WS #15724
Microsoft is scaling its custom Maia 300 inference chips to multi-gigawatt deployments, aiming to reduce reliance on Nvidia hardware. This development signals a long-term erosion of Nvidia's pricing power in the inference market, as hyperscalers prioritize cost efficiency and supply chain diversification. While Nvidia remains dominant in training, the inference segment faces increasing competitive pressure from domestic chipmakers.
Microsoft Maia vs Nvidia AI Chip War
Microsoft is scaling its custom Maia 300 inference chips to multi-gigawatt deployments, aiming to reduce reliance on Nvidia hardware. This development signals a long-term erosion of Nvidia's pricing power in the inference market, as hyperscalers prioritize cost efficiency and supply chain diversification. While Nvidia remains dominant in training, the inference segment faces increasing competitive pressure from domestic chipmakers.