces the Maia 300

Microsoft announced its third-generation in-house AI chip, the Maia 300, on October 12, 2026. The company describes it as the "primary AI inference system" for its products and services. The chip is designed to improve the cost and efficiency of running AI workloads in the cloud, while also supporting the latest reasoning models and agentic tasks. Production rollout to AI workloads is planned for 2027.

Efficiency Gains Over GPU Systems

Scott Guthrie, executive vice president of Microsoft's cloud group, said Maia 300 is expected to deliver 40 percent more tokens per megawatt compared to leading GPU-based systems. In a full 96-accelerator configuration, it delivers up to 7x the rack-scale token throughput of its predecessor, the Maia 200. Microsoft says these improvements will allow it to rely mainly on its own AI chips instead of those from Nvidia and AMD, scaling the Azure Maia family to multi-gigawatt deployments starting in 2027.

Chiplet Design Built on TSMC N2P

Maia 200 was announced in January and already runs production workloads across Microsoft Copilot, GitHub Copilot, Microsoft Foundry, and the company's own AI models. Maia 300 uses a chiplet architecture built on TSMC's N2P process node, separating compute, memory controller, and I/O functions. The Azure hardware and infrastructure team said this increases silicon area dedicated to compute and lets each function use the technology best suited to its role. The new chip targets long-context reasoning, latency-sensitive conversational experiences, and sustained multi-step agentic tasks.

A Push for Full-Stack AI Control

Microsoft AI CEO Mustafa Suleyman said the company has to work across the full stack, down to the silicon, and needs a platform it owns and co-designs with its models. Running Microsoft AI models on the Maia chip is described as a critical focus for the team. The chip is also part of Microsoft's broader effort to catch up with Google DeepMind, OpenAI, and Anthropic in AI model development.

Sources: The Verge Ai