From equipment quote to collateral schedule
Reconcile price, configuration and ownership before applying an advance rate to the invoice total.
Separate the eligible hardware
Illustrative eligible advance
8.00m eligible hardware + 2.00m other quote costs.
Assumptions & calculation
Advance = eligible hardware × advance percentage. Total project quote fixed at 10m; remaining quote costs and any unfunded hardware require other funding. Eligibility is an assumption, not a market rule.
Original InfraCompute hypothetical model. All monetary inputs use one consistent, unspecified currency; m means millions. Values are not market quotes, appraisals or recommended financing terms. Inputs are not saved.
Reconcile the quote before valuing it
A GPU project quote can combine physical equipment, configuration detail and services. Begin by identifying which lines carry price and which merely describe components already included elsewhere. Reconcile quantities and extensions to the supplier total. An eight-GPU server followed by eight unpriced GPU configuration entries is one system, not two sets of financed accelerators.
Next separate the equipment purchase from deployment, support, software and other charges. Some of those costs are essential to the project while having little independent resale value. The financing agreement may permit them to be funded, but that does not make them equivalent collateral. Preserve both the project-cost view and the recoverable-asset view.
Describe what a buyer would receive
Record the system manufacturer, model, GPU form factor and memory, network fabric, storage, cooling and support entitlement. Confirm whether the intended cluster includes management and storage nodes as well as GPU servers. A count of accelerators cannot establish that the network and supporting equipment are sufficient for the customer's workload.
At delivery, map serial numbers and actual configurations back to the approved order. If the supplier substitutes components, record the engineering acceptance, revised price and effect on the equipment schedule. A credit approval against one configuration should not silently become funding for another.
Use a consistent value base
American Compute's BOM guide reports GPUs at 76–82% of hardware value in the cluster quotations it analysed at 22 nodes and above. That is a source sample finding, not a universal allocation rule. Its denominator is hardware value; applying the percentage to an invoice that includes substantial services would change the meaning.
Compare like-for-like quotes and independent price evidence for the specified configuration. Treat list prices, reseller offers and completed transactions as different evidence types. Differences can reflect warranty, location, lead time, installation or credit terms as well as equipment value. An apparent discount is difficult to evaluate until those differences are reconciled.
Expose the financing consequence
Suppose a hypothetical quote totals ten million, comprising eight million of eligible hardware and two million of services and other costs. A 70% advance against eligible hardware supplies 5.6 million. Applying the same percentage to the whole quote would produce seven million. The 1.4 million difference is a funding-base decision, not extra collateral.
Show that decision explicitly in the funding plan and security schedule. The borrower can then identify the cash needed for excluded costs, and the lender can explain the basis of its exposure. Revisit the reconciliation at every material amendment and before the final draw.
Questions for the credit paper
- Which lines are priced, included or optional?
- What part of the invoice is eligible hardware?
- Can the delivered system be matched to the approved order?
- Do the price comparisons include the same support and delivery terms?
About this guide
Original InfraCompute analysis informed by the American Compute material below. The article structure, examples and interactive visuals are InfraCompute’s work. Source-specific findings remain attributed to American Compute; attribution does not imply affiliation or endorsement.
- American Compute, How to Read the BOM ↗
Bill-of-materials guide, August 2026. Source configurations and indicative costs are examples.
We reviewed the published edition and selected public references. We have not independently verified American Compute’s proprietary datasets, every market claim or confidential financing terms. Frameworks and hypothetical models describe questions to investigate; they do not establish protections in a particular financing. Legal and engineering conclusions depend on the relevant documents, system and jurisdiction.
Read our research methodology →