SpaceX has signed a binding commitment for roughly $40B of financing covering artificial intelligence chips and the data center equipment around them. The company called it the largest committed GPU financing ever arranged.
An entity managed by Apollo Global Management buys the hardware and leases it back. Each tranche carries a lease of about five years once delivered, with the deal expected to close by the end of 2026.
Hardware arrives through the first half of 2027. Apollo said the money will help deliver Nvidia’s next-generation Vera Rubin chips for what it called the terrestrial compute buildout.
Why lease instead of borrow
Chief financial officer Bret Johnsen framed the structure as diversification. Traditional debt markets have grown expensive for anything tied to data centers, while SpaceX needed to fund a buildout that keeps accelerating.
He also said the company sees payback periods under a year on its AI infrastructure, a claim that sits awkwardly beside the five-year lease.
The AI segment, which covers Grok and the Cursor coding tool alongside cloud compute, brought in $7.6B of the $14.6B in third-quarter revenue. Capital spending climbed to nearly $20B from $6.9B a year earlier.
The neocloud backdrop
Musk said on X that debt markets have shut off money for companies lacking revenue outside AI data centers. He predicted widespread failure across more than 200 neoclouds.
SpaceX leases out its own compute, putting it in direct competition with the businesses he expects to implode.
