SpaceX is seeking $40 billion in financing to purchase Nvidia AI chips, according to a Bloomberg report. Apollo Global Management would lead the transaction, structured as $10 billion of bank loans plus $30 billion of investment-grade debt. None of the parties — SpaceX, Nvidia, or Apollo — has confirmed the deal, which would not be expected to close until 2027.

The report is credible on its face. SpaceX is building an AI infrastructure business with billions of dollars in contracted revenue, and CEO Elon Musk has stated his hardware preference openly. If the financing proceeds, it signals that the AI capital cycle still has substantial runway.

How big is SpaceX as an Nvidia customer?

SpaceX operates a compute-leasing business with three disclosed contracts:

• Anthropic: $1.25 billion per month for access to roughly 325,000 Nvidia GPUs
• Google: $920 million per month for access to around 110,000 Nvidia GPUs
• Reflection AI: $150 million per month for access to Nvidia GB300 chips inside SpaceX's Colossus 2 data center

On its second-quarter earnings call, management disclosed $6.7 billion of additional cloud services revenue booked during the opening weeks of the third quarter, with capacity ramping over six months. SpaceX subsequently disclosed another unnamed hosting customer expected to generate $1.1 billion of monthly revenue beginning in December.

During that same Q2 call, Musk said SpaceX has decided to build "exclusively" on Nvidia because he believes the Vera Rubin architecture is the best AI computer. SpaceX expects its cumulative compute capacity to approach 10 gigawatts by the end of 2027 — an infrastructure target that carries a large price tag on an aggressive timeline.

Why does SpaceX need $40 billion of debt?

The company's cash profile explains the borrowing. SpaceX spent $18.4 billion on capital expenditures in the second quarter alone. Free cash flow through the first six months of the year was negative $25 billion as the company invested simultaneously in AI data centers, Starlink, and Starship.

SpaceX would not be an outlier. AI hyperscalers — Amazon, Alphabet, Meta Platforms, and Oracle — collectively issued roughly $194 billion of bonds through early July to finance AI infrastructure. Lindsay Rosner, who leads Multi-Sector Investing in Goldman Sachs Asset Management, says the hyperscalers could issue another $420 billion of debt in 2027 as build-outs remain a priority.

The bond issuance does not indicate distress. These companies generate enormous cash flow, but the infrastructure they plan to build has grown so large that outside capital has become necessary. SpaceX's reported raise fits that pattern directly.

What does the deal mean for Nvidia?

Nvidia supplies much of the data center infrastructure the hyperscalers are financing. In its second quarter, the company generated $89 billion of data center revenue, up 117% year over year. Hyperscalers contributed $48.7 billion of that segment revenue, an increase of 102% year over year.

Investors should not treat the $40 billion financing as completed. But the idea fits the broader trend: AI infrastructure spending has become capital-intensive enough that the world's largest technology companies increasingly use debt to supplement cash flow.

If SpaceX raises tens of billions of dollars specifically for Nvidia hardware after committing to Vera Rubin, it reinforces a position that is difficult to ignore: Nvidia remains one of the largest toll collectors of the AI infrastructure boom.