Main finding
The three-year budget splits into access ($12M), integration and customization ($5M), and validation and deployment ($3M). It buys a customized right to use an existing platform, not ownership of it, and the proposal is presented as an opportunity rather than a confirmed commercial term.
What supports it
- The internal cost model reproduces the split arithmetically; the three components sum to the stated total without overlap.
- Licensing practice separates access, integration and enablement, because each is delivered by a different team at a different time.
- Access is the largest share because the underlying capability already runs and its cost was carried by the provider.
- Validation is the smallest line but the one that decides whether the licence is renewed.
What this does not claim
- No supplier quotation exists. Every figure is a preliminary planning estimate that negotiation may move materially.
- Ownership of the technology, model weights and source code stays with the provider, and the right ends with the term.
- Nothing here is an offer, a commitment to purchase or a valuation of the technology.
Planning assumptions used
- A single three-year term, with customization scoped to the six priority applications.
- Validation and enablement cover our data integration, testing and staff training.
- All figures remain unconfirmed planning assumptions.
Open before signature
- What happens to the customized layer at the end of the term?
- Are overage fees, additional users or new models priced separately?
- Which deliverables convert from estimate to milestone payment at signature?
Sources
Figures marked illustrative are planning assumptions. This note separates published evidence from project assumptions and does not present a supplier quotation, a valuation or a forecast of returns.