API
< Back to all clusters
[TECHNOLOGY] · United States · 3 sources

started · updated

Microsoft announces Maia 300 AI accelerator

Microsoft has unveiled the Maia 300, the third generation of its proprietary AI accelerator, designed to handle memory-intensive models and complex logical reasoning. According to Scott Guthrie, Executive Vice President of Cloud + AI at Microsoft, the chip is intended to serve as the primary platform for AI inference within the company’s own products.

The hardware utilizes a modular chiplet architecture manufactured by TSMC using a 2-nanometer class process. The accelerator features 288 GB of HBM4 memory and achieves a memory bandwidth of 18 TB/s. Microsoft aims for a density of 96 accelerators per rack, which is expected to deliver up to seven times the token throughput per rack compared to the previous Maia 200 generation.

Production for use in Microsoft data centers is scheduled to begin in 2027. The company anticipates that the new hardware will provide approximately 40 percent more tokens per megawatt than leading GPU systems, focusing on high efficiency for AI workloads.

Entities

Dell Technologies · Microsoft · Scott Guthrie · TSMC