Skip to content
Trending storyDeveloping

AI inference's hidden storage burden

1 reports1 sourcesUpdated 3 hours ago

What you need to know

AI summary

WD measured that an 8-second 720p video clip, which produces a 3.97MB output, generates more than 500MB of back-end logs during AI inference. The company says this means inference leaves far more data behind than users see. The figure covers only that one clip; the coverage does not specify other workloads or formats.

Generated by AI from the reporting · updated 2 hours ago

Timeline

Follow the reports to see every side of the story.

Oct 22
  1. MIT Technology Review · AI
    AI's hidden storage bill: What inference leaves behind

    Inference leaves far more back-end data than users see—WD measured an 8-second 720p clip (3.97MB output) generating over 500MB of logs.

Heat trend for this story

Not enough continuous data yet to chart a trend.