Microsoft is announcing its third-generation Maia in-house AI chip, the Maia 300, which the company describes as the primary AI inference system for its products and services. The new hardware is designed to improve the cost and efficiency of running AI workloads in the cloud while handling the latest reasoning models and agentic tasks. Scott Guthrie, executive vice president of Microsoft's cloud group, stated that the company expects the Maia 300 to deliver 40 percent more tokens per megawatt compared to leading GPU-based systems. This efficiency gain is part of a broader design philosophy focused on optimizing performance within power constraints. The rollout of the Maia 300 to production is scheduled to begin in 2027. By developing its own silicon, Microsoft aims to reduce its dependence on external suppliers like Nvidia for its core AI infrastructure.