Expanse raises $5.3M to predict AI workloads before they run
What's the deal? Expanse, an AI infrastructure startup based in San Francisco and London, has raised $5.3 million in a seed round led by Crane Venture PartnersDealroom has a profile for this one. Try Dealroom →. PXN VenturesDealroom has a profile for this one. Try Dealroom → and angel investors, including former DeepMindDealroom has a profile for this one. Try Dealroom → researchers, also participated.
What's the endgame? Expanse's software predicts what an AI workload needs before it runs, recommending the optimal GPU, CPU, memory, and runtime configuration before resources are committed. The aim is to help organisations complete more work using infrastructure they already own, cutting failed jobs and unnecessary compute spend.
How does it work? Unlike monitoring tools that explain what happened after a workload finishes, Expanse acts before execution. It installs into a customer's cloud or on-premises infrastructure, and its models analyse workloads without code or telemetry leaving that environment.
Why now? GPUs are among the world's most constrained resources, yet much capacity sits idle. Industry estimates suggest about 30% of cloud spend is lost to over-allocation, while Microsoft ResearchDealroom has a profile for this one. Try Dealroom → has reported GPU utilisation of around 50% across many internal deep learning workloads.
New GPU capacity can take months to procure, and power, cooling, and planning limits make expanding data centres harder. "The machine should carry the uncertainty, not the person," said co-founder and chief executive officer Ismaeel Bashir.
Who's behind it? Expanse was founded by four University of EdinburghDealroom has a profile for this one. Try Dealroom → graduate engineers who previously built large-scale compute infrastructure in quantitative finance and national supercomputing. At the Edinburgh Parallel Computing CentreDealroom has a profile for this one. Try Dealroom →, Bashir developed a resource prediction system for high-performance computing workloads; co-founders are Nikodem Bieniek, Eren Mendi, and Yafet Melake.
In one production deployment, Expanse identified nearly $8 million of idle compute capacity in a single month.
The signal: The round lands as the industry shifts focus from building bigger clusters to using existing ones better. "The future of AI won't be defined only by who builds the biggest clusters, but by who uses them most intelligently," said Scott Sage, co-founder and partner at Crane Venture Partners.
Read more: Business Insider
Image credit: Expanse
