Microsoft says Maia 200 is in production and plans Maia 300 deployment in 2027
Microsoft says Azure Maia 200 is serving production workloads, including Microsoft Copilot, and that Maia 300 will begin production deployments in 2027. The company plans to scale the Maia family to multi-gigawatt deployments starting that year, positioning it as its primary inference system for Microsoft products and services.
Microsoft describes Maia 300 as an inference-first system for long-context, memory-intensive, decode-heavy workloads, including sustained multi-step agent tasks. The company claims 40% more tokens per megawatt than leading GPU-based systems and up to seven times Maia 200’s rack-scale token throughput in a 96-accelerator configuration. These are Microsoft’s stated comparisons, not independently verified results. The announcement also describes eight HBM4 stacks, 18.0 TB/s peak memory bandwidth, and a 950-watt TDP.