AI news · Microsoft · chips
Microsoft plans to answer AI requests mainly on its Maia chips
Microsoft says its third in-house AI chip will become the main system answering requests in its products, starting in 2027.
Microsoft unveiled Maia 300, its third in-house AI chip, on Monday 12 October 2026. It plans multi-gigawatt deployments of its Maia chips from 2027 and wants them to be the “primary inference system” for its products. Inference is the work of answering each request once a model is trained, such as every Copilot reply.
The company expects Maia 300 to produce 40 per cent more tokens, the small chunks of text a model writes, per megawatt than leading systems built on GPUs, the chips most AI runs on today. A full rack holds 96 chips and, Microsoft says, handles up to seven times the output of a Maia 200 rack. The chip is made on TSMC’s N2P process, one of its newest chip-making methods, The Verge reports.
Nvidia keeps a role. Microsoft still trains models with Nvidia and AMD hardware, and analysts at Evercore ISI estimate Maia will be about 10 per cent of Microsoft’s AI chip spending by the first quarter of 2027, CNBC reports.
If Maia delivers the savings Microsoft expects, it lowers the cost of every AI reply Microsoft sells, from Copilot in Office to its cloud services, and that cost shapes what customers pay.
Sources
- Microsoft: Microsoft to expand Azure Maia system to multi-gigawatt deployments. Checked
- The Verge: Microsoft’s new AI chip makes it less reliant on Nvidia’s GPUs. Checked
- CNBC: Microsoft will mainly use its Maia chips for AI inference. Checked
Alexandre S.
Alexandre S. is a creative ad strategist and copywriter with ten years' experience in digital marketing across multiple channels.