The loan runs for years.
A lender finances the GPUs over three years and expects a payment every month.
GPUs financed on what they earn.
We put financed GPUs to work, measure what each one earns,
and run them when the borrower cannot.
Loans run for years.
Customers sign for months.
How Valnce worksCustomers sign for months.
Compute buyers sign for six to twelve months. Between contracts, and after the last one, nobody is paying for the GPUs. So lenders only finance operators with three-year contracts from big companies.
Valnce sells every hour.
We sell the idle months as inference, and run the GPUs ourselves if a contract ends or the borrower defaults. Every minute is logged, so the lender sees what the hardware earns.
How we make GPUs financeable.
We put financed GPUs to work, record what each one earns, and run them if the borrower cannot. The lender gets a record to underwrite against, not only a contract.
01 / Work
Put idle hours to work.
Between contracts and inside the gaps, our router sells the hours as open-model inference. Nothing sits at zero while the loan runs.
A record built for
credit committees.
Talk to usPer GPU, per minute.
Every minute we run is logged against the card that ran it, with what it earned.
Every fleet adds to it.
Each loan we work on extends the history, across models, ages, and sites.
Built for the lender.
Inference providers report to developers. We report to the people who financed the hardware.
Lenders finance the contract because nobody shows them what the hardware earns. We are building the record every GPU loan is underwritten on.