Microsoft's Maia 300 AI Chip Promises 40% Increased Token Efficiency per Megawatt

Buzzline

At the OCP Global Summit held in San Jose, Microsoft introduced its Azure Maia 300 AI chip, aiming to lessen its dependency on Nvidia and AMD processors. The company asserts that the Maia 300 achieves a 40% increase in tokens per megawatt compared to top GPU systems, with a 96-chip configuration providing sevenfold the output of the Maia 200. The Maia 200 is already utilized in Microsoft 365 Copilot and OpenAI models. Microsoft intends to initiate multigigawatt rollouts beginning next year. Industry experts suggest that hyperscale operators will likely continue to procure Nvidia chips, which remain highly sought after. Disclosure: This video contains stock footage and content created or enhanced using AI-assisted tools.  ai-generated

Transcript

00:00Microsoft just unveiled a new AI chip, and it is aimed squarely at NVIDIA.

00:04At the OCP Global Summit in San Jose, Microsoft revealed the Azure Maya 300.

00:09As it works to reduce its reliance on NVIDIA and AMD,

00:13Microsoft claims the Maya 300 delivers 40% more tokens per megawatt than leading GPU-based systems,

00:20and says a 96-chip rack offers seven times the throughput of the Maya 200.

00:25The Maya 200 already powers Microsoft 365 Copilot and OpenAI models.

00:31Microsoft plans multi-gigawatt deployments beginning next year.

00:35Still, analysts note big cloud companies will keep buying NVIDIA chips, which remain the most in demand.

00:41The performance numbers come from Microsoft.

00:44Disclosure. This video contains stock footage and content created or enhanced using AI-assisted tools.

Category

Transcript

Recommended