Microsoft Unveils Maia 300, Its Next-Generation AI Inference Chip

Microsoft announced Azure Maia 300 on October 12, 2026, the third generation of its in-house AI accelerator. The company says it will scale the Maia family to multi-gigawatt deployments starting in 2027, making it the primary inference system behind Microsoft products and services. Maia 300 is designed for long-context reasoning, low-latency conversational AI and sustained multi-step agentic tasks. It uses a multi-chiplet design that separates compute, memory-controller and I/O functions, with compute chiplets built on an advanced TSMC 2nm-class process. A full rack holds 96 accelerators within a 167 kW power envelope, providing 27.6 TB of HBM memory and 1.7 PB/s of aggregate bandwidth. Microsoft expects up to 7x the rack-scale token throughput of Maia 200 and 40% more tokens per megawatt than leading GPU-based systems. As The Verge notes, the move reduces Microsoft's dependence on Nvidia and AMD hardware. Maia 200, announced in January, already runs production workloads such as Microsoft Copilot, GitHub Copilot and Microsoft Foundry.