Launch HN: Expanse (YC P26) – Unlock Wasted GPU Capacity | Hacker News

YC-backed startup Expanse launched software for HPC and GPU clusters that predicts resource needs and failure risks before jobs run.

The tool analyzes source code, submission scripts, and telemetry to improve walltime prediction, detect likely failures, and suggest optimizations.

Expanse says many clusters are heavily over-provisioned, citing research showing significant wasted compute on a national HPC system.

The product also adds live observability and failure diagnosis for environments using SLURM, Kubernetes, DCGM, and CUPTI.