Microsoft’s custom AI chip, internally known as “Braga” and officially named Maia, is reportedly facing production delays that will push its mass rollout to 2026—at least six months later than originally planned. According to The Information, which cites three people familiar with the matter, the company had initially targeted 2025 for large-scale deployment.
The delay reportedly stems from a combination of unexpected design changes, staffing limitations, and high employee turnover. Microsoft had planned to deploy the Maia chip in its data centers this year as part of a strategy to reduce reliance on Nvidia’s increasingly costly AI chips. However, those plans have now been postponed.
“Once production begins, the Braga chip is expected to underperform compared to Nvidia’s Blackwell series, which debuted in late 2024,” the report notes. This setback could place Microsoft at a disadvantage as rivals continue to advance their own chip development efforts.
The Maia chip was first introduced in November 2023, aimed at optimizing AI and general-purpose computing workloads across Microsoft’s cloud infrastructure. Like other major tech companies, Microsoft has been investing heavily in custom silicon to enhance performance and cut costs in its data centers.
Google and Amazon have already made significant progress in this area. Google launched its seventh-generation Tensor Processing Units (TPUs) in April 2025, which are being widely adopted across its services. Amazon, meanwhile, announced its next-generation Trainium3 chip in December 2024, with a launch expected later this year.
Despite unveiling Maia last year, Microsoft has yet to scale its production to meet growing demand or match the rollout pace of competitors. The reported delay could hinder the company’s goal of becoming more self-reliant in AI hardware and competing more directly with Nvidia.
While Microsoft continues to develop its in-house chip technology, the postponement of Maia’s mass production signals the challenges of building competitive silicon in a fast-evolving AI landscape.
