Microsoft has unveiled Maia 300, its third-generation in-house AI accelerator, and plans to make the Maia family a primary inference platform across its products and services starting in 2027.

CEO Satya Nadella said Microsoft will scale Maia to multi-gigawatt deployments next year. The company says Maia 300 can deliver about 40% more inference tokens per megawatt than leading GPU-based systems and up to seven times the rack-scale token throughput of Maia 200.

Reuters reports the processor uses a chiplet design with key compute functions built on TSMC’s 2-nanometer technology, while other functions use 3-nanometer silicon. A Maia 300 rack will contain 96 chips and draw roughly 167 kilowatts.

The launch pushes Microsoft deeper into the same custom-silicon race BitcoinVersus has followed through Amazon’s AI chip expansion, Intel’s effort to route AI workloads to the right accelerator, and new AI rack power architectures. Microsoft is not abandoning Nvidia or AMD, but its own chips give it another way to control cost, power use, and supply as AI infrastructure scales.

Today we are announcing our biggest bet yet on our own silicon.

Starting next calendar year, we will scale our Maia family to multi-gigawatt deployments, making it the primary inference system for our own product and services.

And we’re introducing Maia 300, which delivers 40%… pic.twitter.com/L490XvOys9

— Satya Nadella (@satyanadella) October 12, 2026

Satya Nadella announced Microsoft’s plan to scale Maia into multi-gigawatt deployments.

This explainer on Maia 200 provides context for the architecture Microsoft is now scaling forward with Maia 300.

Editor’s Note: Performance comparisons are Microsoft claims unless independently benchmarked.