SpaceX has just signed one of the largest AI hardware deals of the year, and it did not buy the chips outright. On October 26, 2026, the company said an entity managed by Apollo Global Management will buy AI chips and server hardware and lease them to SpaceX, under a deal with total principal payments of about $40 billion.
The structure matters as much as the number. Instead of raising standard corporate debt or paying cash up front, SpaceX will make quarterly lease payments over roughly five years for each batch of equipment delivered. That is a new way to fund the AI race, and it shows how expensive that race has become even for a company with rocket and satellite revenue behind it.
What SpaceX Actually Announced
According to a regulatory filing reported by Reuters on October 26, SpaceX entered a binding commitment with an Apollo-managed entity. Apollo's entity will purchase graphics processing units and related data-center equipment, then lease that equipment to SpaceX. SpaceX will repay about $40 billion in principal through quarterly installments over approximately five years, plus interest at a fixed annual rate that has not yet been set, and certain other expenses. Each lease will run a little over five years from the day its equipment arrives, and individual leases are expected to begin in the first half of 2027. The master lease is expected to be finalized before the end of 2026, subject to closing conditions. Hardware will arrive in batches, not all at once, so the leases start as deliveries land.
Why Vera Rubin Chips Are Central
Apollo said the transaction will help deliver Nvidia's next-generation Vera Rubin chips for SpaceX's terrestrial compute buildout. SpaceX absorbed its xAI unit in February 2026, so the chips will support AI work inside the combined company. Chief executive Elon Musk said on X on October 25 that SpaceX had closed $40 billion in debt financing for graphics processing units. He added that debt markets have almost entirely shut off access to money for any company that does not have substantial revenue independent of AI data centers, while noting that SpaceX is investment grade and has Starlink and launch revenue.
What the Lease Structure Changes
A lease lets SpaceX fund a massive hardware purchase without issuing conventional bonds. It also spreads the cost over the useful life of the equipment, with quarterly payments that generally include one-twentieth of the financed cost plus interest. SpaceX chief financial officer Bret Johnsen had said the company still expects fourth-quarter capital expenditure of $35 billion to $40 billion, Reuters reported. Musk said the new financing mainly targets chips arriving in the first and second quarters and would not be a major factor this quarter.
How This Fits the Bigger AI Funding Wave
SpaceX is not alone in raising huge sums for AI infrastructure. Meta recently announced plans to raise up to $95 billion in equity for its AI expansion, and other AI labs are pursuing record-sized listings and national computing projects. The pattern is the same: the hardware bill for frontier AI is now measured in tens of billions of dollars. The difference here is the instrument. Meta chose equity. SpaceX chose a lease.
What to Watch Next
Three things will show whether this deal works as planned. First, whether the master lease closes before the end of 2026 as expected. Second, whether the first batches of Vera Rubin hardware arrive in the first half of 2027 on schedule. Third, what fixed interest rate is ultimately set, since that will determine the true total cost above the $40 billion in principal. If deliveries slip or rates rise, the lease could become more expensive than it looks today. If they land on time, SpaceX will have secured a large supply of next-generation AI chips without a traditional bond issue.
For now, one thing is clear: the AI infrastructure race is no longer just about who builds the best model. It is about who can finance tens of billions of dollars of chips, and SpaceX has just shown a new way to do it. Do you think leasing AI chips is smarter than buying them outright, or does it just hide the real cost? Share your view in the comments.

Comments
Post a Comment