Microsoft announces the third generation of its proprietary AI chip: the Maia 300 delivers 40% more tokens per megawatt than leading GPU-based systems, according to the company. Its predecessor, the Maia 200, is already used in applications such as Microsoft 365, Teams, and Copilot. The new generation will begin rolling out in 2027, and Microsoft aims to make the Maia line the primary inference platform for its models. The information comes from Yahoo Finance.