The GPU Financing Professional
Lending Against AI Compute: Delayed Draw Facilities, Residual Values, Capacity Contracts, Collateral and Structures
-
- $12.99
-
- $12.99
Publisher Description
On Monday 16 March 2026, the first $75,000,000 leaves the lenders' accounts. It is drawn against 256 servers that have just passed acceptance in a leased colocation hall near Dallas. Nobody is worried that the GPUs will not work. The question in the credit file is simpler and harder: who will pay for these machines, month after month, for 48 months, and what will the cluster fetch if the customer stops?
The GPU Financing Professional is a practical handbook to lending against GPUs and AI compute capacity in the United States, built around one fictional financing followed from the first draw to the last installment: a $150,000,000 delayed draw term loan to a special purpose vehicle that buys a cluster of 4,096 GPUs and rents it for 48 months to an investment-grade customer. It covers why GPUs became collateral and how large the market grew, what a cluster is worth and how fast it falls, sponsors, anchor customers and supplier backstops, the capacity contract (take-or-pay, service credits, termination for convenience, assignment and step-in), delayed draw mechanics and the borrowing base, sizing on cash flow and debt service coverage, residual value scenarios and loan to value over time, Article 9 collateral, fixtures and landlord access, useful lives, impairment and bonus depreciation, the choice between a term loan, asset-backed securities, a sale-leaseback and vendor support, export controls and power, monitoring and covenants, and termination, default and remarketing, as of November 2026. Every figure of the case is computed by a model.
What is inside:
- 13 chapters that follow one financing from the first draw to the last installment
- One complete case: a fictional sponsor, borrower, customer and lender, with every number recomputed
- Three residual value scenarios and the loan to value curve month by month
- 8 ready-to-use tools, from the term sheet checklist to the default playbook
- 25 questions credit committees ask, with plain answers
- A glossary of 50 terms
Written for analysts and originators at private credit funds, banks and equipment finance companies, finance teams at GPU cloud providers, and the lawyers and valuers who work on these loans. The case, its people, entities and figures are fictional teaching assumptions, not market data. Rules and market figures are paraphrased as of November 2026, and proposals are identified as proposals. This book is for education only. It is not investment, legal, tax or accounting advice and does not replace professional advice on your own financing. It is not endorsed by or affiliated with any lender, GPU maker, cloud provider or regulator.
How this book was made: written with the help of artificial intelligence tools, under the author's direction; every figure was computed by a model and every rule was checked against its source.