Key Highlights

  • Microsoft is targeting a public unveiling of its next-generation Maia 300 AI accelerator as soon as September.
  • Microsoft has entered talks with TSMC to secure Manufacturing capacity for more than 300,000 Maia 300 units for delivery in 2027, with a longer-term ambition of more than one million units.
  • The scale of the order marks a sharp increase from the tens of thousands of Maia 200 chips produced to date, following a troubled rollout for Microsoft's earlier chip generation.
  • Microsoft is working to persuade major cloud customers, including Anthropic, to adopt the Maia 300, even as Anthropic has separately confirmed it is forming its own internal team to design custom chips for its models.

Microsoft Targets September Reveal for Next-Generation AI Chip

Microsoft ( Nasdaq :MSFT) is targeting a public unveiling of its next-generation Maia 300 AI accelerator as soon as September, according to a report citing people with direct knowledge of the plans. Microsoft has entered talks with Taiwan Semiconductor Manufacturing (NYSE:TSM) to secure manufacturing capacity for more than 300,000 units, with delivery planned for 2027, an order that would substantially exceed the tens of thousands of Maia 200 chips produced to date. Microsoft's longer-term ambition is capacity for more than one million units, though component supplies and ongoing packaging negotiations could constrain that target. Andrew Wall, general manager for Microsoft's Azure Maia program, said Microsoft "continues to invest in custom silicon as part of our long-term AI infrastructure strategy," adding that reported production figures "don't reflect the scale of our program."

Maia 300 Marks a Turn From Troubled Predecessor

The scale of Microsoft's ambition marks a sharp shift from the Maia program's recent history. The Maia 200 was delayed after early tests fell short of internal targets and has since been deployed in only a small number of data centers. Microsoft Chief Executive Officer Satya Nadella told investors on the company's fiscal fourth-quarter Earnings Call on July 29th that the chip delivers 30% better performance per dollar compared with existing hardware and is scaling to support both OpenAI and Microsoft's own MAI models. The Maia 200 was built by TSMC using 3-nanometer technology and incorporates a significant amount of SRAM memory, which can provide speed advantages for AI systems handling large volumes of user requests.

Reducing Nvidia Dependence Remains a Stated Priority

Reducing dependence on Nvidia's processors is a stated priority for Nadella, with the Maia line central to that effort. Microsoft believes its custom chips can run both its own models and OpenAI's models at lower cost, and the company is expanding internal usage through Azure AI Foundry and Copilot while pitching the technology to large external cloud customers, including Anthropic. That effort carries a complication: Anthropic confirmed earlier this month that it is forming its own internal semiconductor team to design custom chips for its Claude models, making the AI company simultaneously a potential Maia 300 customer and an emerging long-term competitor in custom silicon. Microsoft's approach mirrors a broader industry trend, with Google beginning to recognize Revenue from direct sales of its custom Tensor Processing Units in the quarter ended June, and Amazon seeing growing adoption of its own Trainium chips.

Execution Risks Remain as Broader Financial Backdrop Supports the Push

Microsoft's fiscal fourth-quarter revenue reached $90.0 billion, up 18% year-over-year, with Azure and other cloud services growing 43%, giving the company financial room to pursue the expanded chip strategy. Still, execution risks remain: projects concentrated on TSMC's N3 manufacturing process and CoWoS advanced chip-packaging technology face Supply tightness through 2027, a constraint directly relevant to Microsoft's Maia 300 ramp. The specific process node and packaging configuration Microsoft is pursuing have not been publicly confirmed, and the September reveal timeline has not been officially announced by the company.