Microsoft plans to significantly expand its in-house Azure Maia AI chip family, with multi-gigawatt deployments starting in 2027. The company aims to make its custom silicon the primary platform for running AI models across products such as Microsoft Copilot and services from its OpenAI partnership. The move is intended to improve AI performance and energy efficiency while reducing reliance on costly third-party processors.
The next-generation Maia 300 is expected to begin production deployments in 2027. It is designed to meet growing demand for AI workloads and is projected to deliver 40% more tokens per megawatt than leading GPU-based systems.
Built using TSMC’s N2P process technology, Maia 300 uses a multi-chiplet design to balance compute, memory, and power more efficiently at cloud scale.