SpaceX Lines Up $40 Billion in Debt to Buy Nvidia GPUs

SpaceX has raised $40 billion in debt financing to buy graphics processors from Nvidia and other suppliers, Chief Executive Elon Musk said in a post on X. The chips will be delivered in the first and second quarters of 2027, and Musk expects most of the debt to be repaid early without a significant impact on fourth-quarter earnings.

The disclosure appears to confirm a Financial Times report from October 7, which said SpaceX was in talks to secure $40 billion. That report named Apollo Global Management as lead investor and described the package as roughly $10 billion in bank loans plus $30 billion in investment-grade debt. Pacific Investment Management Company was also cited among the institutions providing funds.

Musk said SpaceX could raise the money because it holds an investment-grade credit rating and generates substantial revenue from sources other than AI data centers, including its Starlink satellite internet service and space launches. The same door is now closed to most others. For companies without independent revenue streams, debt markets have been "almost entirely shut off," Musk claimed.

What does this mean for neoclouds?

The neoclouds are startups that rent specialized GPU computing capacity to AI developers and other customers. Many expanded rapidly to meet demand for computing power, often borrowing money to acquire the hardware.

Musk predicted that the tightening credit environment will force a wave of consolidation among these operators. "200+ neoclouds are going to see guppy-level death and consolidation," he said. "Some will make it through, but not many." His forecast implies that many smaller players will be acquired by larger competitors rather than survive independently.

The warning points to a widening divide between financially established hyperscalers — Amazon Web Services, Microsoft and Google, all of which report earnings this week — and their smaller rivals. Tougher lending standards could push smaller operators into mergers or force them to cancel expansion plans, handing an advantage to the incumbents.

How large is the AI borrowing wave?

The numbers behind the boom are large by any measure:

• Morgan Stanley forecast in June that capital expenditure by AWS, Microsoft, Google and Meta Platforms would reach $700 billion this year and over $1 trillion in 2027.
• Goldman Sachs projects borrowing by those companies will reach $250 billion this year and $400 billion in 2027.
• Morgan Stanley expects global AI-related debt issuance to top $570 billion this year.

Even the largest borrowers are funding their buildouts with vast amounts of borrowed money, not cash alone.

Why are investors demanding higher yields?

The scale of the borrowing has already moved credit markets. According to Reuters, investors are now demanding higher yields on AI infrastructure-related corporate bonds because of concerns about the massive volume of new debt. Many investors doubt that revenue from AI services will grow fast enough to justify spending of this size on chips, data centers and the supporting infrastructure those facilities require.

Higher borrowing costs change the economics of every new GPU order. A company that can self-fund, or borrow at investment-grade rates as SpaceX has done, faces a materially lower cost per unit of computing than a leveraged startup facing rising yields.

The consequences would not stay inside the lending market. A wave of neocloud failures would reduce demand for AI chips, potentially cutting into the revenue of suppliers such as Nvidia, whose data center business has been built on exactly this class of buyer.

If Musk's read on credit conditions is right, the next year will sort the AI infrastructure market into a small group of well-capitalized builders and a long tail of sellers, with chip demand increasingly concentrated among the survivors.