Committed Use Contracts
Definition
Committed use contracts are agreements where a customer commits to pay for a defined amount of compute capacity, usage, or spend over a specified term.
Why it matters
Committed use can turn volatile spot GPU demand into financeable revenue, but only if the customer is creditworthy and the contract is enforceable, assignable, and matched to available capacity.
Common misconceptions
- •A forecast, reservation, letter of intent, or framework agreement is not the same as an unconditional minimum-payment obligation.
- •Contracted revenue is not risk-free when termination, service-level credits, capacity delays, or customer insolvency can reduce collections.
- •Nominal contract value can overstate backlog if it includes optional usage, unexercised renewals, pass-through power, or ramp periods.
- •Take-or-pay language does not protect the provider when it cannot deliver the contracted hardware, uptime, network, or location.
Technical details
Key terms
Review minimum spend, term, ramp schedule, take-or-pay language, service-level obligations, termination rights, cure periods, and customer credit.
A reservation or memorandum of understanding is not the same as a binding take-or-pay contract.
The provider must have enough power, cooling, network, and hardware delivery certainty to meet the commitment.
Financing impact
Lenders and investors may lend against contracted revenue, but they should haircut contracts with weak termination protections, unproven customers, usage-only pricing, or capacity dependencies outside the provider's control.
Backlog normalization
Separate binding minimum spend from variable usage, options, renewals, setup fees, taxes, and pass-through power. Build a monthly revenue schedule incorporating ramp, free periods, credits, and termination dates. Compare contracted capacity with installed and deliverable capacity so the same GPU or megawatt is not promised to multiple customers.
Credit and termination analysis
Underwrite the paying entity, guarantor, security deposit, parent support, and concentration rather than the customer's brand alone. Review termination for convenience, insolvency, delayed delivery, repeated SLA failure, change in law, export restrictions, and force majeure. Model both lost revenue and stranded hardware after termination.
Asset-liability matching
Match contract tenor and cash collections against hardware debt, leases, power commitments, and facility obligations. A two-year customer commitment may not support five-year financing if residual utilization and resale value are uncertain. Check assignment rights and lender step-in provisions so contracts remain useful after an operator default or foreclosure.
Capacity-to-revenue bridge
For committed use contracts, bridge physical capacity to billable revenue. Start with contracted or announced units, then deduct capacity not yet delivered, powered, cooled, networked, commissioned, accepted by customers, or available after redundancy and maintenance requirements.
Build a monthly schedule for installed capacity, usable capacity, committed capacity, billed capacity, and collected revenue. This prevents double-counting the same GPU, rack, or megawatt across marketing pipeline, financing collateral, and customer backlog.
Separate high-margin infrastructure revenue from pass-through power, setup fees, burst usage, credits, taxes, and reimbursed costs. Revenue quality depends on margin, duration, collectability, and renewal probability, not only gross contract value.
Contract and counterparty diligence
Review the exact contracting party, guarantor, minimum commitment, ramp schedule, delivery conditions, service levels, termination rights, cure periods, force majeure, assignment rights, deposits, and lender step-in rights.
Customer quality matters because AI demand can be volatile. Underwrite concentration, funding runway, payment history, use case, workload portability, and whether the customer can switch to hyperscalers or newer hardware.
Supplier diligence should cover title transfer, liens, serial-number evidence, warranty, replacement rights, export controls, delivery delay remedies, and whether a reseller actually controls the inventory it promises.
Operating constraints and cost stack
AI compute economics are constrained by power price, power availability, cooling design, rack density, network fabric, facility uptime, maintenance, software orchestration, spare parts, and labor. A GPU fleet can be technically installed but commercially weak if one of these constraints binds.
Stress power-price increases, curtailment, delayed interconnection, transformer lead times, cooling retrofits, customer credits, lower utilization, and hardware failures. Compare gross utilization with contribution margin after power and operating costs.
For financing, match customer contract tenor and hardware useful life to debt amortization. A long loan against short-lived or rapidly repricing hardware can leave residual-value risk with the lender or vehicle.
Refresh, residual value, and monitoring
Track hardware by cohort: model, purchase date, installed cost, memory profile, networking, warranty, utilization, average realized rate, power draw, and expected resale or redeployment value.
Monitor competitive GPU pricing, new chip launches, customer workload shifts, inference versus training mix, cloud spot pricing, and resale market depth. A unit that still functions can become economically stale before physical failure.
Warning signs include revenue booked before acceptance, unclear ownership of hardware, repeated delivery delays, rising service credits, power constraints, low realized utilization, customer nonpayment, and capex needs that are not reflected in the financing model.