Skip to content
AI News HubLIVE
More
Summary only

Your Next AI Query May Travel Where the Power Is

Nvidia and partners plan to build about 25 micro data centers near utility substations in the U.S., each ranging from 5 to 20 megawatts, to dynamically shift computing tasks based on power availability. This 'distributed inference' strategy aims to increase grid flexibility and provide faster grid access for AI inference workloads.

SourceIEEE Spectrum AIAuthor: Dina Genkina
Your Next AI Query May Travel Where the Power Is
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions

Summary only

Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • Nvidia, EPRI, and partners will build 25 micro data centers (5-20 MW each) next to substations across 5 U.S. utilities.
  • Compute tasks can be shifted to substations with spare capacity during overloads or outages, requiring relocation only 0.1% of the time.
  • U.S. data centers could consume 9-17% of electricity by 2030; grid flexibility is crucial.
  • Inference workloads are dynamically routable, while training requires large centers but can pause briefly to reduce peak load.

Highlights and analysis are generated automatically and may contain errors. Check the original source.