Microsoft is introducing its third-generation Maia 300 chip as the primary engine for AI inference in the company’s products and services. Multiple outlets report that Microsoft plans to use the homegrown silicon to handle incoming AI model requests, starting with this newer Maia 300 generation.
The coverage frames the move as a way to reduce reliance on third-party AI accelerators, particularly Nvidia and AMD GPUs. Microsoft positions Maia 300 as a more efficient and capable option for running the latest AI workloads, including reasoning-oriented models and agent-like tasks. Other outlets highlight expected benefits for end users through improved performance and efficiency in services such as Teams, Copilot, and Copilot-related features.
While the articles differ in emphasis—some focus on independence from specific suppliers, others on product impact and technical capability—they align on the core announcement: Maia 300 is positioned as the default inference hardware for Microsoft’s AI at scale, with benefits tied to cost, efficiency, and workload fit.
