AI INVESTING INSIGHT
Microsoft sets multi-gigawatt rollout for custom Maia AI chips
Updated October 12, 2026 • Educational content
Why it matters: Microsoft says Maia 300 is designed for inference and expects 40% more tokens per megawatt than leading GPU-based systems. A deployment at the announced scale could affect AI-chip demand, data-center power needs, and competition among accelerator suppliers.
Important caveat: This is a forward-looking company plan, not evidence that the chips have already reached that scale or that demand for third-party GPUs is falling. Microsoft says Maia 300 production deployments begin in 2027. Broader market conditions were comparatively calm: Reuters described U.S. index futures as little changed earlier today.