GoKawiilMicrosoft unveils Maia 300 AI chip to replace most Nvidia GPU workloads by 2027
Microsoft announced its third-generation Maia 300 AI chip, which it plans to make the primary inference system for its products and services starting in 2027. Executive Scott Guthrie said the chip delivers 40 percent more tokens per megawatt than leading GPU systems and up to 7x the rack-scale throughput of its predecessor, Maia 200, which launched in January and already powers Copilot, GitHub Copilot and Microsoft Foundry.
GoKawiil's interpretation of the reporting above, not reported fact.
The move suggests Microsoft wants to cut its dependence on Nvidia and AMD GPUs for cloud AI inference, which could lower costs and reduce supply constraints tied to chip shortages. Guthrie's claim that Maia will scale to multi-gigawatt deployments by 2027 indicates Microsoft expects custom silicon to handle increasingly complex reasoning and agentic AI tasks at scale, though actual performance versus Nvidia hardware in production remains to be independently verified.
- Maia 300 is built on TSMC's N2P process and uses a new chiplet architecture separating compute and memory.
- Microsoft claims 40% more tokens per megawatt and 7x rack-scale throughput versus Maia 200.
- Microsoft aims to make Maia its primary AI inference system via multi-gigawatt deployments starting in 2027.
Source: theverge.com — Tom Warren, 2026-10-12
Published there as: “Microsoft’s new AI chip makes it less reliant on Nvidia’s GPUs”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.