Microsoft’s new AI chip makes it less reliant on Nvidia’s GPUs

Microsoft announced Maia 300, its third-generation chip for running AI in its data centers. The design splits compute, memory and input-output into separate pieces of silicon, and Microsoft says it delivers 40 percent more tokens per megawatt than leading GPU systems. Starting in 2027, it becomes the main system for serving Microsoft's AI products, reducing reliance on Nvidia.

Summary written with AI by HowTechMoves, based on reporting by The Verge.