Microsoft unveils Maia 300 inference chip

Microsoft announced its third-generation Maia 300 in-house AI chip, built on TSMC's N2P node with a chiplet design, for production rollout in 2027. Microsoft says it will scale Maia to multi-gigawatt deployments and make it the "primary AI inference system" for its products, reducing reliance on Nvidia and AMD.

6 effects · 4 potentially affected companies · The full cascade analysis is part of the Pro plan. Create free account

Microsoft's Maia 300 pushes the largest hyperscalers toward in-house silicon for inference, the fastest-growing AI compute workload. Microsoft claims 40 percent more tokens per megawatt than leading GPU-based systems, which would weaken the case for buying merchant GPUs for inference at Microsoft. Nvidia and AMD face a risk to a major customer's inference demand, though training demand and the 2027 timeline limit near-term effects. TSMC benefits from leading-edge N2P demand and from the advanced packaging that chiplet designs require. Power efficiency per token becomes the main competitive metric, which favors custom ASICs and the vendors that supply them.