SpaceX has secured a binding financing commitment for roughly $40 billion in AI chips and related data center equipment. The arrangement gives the company another way to fund its expanding AI infrastructure, while an entity managed by Apollo Global Management will own the equipment and lease it to SpaceX.
A lease tied to equipment deliveries
Under the agreement, Apollo’s managed entity will purchase the chips and related server hardware, then lease that equipment to SpaceX. The leases will run about five to five-and-a-quarter years, with each term beginning when its tranche of equipment is delivered, according to SpaceX’s regulatory filing.
The transaction is expected to close by year-end, subject to closing conditions. SpaceX expects the leases to start in the first half of 2027 as GPUs and related equipment arrive. That timing ties the financing to the gradual delivery of hardware rather than a single start date for all of the equipment.
SpaceX described the arrangement as the “largest committed GPU financing” in a statement. CFO Bret Johnsen said it diversifies the company’s funding sources and covers a notable portion of the money needed to deliver its growth plans. He also said SpaceX is seeing “less than one-year payback periods” on its AI infrastructure investments as it builds computing capacity for Grok and its cloud services.
AI infrastructure is driving spending
The financing arrives as SpaceX expands its AI operations alongside Starlink and its launch businesses. The company’s AI segment includes the Grok chatbot, the Cursor coding tool and its cloud compute business. In the third quarter, that segment exceeded investor expectations and generated $7.6 billion of SpaceX’s $14.6 billion in revenue, more than Starlink and rocket launches combined.
SpaceX’s capital expenditures climbed to nearly $20 billion, compared with $6.9 billion a year earlier, as it worked to expand its compute footprint. On the company’s Oct. 20 earnings call, Johnsen said SpaceX expected to spend $35 billion to $40 billion in the fourth quarter alone.
The financing therefore addresses a substantial need identified by the company as it scales its AI infrastructure. A lease lets SpaceX use the equipment while the Apollo-managed entity buys and owns it under the agreement described in the filing. SpaceX says the arrangement diversifies its funding sources; the source does not provide further terms about payments or the overall cost of the leases.
Apollo points to next-generation Nvidia chips
Apollo said the financing will help deliver Nvidia’s next-generation Vera Rubin chips for SpaceX’s “terrestrial compute buildout.” The hardware and its delivery schedule are central to when the leases begin: each lease starts when its tranche arrives, and SpaceX expects those starts in the first half of 2027.
For SpaceX, the deal links capital for AI computing with an expansion that the company says supports Grok and cloud services. For Apollo, the commitment is connected to supplying equipment for that buildout. The agreement remains subject to closing conditions, and the company expects it to close by year-end.
Contrasting financing conditions in AI computing
The deal follows a difficult period in debt markets for companies building AI infrastructure. The source describes investors as growing wary of rapidly rising capital expenditures and limited profits in the sector. Against that backdrop, SpaceX has pointed to its investment-grade status and cash flow from Starlink and rocket launches as features that distinguish its financing position.
Elon Musk argued on Sunday that debt markets have largely shut out companies without major revenue streams beyond AI data centers. He said that could put pressure on more than 200 neocloud companies whose business is renting GPUs for AI workloads. “That means the 200+ neoclouds are going to see guppy-level death and consolidation,” Musk wrote on X. “Some will make it through, but not many.”
SpaceX’s financing announcement thus sits within a broader debate about how AI computing companies pay for expansion. The company has a large AI business and substantial planned spending, while also pointing to revenue from other operations. Its lease with an Apollo-managed entity is one way to obtain chips and server hardware for that buildout, with the first lease terms tied to deliveries expected in 2027.