GPU Project Finance & Modelling
GPU infrastructure is a capital-intensive asset class with accelerating depreciation, volatile utilisation, and pricing that moves quarterly. Generic financial models built on static assumptions produce dangerously wrong answers. Ours don't.
Every model draws on live pricing data from 70+ providers, real-world cluster economics, and assumptions stress-tested against actual operator performance.
What's included
Project Finance Models
Full project finance structures for GPU cluster deployments: debt sculpting, DSCR and LLCR cover ratios, cash cascade waterfalls, reserve accounts, and equity return analysis. Built to institutional standards that satisfy lender technical advisors and investment committees.
Cluster Unit Economics
Granular cost modelling from GPU acquisition through power, cooling, networking, staffing, and maintenance. Revenue modelling across contract types (reserved, on-demand, spot) with utilisation sensitivity. We model the economics at the rack, row, and facility level.
TCO & Depreciation Analysis
GPU hardware depreciates faster than any other infrastructure asset. We model total cost of ownership across GPU generations, accounting for technology refresh cycles, secondary market residual values, and the real cost of running mixed-generation fleets.
Business Plan Financials
Three-statement financial models (P&L, balance sheet, cash flow) for neocloud business plans, fundraising decks, and lender presentations. Integrated with operational assumptions and scenario analysis so investors can stress-test the numbers themselves.
Pricing & Revenue Sensitivity
What happens to your IRR if H100 spot prices drop 30%? If utilisation falls from 85% to 65%? If your largest customer churns? We build scenario and sensitivity frameworks that answer the questions investors and lenders actually ask.
How we work
Assumption Workshop
Work through your operational assumptions together: hardware mix, deployment timeline, pricing strategy, customer pipeline, and financing structure. Challenge anything that doesn't hold up against market data.
Model Architecture
Design the model structure: inputs sheet, timing, capital expenditure, revenue build, operating costs, debt schedule, tax, and returns analysis. Agree on scenario definitions and sensitivity parameters.
Build & Populate
Build the model in Excel following institutional methodology (Camilla Culley standards where applicable). Populate with your specific inputs and Disintermediate's proprietary market benchmarks.
Stress Test & Audit
Run full scenario and sensitivity analysis. Audit every formula path. Produce a model integrity report documenting assumptions, sources, and known limitations.
Handover & Walkthrough
Deliver the model with full documentation. Walk your finance team, investors, or lenders through the structure, assumptions, and outputs. The model is yours to own and extend.
What you'll have at the end
Financial Model (Excel)
Fully auditable Excel model with inputs, assumptions, debt schedule, three-statement financials, and returns analysis. Institutional-grade, not a template.
Scenario & Sensitivity Pack
Base, upside, and downside scenarios with tornado charts on key drivers: GPU pricing, utilisation, power costs, customer concentration, and capex timing.
Assumption Documentation
Every assumption sourced and documented. Market benchmarks from Disintermediate's proprietary data clearly separated from management inputs.
Investor Summary
Two-page financial summary suitable for investment committee papers, lender presentations, or board materials.
Model Walkthrough
Recorded walkthrough session covering model structure, key assumptions, scenario outputs, and how to update inputs as your business evolves.
Who engages on financial modelling
- Neocloud operators building bankable models for debt financing or equity fundraising
- Infrastructure investors needing independent model validation or build-from-scratch analysis on target assets
- Lenders and technical advisors requiring borrower model review with GPU-specific market expertise
- Founders preparing financial projections for Series A/B fundraising in GPU infrastructure
- Sovereign programmes building investment cases for national AI compute capacity
Depending on model complexity and data availability
Culley methodology, fully auditable, lender-ready