SpaceX lines up roughly $40B in GPU financing with Apollo

SpaceX has lined up roughly $40 billion in financing for AI chips and data-center hardware, according to a securities filing made public Monday — a deal the company is billing as its “largest committed GPU financing” to date.

The structure routes the purchase through a vehicle run by Apollo Global Management, the New York private-equity firm: Apollo’s entity buys the chips and server hardware outright, then leases that equipment back to SpaceX for roughly five to five-and-a-quarter years per batch, with each lease starting once that batch ships. Closing is targeted for year-end, subject to the usual conditions, and the leases themselves are set to begin in early 2027 as the hardware arrives. Apollo said the arrangement will fund delivery of Nvidia’s next-generation Vera Rubin chips for what it called SpaceX’s “terrestrial compute buildout.” CFO Bret Johnsen cast the lease as a new channel for funding SpaceX’s growth plans and pointed to payback periods under a year on the company’s AI infrastructure outlays.

The move comes on the heels of a standout quarter for SpaceX’s AI unit. That business — the Grok chatbot, the Cursor coding tool, and SpaceX’s cloud compute operation — beat investor expectations and generated $7.6 billion of the company’s $14.6 billion in third-quarter revenue, outpacing Starlink and rocket launches combined. Capital spending jumped to nearly $20 billion for the quarter, up from $6.9 billion a year earlier, and Johnsen told investors on the Oct. 20 earnings call that SpaceX plans to spend another $35 billion to $40 billion in the fourth quarter alone.

The financing lands in a rougher patch for AI-infrastructure debt, with investors increasingly wary of runaway capital spending and thin profits elsewhere in the sector. Musk framed the deal as proof of SpaceX’s edge over pure-play AI compute rivals, citing the company’s investment-grade rating and the steady cash SpaceX pulls in from Starlink and its launch business. He argued that lenders have largely stopped extending credit to companies with no major revenue outside AI data centers, a warning aimed at the more than 200 “neocloud” firms that exist solely to rent out 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.”

Why it matters

The lease gives SpaceX a non-debt channel for funding its AI expansion, and Musk’s framing signals mounting pressure on single-purpose GPU-rental “neocloud” operators that lack SpaceX’s diversified revenue base as credit conditions tighten.