Microsoft plans to significantly expand its custom-designed Maia chips in Azure data centers starting next year, reducing reliance on Nvidia GPUs. The company will make Maia 300 chips its primary inference system for its products and services. This shift benefits Marvell Technology, which partners on the Maia chips, while pressuring Nvidia as major cloud providers develop custom AI accelerator alternatives.
Axe note: Microsoft's move to custom Maia chips threatens Nvidia’s data center dominance, helping Marvell and Alphabet’s AI ambitions.