SpaceX Just Borrowed $40 Billion to Buy Computer Chips and That's the Good News

A graphics processing unit displayed in a bright, modern technology lab

Date: October 26, 2026 | Byline: Dakota Jones
Desk: Economic, Business & Financial, Corporate & Industry

Intro

SpaceX says it has closed $40 billion in debt financing to buy graphics processing units, or GPUs. The chips that power artificial intelligence systems. That is an enormous amount of borrowing for computer hardware. It is also a pointed reminder of who can still get money in the AI boom.

On Sunday, October 25, Elon Musk said on X that SpaceX had secured the financing. His explanation: the company has investment-grade status and revenue from Starlink satellite internet and rocket launches. In other words, lenders can look beyond AI and see businesses already bringing in money. That distinction is becoming increasingly important.

The chip purchase is the headline. The credit divide is the story.

What Happened

Elon Musk at an event in Washington, D.C.

Musk said the financing is primarily callable debt, meaning the borrower can repay it before maturity. He said the GPUs are expected to arrive in the first and second quarters of 2027, and described the deal as “not a major factor this quarter.”

Bloomberg had previously reported that SpaceX was in talks with banks and investors to raise $40 billion for Nvidia chips. The reported structure involved a special purpose vehicle, or SPV. A separate entity used to hold assets or financing. The aim was to keep the debt off SpaceX’s balance sheet and preserve its investment-grade credit ratings. Bloomberg, as reported by CNBC, said Goldman Sachs was leading talks with investors including Apollo Global Management.

SpaceX is not shopping for chips on a whim. Its AI operation, called “Super Intelligence,” or SI, is now the company’s largest business unit. In the third quarter, AI revenue rose roughly nine-to-ten-fold, reaching $7.56 billion, compared with $813 million a year earlier. Total revenue climbed 161% year over year to $14.58 billion for the three months ended September 30.

The spending is just as striking. CFO Bret Johnsen said SpaceX expects $35 billion to $40 billion in capital expenditure in the fourth quarter. Capital expenditure, often shortened to capex, is money spent on long-lived assets such as equipment and infrastructure. Apparently, in today’s AI race, “long-lived equipment” includes a truly staggering number of expensive chips.

Rows of servers in a bright, modern data center

Why It Matters to You

The headline isn’t just about SpaceX, Nvidia, or the latest AI spending spree. It’s about which companies can borrow to build the infrastructure behind AI, and what lenders require in return.

Musk said debt markets have “almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.” That is the heart of the matter. SpaceX can point to Starlink and rocket launches as revenue sources beyond AI. AI-focused cloud infrastructure providers, commonly called neoclouds, have a harder case to make if their business depends on selling compute capacity and financing more of it.

Musk predicted that more than 200 neocloud companies (including firms such as Nebius, CoreWeave, and Iren) will face widespread failures and consolidation, with only a small number surviving. He later said SpaceX would not buy any neoclouds, calling the idea “pointless.” Those are Musk’s predictions, not a final accounting of the sector’s fate. Still, the market heard him: Nebius, CoreWeave, and Iren shares were all down more than 4% in premarket trading Monday after his comments.

Investors have already started betting against U.S. neocloud and data center companies amid concerns that their debt-fueled AI buildout is unsustainable. The warning here is not that AI demand has vanished. It is that demand alone may not be enough to keep financing available. Lenders want a credible path to revenue, and, increasingly, they want that revenue to come from somewhere besides AI data centers.

That makes SpaceX’s deal both a financing story and a competitive advantage. A company with several revenue streams can potentially borrow on terms that are unavailable to a company selling AI compute alone. For smaller providers, the environment may become less “build now, grow later” and more “show the cash flow first.” Not quite as catchy. Considerably more useful.

The Bigger Bet: Chips, Factories, and More Spending

SpaceX has expanded aggressively into AI this year. After a blockbuster IPO, its S-1 statement listed “manufacturing our own GPUs” among planned capital spending. The company’s ambitions extend beyond buying chips: Terafab, the chip-manufacturing joint venture between SpaceX and Tesla in Austin, should produce its first test chip around May or June 2027, with volume production targeted for Q3 2028.

A semiconductor wafer and test chip on a precision workbench

Musk also told investors last week that a merger of SpaceX and Tesla would be “a good idea.” That comment belongs in the wider context of a company expanding across AI computing, chip manufacturing, satellite internet, and launch services. It does not, by itself, tell us that a merger will happen.

For you, the practical takeaway is that AI infrastructure remains an expensive, high-stakes buildout, and that the companies funding it are not all standing on equal ground. The availability of capital can shape who gets to expand, who has to slow down, and who may be pushed toward a sale or consolidation.

FAQ

What is a GPU?
A graphics processing unit is a chip designed to handle many calculations in parallel. That makes GPUs useful for graphics and for training or running many AI systems.

Does callable debt mean SpaceX has to repay early?
No. It means the borrower can repay before maturity. Musk said the financing is primarily callable debt.

When are the chips expected to arrive?
Musk said they are expected in the first and second quarters of 2027.

Are neocloud companies certain to fail?
No. Musk predicted widespread failures and consolidation among more than 200 providers. That is a forecast, not a confirmed outcome.

Verdict

SpaceX’s $40 billion GPU financing shows how much capital the AI buildout requires, and how much access to that capital depends on the borrower’s business beyond AI. SpaceX has Starlink and rocket-launch revenue to point to. AI-only infrastructure companies face tighter credit and a tougher question: can their revenue support their borrowing?

The chips are the visible purchase. The more consequential asset may be SpaceX’s access to financing. For AI companies, investors, and customers, watch the revenue behind the compute, not just the size of the next data center announcement. For more coverage from LD3 Productions, visit our news page.

Sources