Microsoft announced its third-generation Maia 300 AI chip, designed to become the primary inference system for its services by 2027. The company says the chip offers 40% more efficiency than leading GPU systems.
Microsoft has announced the Maia 300, its third-generation in-house artificial intelligence chip. The company states that the new hardware is designed to serve as the "primary AI inference system" for Microsoft’s products and services, a move intended to reduce dependence on external suppliers like Nvidia and AMD. Production rollouts for the Maia 300 are scheduled to begin in 2027.
According to Microsoft, the Maia 300 is built to improve the cost and efficiency of running AI workloads in the cloud. Scott Guthrie, executive vice president of Microsoft’s cloud group, stated that the company expects the chip to "deliver 40 percent more tokens per megawatt compared to leading GPU-based systems." Guthrie added that in a full 96-accelerator configuration, the new system "delivers up to 7x the rack-scale token throughput of Maia 200."
The chip utilizes a new chiplet architecture built on TSMC’s N2P process node. This design separates compute, memory controller, and I/O functions. "This increases the silicon area dedicated to compute performance and allows each function to use the technology best suited to its role," said the Azure hardware and infrastructure team. Microsoft notes that the chip provides memory bandwidth of 18 terabytes per second.
Microsoft CEO Satya Nadella described the launch as "our biggest bet yet on our own silicon" in a post on X. The company plans to scale its Azure Maia family to multi-gigawatt deployments starting in 2027. This expansion aims to make Maia the primary system for handling AI requests across Microsoft's ecosystem, including Microsoft Copilot and GitHub Copilot.
Mustafa Suleyman, CEO of Microsoft AI, emphasized the need for vertical integration. "We have to work across the full stack, all the way down to the silicon," Suleyman said. "At that scale we need a platform that we own, and that we co-design with our models. That’s why running our Microsoft AI models on top of the Maia chip is a critical focus for our team."
The Maia 300 succeeds the Maia 200, which was announced in January 2026. The Maia 200 is currently running on production AI workloads, including Microsoft Copilot and Microsoft’s own AI models. According to Microsoft, Maia 200 is being deployed across the U.S., Europe, and Australia, with GitHub Copilot and Microsoft Foundry set to move onto the platform by the end of the year.
Despite the push for in-house hardware, Microsoft maintains its partnership with Nvidia. Nadella and Nvidia CEO Jensen Huang recently shared the stage to launch Windows PCs built on Nvidia chips. Microsoft also plans to deploy next-generation systems based on Nvidia’s Vera Rubin and AMD’s Helios. However, the shift to Maia reflects a broader industry trend where tech giants like Google, Amazon, and Meta are developing custom AI processors to control costs and supply chains.
The introduction of the Maia 300 signals Microsoft’s intent to gain greater control over its AI infrastructure costs and performance. By designing its own chips, Microsoft aims to address the growing energy demands of AI inference, particularly for agentic tasks and long-context reasoning. While Nvidia remains a key partner for training and other workloads, the scaling of Maia to multi-gigawatt levels by 2027 could significantly alter the balance of power in the AI semiconductor market, reducing Microsoft's reliance on third-party GPUs for its core services.
Sources
This article was drafted with AI assistance and checked against the sources above. Company claims are reported as claims. Cover image is AI-generated.
The day's checked news, by email
One short email a day with the top stories. Free, and you can unsubscribe in one click.
Or follow for the latest news:
IMF Managing Director Kristalina Georgieva stated that AI is expected to impact approximately 40% of jobs globally. She noted that while AI supports economic growth, it may augment, downgrade, or eliminate positions.
Anthropic has launched an expanded Cyber Verification Program with three access tiers for security professionals, integrating previous programs to provide advanced cyber capabilities.
