SpaceX Is Borrowing $20 Billion to Fund Its AI Ambitions. Is That a Bold Move or a Dangerous One?

When a company raises $86 billion in the largest IPO in history and then turns around five days later to borrow another $25 billion, one of two things is true: It has identified an opportunity so large that no amount of capital is enough, or it has taken on obligations it cannot fund from operations.

With Space Exploration Technologies Corp (NASDAQ: SPCX), both are true simultaneously, and that tension is exactly what the last 10 days of share transactions have been processing.

The anatomy of the SpaceX debt

SpaceX's $25 billion bond offering, priced Tuesday in five tranches with maturities ranging from five to 30 years, is the company's first-ever investment-grade dollar bond issuance. The primary purpose of the raise is not to build new rockets. It is to refinance a $20 billion bridge loan that SpaceX took out in March, when it absorbed Elon Musk's X and xAI in an all-stock deal -- and those companies' combined $17.5 billion in existing debt came along with them.

Image source: Getty Images.

The sequencing matters. SpaceX went public, raised $86 billion, and the very next week turned to the bond market because the bridge loan needed to be repaid and the AI infrastructure build-out requires capital that the IPO proceeds don't fully cover. The offering attracted close to $85 billion in orders, a genuine sign of institutional demand. But bond investors required a premium over Treasuries -- described as "large" -- to get the deal done. That premium is what sophisticated fixed-income buyers charge when they're not certain a company's cash flows fully support its debt load.

Oppenheimer analysts, in initiating coverage, projected SpaceX will carry more than $400 billion in net debt by 2031. That number assumes the AI capital spending cycle continues at its current pace, which is precisely the assumption that deserves scrutiny.

What AI revenue looks like here

SpaceX announced a $6.3 billion AI infrastructure deal with a start-up called Reflection AI on the same day it announced the bond sale, framing it as validation of the AI strategy. What the Bloomberg terminal data revealed: Reflection AI's records value the company at $3.6 million. SpaceX is lending $150 million per month in compute to a start-up worth less than one hour of SpaceX's IPO proceeds, under a contract it can terminate after three months. The deal didn't stabilize the stock. Shares fell for a third consecutive session on the news.

The deeper problem is that xAI -- the division that SpaceX absorbed and is now borrowing billions to expand -- generated $818 million in revenue against $2.47 billion in operating losses in Q1 2026 alone. Grok, xAI's large language model, has not demonstrated measurable market share against OpenAI or Alphabet's Google Gemini. SpaceX is taking on long-dated debt, payable over 30 years, to fund a bet on an AI product that is currently losing $3 for every $1 it earns.

The broader AI valuation problem investors should look at

This is where SpaceX's balance sheet decisions collide with a marketwide reckoning that the June sell-off has started to force. The entire premise of AI-driven valuations in 2024 and 2025 rested on a chain of assumptions: that infrastructure spend would translate to revenue, that revenue would compound fast enough to justify current multiples, and that the biggest spenders would capture the most value. That chain hasn't held.

Microsoft's Copilot, after two years of heavy marketing, is still a story about enterprise seat licenses -- not the mass-market productivity revolution that justified its valuation premium. Google announced $175 to 185 billion in capex this year, and its stock sold off before recovering. Amazon is deploying $200 billion on a timeline where returns are years away. The pattern is consistent: AI companies are spending as if demand will arrive faster and at higher margins than current economics support.

SpaceX is now borrowing at scale to join that race in a market already asking whether the leaders can justify what they've already spent. The stock has erased $400 billion from its post-IPO peak and trades at $157.60, near its first-day close. That's not panic. It's a reassessment.

The bull case is real: cheap long-dated capital, and Grok training on Starlink data from 10 million subscribers is a genuine moat. But the bear case is also recognizable -- using one extraordinary business as collateral to fund AI bets at valuations that may never be recovered. That's what Softbank did. It lost $27 billion in fiscal 2022.

For investors holding SpaceX stock, you're not just owning a launch and satellite company anymore. You're owning an AI conglomerate with $29.1 billion in long-term debt and projections pointing toward $400 billion more. The IPO prospectus told you that was the plan. The question is whether you signed up for it.

Micah Zimmerman has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Alphabet, Amazon, and Microsoft. The Motley Fool has a disclosure policy.