SpaceX has borrowed $43 billion using Nvidia GPUs as collateral. Under the structure, a separate company buys the chips and SpaceX makes quarterly payments to use them, and lenders have backed the deal.
SpaceX is borrowing $43 billion to acquire Nvidia AI chips, using the chips themselves as collateral for the loan. The financing, named Project Saturn, is structured so that the chips are leased back to SpaceX.
The arrangement raises questions for lenders about whether the collateralized chips could be recovered if SpaceX defaults, particularly if they are deployed in its planned orbital data center. The structure also reflects how large AI infrastructure is being financed through asset-backed arrangements.
The financing covers Nvidia GPUs acquired through the Saturn Funding Trust, a fully owned subsidiary of WHCO, with SpaceX responsible for all operating expenses. The AI1 satellite it has revealed carries a 120-kilowatt payload, and the planned Starbase supercomputer is priced at $24 billion.
SpaceX pays the Saturn Funding Trust quarterly for use of the chips under a five-year, triple-net, hell-or-high-water finance lease, a structure similar to aircraft leasing. The company said it will deploy the collateralized chips only on Earth, while rating agency Moody's stated in its report that the chips will be deployed in SpaceX's U.S. data centers.
The arrangement applies the aircraft-leasing model, in which a lessor owns high-value equipment and leases it to operators, to AI accelerators. It reflects how capital-intensive AI infrastructure is increasingly financed through asset-backed structures rather than conventional corporate debt.
For most users, the financing will not change how they access SpaceX's Starlink service directly. The arrangement matters mainly to investors and lenders, whose exposure depends on whether the collateralized chips can be recovered if the company defaults.
The deal shows how AI computing hardware is becoming a financeable asset class, with the chips themselves serving as the security behind large loans. Readers following AI infrastructure should watch how lenders and rating agencies treat such collateral as deployment plans take shape.