Expanse (YC P26) Launches Tool to Boost GPU Cluster Utilization by Predicting Resource Needs
By
ismaeel_bashir
Hot, fresh, and worth queueing round the block for.
Summary
Expanse is a startup (YC P26) founded by Ismaeel, Eren, Yafet, and Nikodem that addresses the problem of low GPU/HPC cluster utilization (30-40%). Their solution analyzes source code, job submission scripts, and hardware to predict actual resource needs before workloads hit the cluster, flag potential failures, and surface line-level optimizations for researchers. The tool works with Kubernetes and SLURM schedulers to unlock wasted GPU capacity.
Key quotes
· 3 pulledWe read the source code, job submission script, and the hardware a workload is about to run on to predict what the job actually needs before the cluster sees it.
We also flag failures we think are about to happen and surface line-level optimisations the researcher can apply themselves.
Datacenters run at roughly 30% to 40% effective utilisation. Users request more resources than
You might also wanna read
Snowflake commits $6 billion to AWS Graviton CPUs and AI infrastructure over five years
Snowflake has announced a $6 billion, five-year commitment to use Amazon's custom Graviton CPUs and AI accelerators on AWS. The partnership
theregister.com·2d ago
AWS rebuilds OpenSearch Serverless from ground up to support AI agent workloads with zero-idle scaling
AWS has completely rebuilt its OpenSearch Serverless architecture to better support AI agent workloads, which have bursty usage patterns wit
bit.ly·3d ago
Anthropic partners with SpaceX to boost compute capacity, raises Claude usage limits
Anthropic has announced a partnership with SpaceX to substantially increase compute capacity, enabling higher usage limits for Claude Code a
anthropic.com·25d ago
Cloudflare enables AI agents to autonomously create accounts, buy domains, and deploy code
Cloudflare now allows AI coding agents to create accounts, purchase domains, set up paid subscriptions, and obtain API tokens autonomously o
blog.cloudflare.com·26d ago
General Compute Launches ASIC-Based Inference Cloud for Faster AI Agent Performance
General Compute is an inference cloud built on ASICs (purpose-built alternatives to Nvidia GPUs) designed specifically for AI inference, not
Product Hunt·1mo ago
IonRouter: OpenAI-Compatible API for AI Models at Half Market Rate
IonRouter is an OpenAI-compatible API service that allows teams to access various AI models (LLMs, vision, video, TTS) at half the market ra
Product Hunt·2mo ago