Microsoft’s ambitions to dethrone NVIDIA in the AI chip market are facing significant hurdles. The tech giant’s first in-house AI chip, code-named ‘Braga,’ is encountering delays, and its performance is underwhelming, as initial tests reveal it lags behind NVIDIA’s existing hardware. 
Originally slated for production by 2025, Braga’s mass production is now set for 2026, marking a major setback of up to six years.
The chip project, which is part of Microsoft’s broader push to reduce its reliance on NVIDIA’s in-demand GPUs, had aimed to compete directly with NVIDIA’s highly successful AI chips. However, Braga’s performance is falling short of expectations. Reports indicate that it performs at a lower level than NVIDIA’s Blackwell architecture, which launched in 2024, raising doubts about Microsoft’s ability to catch up with Team Green’s dominance.
Despite pouring significant financial and technical resources into this project, Microsoft is struggling to meet its original goals. In-house chip design, while attractive to tech companies seeking to reduce dependency on third-party suppliers, requires deep expertise. The failure to deliver a competitive product thus far highlights the challenges even Big Tech faces in this high-stakes race.
Microsoft’s chip project isn’t alone in the quest for custom AI hardware. Competitors like Google and Amazon have already developed in-house chips for their data centers, but NVIDIA continues to maintain a strong lead. NVIDIA’s CEO, Jensen Huang, has downplayed the threat posed by custom silicon, pointing out that his company is already meeting the demand with better products. As the AI race slows down, Microsoft will need to rethink its strategy and approach to chip development if it hopes to remain competitive in the long run.