Eleven files from inside the
paperwork, where the risk of the AI build actually has to be allocated. Power diligence,
borrowing-base tests, and what a GPU is assumed to be worth in year four.
Mike Niekoop
Partner · Norton Rose Fulbright
structures adapt risk moves faster
File 01 · The framing
Financing structures are
adapting. The risk is moving faster.
The old profile
One asset,
contracted capacity, a strong customer. Traditional project finance fits this.
The AI profile
Shorter-duration
hardware. Rapid obsolescence. Separate equipment financing. And revenue that depends on
an ultimate AI customer the lender never contracts with.
Lenders must assess
how cash flows, credit support, deployment obligations and termination rights interact
across the DatacentreCo, the GPUCo and the end customer.
65
out of 100 toward structures straining under AI-era risk
a strain, not a failure
Contributed · Mike Niekoop · Partner, Norton Rose Fulbright01 / 13