By Trefis Team
Publication Date: 2026-10-13 15:35:00
Microsoft (MSFT) says its own Maia chips will become the main hardware for running its AI models, reducing its reliance on Nvidia. The news lands while Microsoft is short of capacity: management has told investors for several quarters that demand exceeds supply, and has been spending heavily to add more. So how much of Microsoft’s AI work will move to its own chips, and how soon?
MSFT metrics > Market Cap $3.9T · Revenue $332B · Growth 17.8% · Op Margin 47% · P/E 29.7x
What Happened
On Monday, Microsoft said it intends to make Maia its primary architecture for running AI models. It also said it will eventually have multiple gigawatts’ worth of Maia chips. Maia will handle most of the AI work behind Copilot and the company’s other products. The third-generation Maia 300, introduced the same day, rolls out in 2027 and targets lower costs and greater efficiency in running AI workloads. The business closest to the move is Intelligent Cloud, the segment that…
