At its Ignite 2026 event, Microsoft announced that Maia 200—its second-generation in-house AI accelerator—has officially entered large-scale production and is now running the company's largest frontier AI models in its data centers. Designed for AI inference and built on TSMC's 3nm process, the chip delivers up to 1.7x higher token throughput per watt than leading GPU infrastructure in built-in AI features such as Teams and Copilot. It is currently deployed in the United States, Europe, and Australia to handle related production workloads, with support for more products to follow. Microsoft plans to scale its Azure Maia systems to multi-gigawatt-scale deployment by 2027, positioning them as the primary inference infrastructure for its products and services.
