Microsoft to Launch Maia 300 AI Chip in September 2026

Microsoft accelerates custom chip roadmap

Microsoft is preparing to unveil its Maia 300 AI chip as early as September 2026, marking the third generation of the company's custom silicon designed to power artificial intelligence workloads, according to a report from The Information citing people with direct knowledge of the plans.

The announcement would come nearly three years after Microsoft first introduced the Maia line in November 2023. The company has been working to catch up with rivals Alphabet and Amazon, both of which have made significant progress scaling their in-house chip operations.

Manufacturing and production targets

Microsoft is in active discussions with Taiwan Semiconductor Manufacturing Company (TSMC) to secure manufacturing capacity for more than 300,000 Maia 300 units scheduled for delivery in 2027, The Information reported. The company's longer-term ambition extends to securing capacity for over 1 million chips, though component supply constraints and ongoing capacity negotiations with TSMC could limit those plans.

The Maia 200, Microsoft's second-generation chip unveiled in January 2026, was built by TSMC using 3-nanometer process technology and incorporated substantial amounts of SRAM—a type of memory that can deliver speed advantages for AI systems processing high volumes of user requests.

Competing in the custom AI chip market

Microsoft's chip development efforts are part of a broader industry shift as major cloud providers work to reduce their dependence on Nvidia's expensive processors. Google has already begun recognizing revenue from direct sales of its Tensor Processing Units in the quarter ended June 2026, while Amazon has seen growing adoption of its Trainium and other custom processors.

Microsoft is also working to persuade major cloud customers, including AI company Anthropic, to adopt the Maia 300 chip for their workloads.

Why it matters

The race to develop custom AI chips represents a fundamental shift in how hyperscale cloud providers manage both costs and competitive positioning. Success in deploying proprietary silicon at scale could reduce Microsoft's chip procurement costs by billions of dollars annually while giving the company greater control over performance optimization for its AI services. The ability to deliver 300,000 chips by 2027 would signal Microsoft has overcome early production challenges that left it trailing competitors in the custom chip market.

Neither Microsoft nor TSMC responded to requests for comment on the plans, which were first reported by The Information.