SpaceX’s $40 Billion Financing Faces Hurdles as Borrowing Costs Rise

Key Facts

SpaceX is working to finalize a $40 billion off-balance-sheet debt structure to fund its expansion in artificial intelligence, according to Bloomberg Law. The talks have faced market volatility and higher borrowing costs across the sector, alongside more complex pricing negotiations. The structure is aimed at financing Nvidia chip purchases through a special-purpose vehicle. According to the report, the arrangement would allow the obligations to remain off SpaceX’s books and help preserve its investment-grade ratings. The report did not identify global credit-market volatility as the primary cause of the challenges, instead linking them to higher borrowing costs and pricing complexity.

Bankers from Goldman Sachs are leading discussions on the structure with investors led by Apollo Global Management, according to Bloomberg Law. Apollo-backed Atlas SP is set to manage the proposed structure. The arrangement would put a special-purpose vehicle at the center of the chip purchase and its financing, separating the related obligations from SpaceX’s books. That separation is intended to help the company preserve its investment-grade ratings, according to the report. But market volatility has raised borrowing costs across the sector and complicated negotiations over the financing price.

The talks come as SpaceX has expanded its debt financing; the company said in its quarterly report that it issued $25 billion in senior unsecured notes in June. The weighted average interest rate on those notes was 5.855% as of September 30, 2026. The company also reported that the total principal of its debt and that of its subsidiaries was $38.006 billion on September 30. The proposed $40 billion structure would add substantial new financing in this context, although it is intended to keep the obligations off balance sheet. Financing terms and pricing remain under discussion, according to Bloomberg Law.

The proposed borrowing is tied to SpaceX’s expansion of the infrastructure needed for its AI activities, with financing directed toward Nvidia chip purchases through the special-purpose vehicle. Third-quarter results showed that AI had become SpaceX’s largest segment by revenue, while Starlink remained its biggest contributor to profit, according to the Associated Press. The company posted a profit of $4.4 billion in the three months ended September 2026, following a loss in the same period a year earlier. Revenue rose to $14.6 billion, up 161% year over year. The results provide financial context for the expansion, while details of the debt deal’s pricing remain under negotiation.

The AI plans are underpinned by expected demand for computing services; Reuters reported in July that Colossus data center contracts with Anthropic, Google and Reflection AI are expected to generate more than $28 billion annually. That compares with AI revenue of about $3.2 billion in 2025. Analysts have cautioned that the contracts include termination provisions, limiting the extent to which the expected revenue can be treated as guaranteed. EL7 previously reported in July that Foxconn had won a contract to supply AI servers to SpaceX, valued at $52 billion in the headline. These developments form part of the company’s expansion of its computing capacity, while the proposed financing for chip purchases depends on finalizing the structure and settling its terms.

In trading, SPCX closed at $159.31 on October 21, 2026, then at $144.75 on October 22, according to EL7 data. The October 22 close was below the previous day’s close, but the data available here do not establish that the move resulted from the financing talks. The stock had risen 7.5% in after-hours trading on October 21 following the third-quarter results announcement, according to the Associated Press. More broadly, the agency said concerns about rising borrowing costs rattled AI stocks in the previous week, while higher U.S. Treasury yields increased financing costs for borrowers. On the outlook for expansion, Reuters said selling computing capacity to institutional customers is already underway, while opportunities in orbital AI remain longer term because they depend on introducing Starship rockets, reducing launch costs and making technological progress.