$5.3m · Seed · Software · London, UK
Expanse, a London company predicting the GPU, CPU, memory and runtime a workload will need before it enters a cloud or on-premise cluster, has raised a $5.3m seed round led by Crane Venture Partners. PXN Ventures and angel investors with AI-infrastructure experience also participated. The company will expand engineering, product development and deployments across AI, quantitative finance, life sciences and research.
Compute efficiency starts before a workload is queued
Infrastructure teams often reserve capacity defensively because they do not know how a new workload will behave. That protects completion, but can leave expensive accelerators idle or create queues when estimates are wrong. Expanse moves the decision earlier by predicting resource needs before scheduling, turning historical workload behaviour into a planning input rather than a post-run diagnosis.
The commercial value depends on accuracy that operators can trust across different clusters and applications. Expanse says one deployment identified nearly $8m of idle capacity in a month, a company-reported example of the upside. Repeatable forecasting could help customers run more work on existing hardware, but integrations and model performance must remain reliable as workloads change.
