Microsoft to Unveil Maia 300 AI Chip in September, Securing Production for 300,000 Units

Microsoft plans to unveil its new Maia 300 AI chip as early as September 2026 to power its Azure and Copilot services. To reduce its reliance on Nvidia's expensive processors, Microsoft is negotiating with TSMC to secure manufacturing capacity for over 300,000 units for delivery in 2027, with a long-term target exceeding 1 million units. Microsoft is also pitching the chip to major cloud customers like Anthropic as it plays catch-up to rivals Alphabet and Amazon, which are already scaling their own custom silicon.

Maia 300 chip launch timeline

• ▪The Information first reported Microsoft's plans to launch the Maia 300 chip, citing people with direct knowledge of the matter.
• ▪Microsoft is planning to unveil its new Maia 300 artificial intelligence chip in the fall of 2026, potentially as early as September 2026.

Reducing reliance on Nvidia

• ▪Microsoft is seeking to persuade major cloud customers, including the artificial intelligence startup Anthropic, to adopt its custom Maia chips.
• ▪Microsoft is expanding its in-house semiconductor program to reduce its reliance on Nvidia's expensive processors.

TSMC production capacity negotiations

• ▪Microsoft has held talks with Taiwan Semiconductor Manufacturing Company to secure production capacity for more than 300,000 Maia 300 chips for delivery in 2027.
• ▪Andrew Wall, general manager for Microsoft's Azure Maia, stated that reported production figures do not reflect the actual scale of Microsoft's custom silicon program.
• ▪Microsoft ultimately aims to secure capacity for more than 1 million Maia 300 chips, though component supplies and ongoing negotiations with TSMC could constrain these plans.

Azure AI infrastructure

• ▪Microsoft introduced its second-generation Maia 200 chip in January 2026, which was manufactured by TSMC using 3-nanometer technology and packed with high-speed SRAM.
• ▪Microsoft plans to use the custom Maia 300 chip to strengthen its artificial intelligence infrastructure and power services such as Azure and Copilot.

Rivalry with Google

• ▪Google began recognizing revenue from direct sales of its custom Tensor Processing Units in the quarter ended June 2026.
• ▪Microsoft has lagged behind Alphabet in scaling up its custom in-house chip efforts since first introducing its Maia lineup in November 2023.

Rivalry with Amazon

• ▪Amazon has experienced growing customer adoption of its in-house processors, including its custom Trainium artificial intelligence chips.
• ▪Microsoft's slow start to custom silicon development has left it trailing Amazon in the race to deploy proprietary hardware.

6 sources