Microsoft unveiled its third-generation in-house AI chip, Maia 300, and plans to install multiple gigawatts of it for inference. Reports indicate deployment begins around 2027, with the chip aimed at reducing reliance on Nvidia GPUs.

Microsoft will install more than one million of its latest-generation AI chips in its data centres as it works to reduce its reliance on Nvidia.

Starting next year, the company plans to begin deploying multiple gigawatts of its Maia 300 chip, the latest version of its custom-designed processor, to run AI inference workloads for customers. Microsoft aims to make its data centres more efficient and to reduce some of the costs associated with Nvidia GPUs, which have become the industry standard. The company expects Maia 300 to deliver 40% more tokens per megawatt than leading GPU-based systems.

Microsoft, Amazon and Google are all developing their own AI chips to become more self-sufficient and reduce costs. Microsoft introduced its first-generation chip in 2023 and an upgraded version in January this year. Production of Maia 300 is expected to begin in 2027.

"As we move into the era of Super Intelligence, Microsoft is entering a new phase of infrastructure," the company said.