SpaceX Seeks $40 Billion for Nvidia Chip Purchases
SpaceX is negotiating about $40 billion of financing to purchase Nvidia AI chips, according to Reuters reporting based on two people familiar with the discussions. The proposed package comprises approximately $10 billion of bank loans and $30 billion of investment-grade debt. Apollo is expected to help lead the transaction, while Pimco has been involved in discussions. The financing has not been announced as completed, and the reported closing timetable extends into 2027. For Nvidia, the talks identify a very large potential source of customer purchasing capacity, but their immediate significance is access to funding: the headline amount is neither recognised Nvidia revenue nor evidence that an equivalent fleet has already been delivered.
The two proposed funding channels would draw on different pools of institutional capital. Banks can arrange and distribute loans, while a broader debt placement can bring in investors seeking contractual interest payments over a longer period. Their participation would depend on the borrower, collateral, repayment structure and any protections supporting the transaction. Those terms have not been disclosed in a completed agreement. SpaceX and Nvidia did not immediately comment to Reuters, while Apollo and Pimco declined to comment. The absence of final terms matters particularly for AI hardware, where lenders must judge both the revenue the equipment can generate and its value as newer systems become available during the life of the debt.
The proposed borrowing would follow substantial fundraising already undertaken by SpaceX. Reuters’ Morning Bid account noted the company’s US$86 billion IPO and subsequent US$25 billion bond issuance in June, while reporting that investors had pushed its credit-default-swap spread higher. The earlier proceeds and the proposed chip financing should not be added together as an available cash balance: they relate to separate transactions and spending requirements. The credit response nevertheless supplies a counterweight to the scale of the ambition. Investors can welcome demand for computing while requiring greater compensation for the financial obligations used to create it, particularly when multiple large projects depend on continuing access to external capital.
Some analysts see a strategic rationale for combining the businesses involved. Dan Ives described the proposal as “a smart strategic move for the SpaceX buildout,” in a note quoted by Investopedia, arguing that computing can connect its launch, satellite and AI activities. That is an analyst’s assessment, not a demonstrated return on the proposed borrowing. Nvidia has separately sought to broaden the financing available to its customers through partnerships with major investment firms. The shared premise is that useful computing infrastructure can attract capital beyond the buyer’s own balance sheet. Whether a particular project merits that capital depends on its contracts and operating economics, rather than the appeal of the wider AI market alone.
For the chip supplier, financing and purchase execution remain distinct stages. An agreed debt package could support orders, and delivered systems could then support revenue under the relevant contracts. The customer must still install, power and use the equipment to generate the income needed to service its obligations. It is not possible from the disclosed talks to determine the Nvidia product mix, delivery schedule, effective price or any guarantee Nvidia might provide in this specific transaction. Those omissions prevent a reliable conversion of US$40 billion into a GPU count or a quarterly sales forecast. They also leave open how much of the financing risk would sit with SpaceX, lenders or other participants.
Analysis
Three-quarters of the proposed package would come through the US$30 billion debt component, making institutional credit appetite central to its scale. For Nvidia, the attractive outcome is financed purchasing capacity without a matching assumption of customer risk on its own balance sheet. That outcome cannot be presumed before the guarantees and collateral are known. SpaceX would gain capacity while committing future cash flows to debt service, so the economic return depends on productive use arriving quickly enough to cover both operating costs and capital costs. Large orders become more durable when the customer’s earnings, rather than repeated refinancing, sustain them.