STORY · PRODUKTER_
Expanse: Startup aims to recover wasted GPU capacity with AI predictions
Expanse, a YC-backed startup, has built tools using machine learning to predict how many resources datacenter jobs actually need before they start. The team measured that 59% of compute capacity on a national HPC cluster was wasted because users requested 2-3 times more resources than necessary—the calibration is based on analyzing job code, submission scripts, and hardware telemetry.
WHY IT MATTERS
With 30-40% effective utilization in datacenters, this represents potential billion-dollar losses annually for the industry. A tool that can reduce over-requesting will free up significant capacity, lower costs, and make AI training and HPC work more efficient at global scale.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.