The floating cost behind Energy Vault’s AI power financing

Private lender Eagle Point on 29 October described a US$137.5 million facility supporting Energy Vault’s purchase of 275 megawatts of Rolls-Royce MTU engine capacity for AI infrastructure. The lender says its data-centre underwriting weighs delivery timing, power availability and the durability of demand.[1]

The transaction has a longer history than this week’s announcement. Energy Vault’s subsidiaries signed the agreement on 14 August, according to its SEC filing. The company announced the engine procurement on 10 September; the lender’s law firm, A&O Shearman, described the financing on 8 October.[4][2][5]

Energy Vault expects deliveries from the second half of 2027 through the first half of 2028. It plans to combine the engines with battery storage, power conversion equipment and plant controls. Their modular design is intended to let it add generation as individual AI campus projects develop.[2]

What the money costs

The filed agreement funds supplier instalments as they fall due, with a final maturity of 2 January 2028. Its interest provisions matter: Loans priced against Term SOFR, a dollar borrowing benchmark, carry that rate plus 6.75 percentage points through December 2026, rising to plus 7.50 points thereafter. Those percentages are margins, rather than the complete interest rate.[3]

An annual 1% fee also applies to undrawn commitments after the first funding. Delaying a draw therefore reduces the balance accruing loan interest, but retaining access to unused financing still has a cost.[3]

A&O Shearman partner Andrew Lanius described the commercial logic in the firm’s October statement: long waits for generation equipment and grid connections make future production capacity worth financing now. That is the account of an adviser to the transaction, with an interest in explaining its merits.[5]

The company’s plan is to preserve more corporate cash before placing equipment into customer projects. Its September announcement refers to multi-gigawatt project discussions and names no customer site for these engines. The 275 MW therefore describes contracted equipment supply; the delivery schedule is the next concrete stage in turning it into usable power.[2]

For AI infrastructure buyers, the financing illustrates a cost that begins before servers can run: securing the equipment timetable. Eagle Point is lending against that earlier stage of development, while Energy Vault pursues the projects expected to use the capacity.[1]

Report an error or suggest a correction