Advanced AI Technologies

Microsoft announces Maia 300 AI accelerator

Microsoft announced Maia 300, a custom AI accelerator planned for multi-gigawatt deployment in 2027 and designed for higher token efficiency.

Microsoft announced Maia 300, a new generation of custom AI accelerator hardware, according to InformationWeek’s weekly technology review. The company plans to deploy the chips at multi-gigawatt scale in 2027.

Microsoft expects Maia 300 to deliver 40% more tokens per megawatt than leading GPU-based systems. The claim focuses on inference efficiency, an increasingly important consideration as AI services require large amounts of computing power.

Custom accelerators allow cloud providers to optimize hardware for particular AI workloads and reduce reliance on general-purpose GPU systems. Greater efficiency could help control operating costs and energy consumption as model usage grows.

The development matters because AI infrastructure is shifting beyond model software toward specialized chips and vertically integrated computing systems. The source presents Maia 300 as part of Microsoft’s broader effort to build capacity for large-scale AI deployment.

Original source: InformationWeek

Related RAQI solutions

In-depth analysis on RAQI Insight

Work with RAQI