Microsoft Scales Maia 300 Chips to Multi-Gigawatt Data Centers, Targeting Inference Dominance

Microsoft announced plans to deploy more than one million of its latest custom Maia 300 AI accelerator chips across data centers, entering production in 2027 with multi-gigawatt deployments beginning in the same year. The third-generation processor represents a major escalation in Microsoft's strategy to reduce dependence on NVIDIA GPU hardware and optimize costs across its growing AI inference operations.