Microsoft unveils Maia 300 and plans gigawatt-scale AI deployments in 2027

First reported by TechRadar at · Updated · 11 sources

The third-generation accelerator is set to enter production deployments in 2027 as Microsoft expands its in-house AI infrastructure. The company plans to make the Maia family its primary system for inference across its products and services, using its own chips to handle requests to AI models.

  • Azure Maia 200 is already in production, Microsoft says.
  • Maia 300 uses TSMC’s N2P technology, according to DigiTimes.

Covered by 11 publishers within 7 hours of the first report: 10 news reports and 1 from the company.

Reporting10

Unite.AI Microsoft Announces Maia 300 AI Chip, Multi-Gigawatt Azure Scale-Up · Theo Nash, AI Infrastructure & Compute, AI Research Agent
The Next Web Microsoft says Maia 300 chip will run its AI services from 2027 · Ana Maria Constantin
HPCwire Microsoft Announces Maia 300, Its Next-Gen Custom AI Chip · Andrew Jolly
SiliconANGLE Microsoft debuts multichiplet Maia 300 inference accelerator · Maria Deutscher

From the company1

Discussion

Hacker News 7 points, 3 comments

Related

Topics MicrosoftTSMC