Key Takeaways
- Microsoft plans to deploy its own Maia chips as the primary hardware for running AI products starting in 2027.
- The third generation of Maia chips, Maia 300, is optimized for inference tasks and is expected to generate 40% more tokens per megawatt.
- Each Maia 300 rack contains 96 chips and can deliver up to seven times the output of Maia 200.
Microsoft has announced plans to use its custom Maia chips as the primary hardware for running AI products, starting in 2027. The company unveiled the third generation of its custom AI accelerator, Maia 300, which is designed to optimize inference tasks.
Built for AI inference, Maia 300 is particularly suited for long conversations, AI reasoning, and multi-step tasks. The chip is expected to generate 40% more tokens per megawatt of power compared to leading GPU-based systems, according to Microsoft.
Each Maia 300 rack will contain 96 chips and consume around 167 kilowatts of power, making it a denser system in terms of accelerator count. For comparison, Nvidia’s GB300 racks contain 72 GPUs and draw around 142 kilowatts, highlighting the efficiency of Maia 300.
The Maia 300 is part of a broader industry shift among large cloud providers and AI companies to develop their own chips. Microsoft is following a similar path to companies such as Google and Amazon, which already operate custom AI accelerators. OpenAI has also been developing its own AI chip with Broadcom.
Microsoft is working to secure enough manufacturing capacity for the planned rollout. Earlier reports indicated that the company was in discussions with TSMC to manufacture more than 300,000 Maia 300 chips for 2027. The company now says its deployment plans will be measured in gigawatts, suggesting eventual volumes could be substantially higher.
The Maia 300 is already being used inside Microsoft, running workloads for Microsoft Copilot and supporting both OpenAI models and Microsoft’s own MAI models. Microsoft plans to use Maia 200 for GitHub Copilot and Microsoft Foundry by the end of the year.
By developing its own chips, Microsoft aims to reduce reliance on Nvidia and gain greater control over cost, power consumption, and hardware availability as demand for AI computing continues to rise. The strategy could provide Microsoft with a competitive edge in the rapidly growing AI market.
