Take-or-pay contract
A take-or-pay contract obliges the buyer to pay for a committed quantity whether or not it uses it.
In compute, the customer pays the full reserved fee for dedicated GPUs even if they sit idle.
Three ways compute is sold
| Take-or-pay (reserved) | Full fee for dedicated GPUs; lenders finance against it |
| Committed spend (credits) | Prepaid credits used as needed, unused credits expire; a weaker repayment source than reserved |
| On-demand | Pays for use only; not a repayment source |
From our review of SEC-filed compute MSAs in compute offtake agreements.
How it works
Pay for the reservation, not the use. The customer reserves dedicated GPUs for a term and pays the full fee whether or not it runs workloads. In compute this is the reserved instance; bulk credits are committed spend; on-demand is neither (compute offtake agreements).
Public examples sit under rated debt: CoreWeave’s DDTL 4.0 is backed by a six-year take-or-pay contract with a customer reported to be Meta, and Lambda’s rated financing by take-or-pay capacity for NVIDIA at a single site in Quebec (Rated GPU Financings).
The obligation is only as strong as the contract around it. Termination for convenience, SLA exit triggers, cure periods and change of control can each turn take-or-pay into take-or-leave.
What lenders check
It makes debt service predictable: the debt service coverage ratio is computed on the contracted fee, not on expected utilization.
It does not survive a bankruptcy on its own. An offtaker that files can reject the contract, leaving the borrower an unsecured claim for damages (GPU loan default playbook). Letters of credit and parent guarantees cover part of that gap; lost revenue cover pays part of the missed revenue for 30 days in a qualifying interruption, time to collect or re-sign.
Discuss a transaction
Send us the equipment list and the loan or lease you have in mind. We will come back with an indication and the questions we would need answered.
Prefer email? hello@amcompute.com