Microsoft has unveiled Maia 300, the third generation of its custom AI chips, as the company accelerates its push to power its own AI services on its own silicon. Microsoft says it plans to deploy more than one million Maia 300 chips next year, with multi-gigawatt rollouts beginning in the same timeframe.
The chip is built on TSMC 2nm chiplets, and Microsoft claims it delivers roughly 40% more useful compute per megawatt than leading GPU-based systems — a direct pitch at the efficiency bottleneck constraining the AI boom.
The announcement is the latest escalation in Big Tech's silicon arms race. By designing its own AI accelerators, Microsoft reduces its dependence on external suppliers while tailoring hardware precisely to its cloud and AI workloads. A million-chip deployment next year would make Microsoft's custom silicon one of the largest AI compute fleets in the world.