Microsoft plans Maia 300 AI chip deployments in 2027

Microsoft announced Maia 300, its third-generation in-house artificial intelligence accelerator, on October 12, saying it plans to begin production deployments in 2027 and make the chip family its primary system for running AI across its products and services. Maia 300 is designed for inference, the computing process that runs trained models to answer questions, generate text and perform other tasks. Microsoft says the chip is aimed at workloads including long conversations, reasoning and multistep tasks handled by AI agents. The company plans to scale its Maia deployments to multigigawatt levels beginning in 2027. Microsoft estimates Maia 300 will deliver 40% more tokens per megawatt than leading GPU-based systems. Tokens are units of text processed or generated by AI models. The company has not identified the systems used for comparison, and the estimate has not been independently tested in the supplied reporting. A fully configured rack will hold 96 Maia 300 accelerators. Microsoft says that rack can deliver up to seven times the token throughput of a Maia 200 rack. The comparison is a company estimate.