Microsoft bets its cloud on its own AI chips

Every Copilot question you ask has a hidden Nvidia tax baked into it. Satya just laid out the plan to stop paying it. A high-end Nvidia chip goes for over $30,000, and Nvidia's gross margin runs north of 70%. Microsoft can reportedly build its own Maia chip for about a third of that price. So starting next year, Microsoft's own products run on the homemade one. That's the fleet split hiding in this announcement. Copilot, Teams, Office, Bing, everything with a Microsoft logo moves onto silicon Microsoft makes at cost. The pricey Nvidia racks get saved for Azure customers, who pay rental rates for them. Microsoft keeps the markup when it's the seller and dodges it when it's the buyer. At billions of AI requests a day, that dodge adds up to one of the biggest cost savings in the company's…

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% more tokens per watt compared to leading GPU-based systems. We are already deploying Maia 200 across the US, Europe, and Australia, delivering up to 1.7x more tokens per watt across experiences in Teams and Copilot. Our goal is to build on the full-stack efficiency gains we’ve already seen, while continuing to bring the latest innovations from across the ecosystem into our fleet.

Microsoft’s new AI chip makes it less reliant on Nvidia’s GPUs. The new Maia 300 chip will start rolling out to production AI workloads in 2027

JUST IN: Microsoft has announced a new AI chip, Maia 300. Maia 300 delivers ​about 40% more computing per megawatt than leading GPU-based systems, per Reuters.