Microsoft’s push to develop proprietary artificial intelligence chips has encountered a significant setback, as its next-generation Maia AI chip, codenamed Braga, is facing a production delay of at least six months, reported Information.
This development pushes the anticipated mass production timeline from 2025 to 2026.
The delay could have implications for Microsoft’s strategy to reduce its reliance on external chip suppliers and strengthen its position in the fiercely competitive AI landscape.
Production setbacks and performance expectations
The delay of the Braga chip’s mass production into 2026 marks a notable shift from Microsoft’s initial aspirations.
The company had reportedly hoped to integrate the Braga chip into its data centers as early as this year.
However, a confluence of factors is contributing to these unforeseen delays.
The report indicates that unanticipated changes to the chip’s design, along with staffing constraints and a high rate of employee turnover within the project, are key contributors to the revised timeline.
Compounding the challenge, sources suggest that when the Braga chip eventually enters production, its performance is expected to fall considerably short of Nvidia‘s recently released Blackwell chip, which debuted last year.
This performance gap could present further strategic considerations for Microsoft as it aims to optimize its AI infrastructure.
The drive for custom AI chips
Microsoft’s investment in developing custom processors like Maia underscores a broader trend among major technology companies.
Akin to its Big Tech counterparts, Microsoft has heavily prioritized the creation of in-house chips specifically designed for artificial intelligence operations and various general-purpose applications.
The primary motivation behind this strategic pivot is to significantly reduce the tech giant’s dependence on costly chips supplied by third-party manufacturers, particularly high-demand and expensive offerings from Nvidia.
By designing its own silicon, Microsoft seeks to gain greater control over its hardware stack, potentially leading to improved performance tailored to its unique software and cloud services, along with substantial cost efficiencies in the long run.
Peer progress in custom silicon
While Microsoft navigates these development hurdles, its cloud rivals, Amazon and Alphabet’s Google, have been actively advancing their own custom chip initiatives.
Google, for instance, has achieved notable success with its custom AI chips, known as Tensor Processing Units (TPUs).
In April, Google further showcased its commitment to in-house silicon by unveiling its seventh-generation AI chip, specifically engineered to accelerate the performance of sophisticated AI applications.
Similarly, Amazon, another major player in the cloud computing space, unveiled its next-generation AI chip, Trainium3, in December, with its release slated for later this year.
These advancements by competitors highlight the intense race to innovate in custom AI hardware and underscore the pressure on Microsoft to accelerate its own development and deployment efforts to avoid falling further behind in the custom chip ecosystem.
The post Microsoft’s new AI chip delayed to 2026: report appeared first on Invezz