Microsoft has published the specifications for Azure Maia 300, the third generation of its in-house AI accelerator, and said it will scale the Maia family to multi-gigawatt deployments from 2027. In the announcement, cloud chief Scott Guthrie describes Maia as becoming the primary inference system for Microsoft products and services.
The figures Microsoft gives are system level rather than per-chip compute. Each accelerator carries eight HBM4 stacks delivering 18.0 TB/s of peak memory bandwidth inside a 950-watt power budget, with compute chiplets on TSMC’s N2P node and separate chiplets handling memory control and I/O. A rack holds 96 accelerators in a 167-kilowatt envelope, with 27.6 TB of HBM capacity and 1.7 PB/s of aggregate bandwidth.
The comparative claims are Microsoft’s own and unaudited. It expects Maia 300 to deliver 40% more tokens per megawatt than leading GPU-based systems, and up to seven times the rack-scale token throughput of Maia 200. Microsoft does not name the GPU systems in either comparison, and no third party has measured them.
Maia 200 is already in production, running Copilot and deploying across the United States, Europe and Australia, with GitHub Copilot and Microsoft Foundry to follow by the end of the year. Maia 300 begins production deployment in 2027, which leaves a year before the claims meet shipping hardware.