Microsoft’s ambitious push into custom artificial intelligence (AI) hardware has encountered a significant setback, with the company’s next-generation Maia AI chip—code-named Braga—now expected to enter mass production only in 2026. The delay, first reported by The Information, marks a six-month slip from the original 2025 timeline and could impact Microsoft’s broader AI infrastructure strategy.
According to sources familiar with the project, the delay stems from a combination of unanticipated design changes, staffing constraints, and high employee turnover within the chip development team. Microsoft had initially planned to deploy the Braga chip in its data centers this year to reduce reliance on Nvidia’s high-cost AI processors.
When Braga eventually enters production, it is reportedly expected to underperform compared to Nvidia’s Blackwell chip, which debuted in late 2024 and currently leads the market in AI processing power.
Microsoft introduced the Maia chip line in November 2023 as part of a broader initiative to develop in-house silicon tailored for AI workloads and general-purpose computing. The move mirrors similar efforts by cloud rivals Amazon and Google, both of which have made significant strides with their own custom chips—Trainium and Tensor Processing Units (TPUs), respectively.
While Microsoft has lagged behind in scaling up Maia’s production, its competitors have continued to advance. Google unveiled its seventh-generation TPU in April 2025, and Amazon is set to release its Trainium3 chip later this year.
The delay in Braga’s rollout may pose strategic challenges for Microsoft as it seeks to optimize performance and cost-efficiency across its Azure cloud platform. However, the company remains committed to its long-term vision of developing proprietary AI hardware to support its growing suite of AI services and applications.
Microsoft has not issued an official comment on the reported delay.
