SpaceX just reported its first public quarterly earnings, beat expectations on revenue, narrowed its losses, and still watched its stock crater 8% in a single afternoon. The culprit: a $15.8 billion AI spending spree in one quarter alone that has investors asking, out loud and on the record, whether any of this is actually going to work.
The Number That Broke Everyone's Brain
Let's put $15.8 billion in a single quarter into perspective. That is more than double what SpaceX spent on AI in the entire first quarter of this year. According to CBS News, AI-related investments accounted for 86% of the company's total capital expenditures from April to June, totaling $18.4 billion overall. Eighty-six percent. The rockets, the satellites, the Starlink infrastructure that actually generates revenue right now — all of that combined is the remaining 14%.
On Wednesday's earnings call, Elon Musk was characteristically unbothered. "We're building AI compute capacity at scale faster than anyone else, we believe," he said, also pointing to the recent release of Grok 4.5 as evidence of progress. Investors heard that and apparently decided they'd heard enough Musk optimism to last them a while.
Strong Quarter, Angry Market
Here is the genuinely maddening part. SpaceX actually had a good quarter by any traditional measure. CBS News reports that second-quarter revenue hit $7.8 billion, beating analyst forecasts. The company also narrowed its net loss to $541 million, down from a full $1 billion loss in the same period last year. That is real, measurable improvement.
None of it mattered. Shares fell $10.06, landing at $115.27 by early Wednesday afternoon. The market looked at the revenue beat and the improving losses and then looked at that AI spend figure and decided to panic anyway. Which, honestly, is not entirely irrational when you see what they're planning to do with all that money.
Space-Based Data Centers Are Apparently a Real Plan
The pitch SpaceX is making to justify all of this spending is that it will launch orbital data centers, potentially as soon as next year. Jay Ritter, an IPO expert and professor at the University of Florida's Warrington College of Business, told CBS News that while the orbital data center timeline was actually one of the bright spots investors responded to, the earthbound AI spending is a much harder sell. "The company is spending lots of cash on Earth-based data centers, a business that has a lot of competition," he said.
That is the tension at the heart of SpaceX's current valuation problem. The thing that makes SpaceX genuinely interesting and different from every other tech company is its ability to put things in space. The thing it is currently pouring most of its money into is the exact same AI compute race every other company on earth is running. At a $1.5 trillion valuation, investors expected something a little more original.
Musk's Credibility Problem Has a Price Tag
Ritter put it diplomatically, as professors tend to do. "Elon Musk has a history of overpromising, so investors took some of the optimistic forecasts with a grain of salt," he told CBS News. That is a very polite way of saying Musk has spent years announcing deadlines for things that didn't happen on schedule, and the market has started pricing in a discount for his enthusiasm.
David Trainer, CEO of investment research firm New Constructs, was blunter about the broader AI spending arms race SpaceX has joined. "The entire market is beginning to suspect that the extraordinary spend on AI might not work out for every firm," he told CBS News. "They cannot all be winners, yet they are all spending as if they will be." SpaceX is not just making a bet on AI. It is making a bet on being the specific AI winner in a field where Microsoft, Google, Amazon, Meta, and approximately forty other companies are also betting they will be the winner.
Thursday Could Get Worse Before It Gets Better
Wednesday's 8% drop may not even be the bad day. CBS News reports that up to 911.5 million shares become eligible for sale on Thursday as a lockup period expires, which could flood the market with new supply at exactly the wrong moment. Paul Karger, co-founder and managing partner at private investment firm TwinFocus, told CBS News that a lockup expiration doesn't guarantee mass selling, but it "does increase the potential supply of shares coming to market, which can create near-term pressure on the stock."
And if you were hoping to buy the dip as a savvy contrarian move: Nicolas Owens, an equity analyst at Morningstar who covers SpaceX, noted in a Wednesday analysis that even after the drop, Morningstar still views SpaceX shares as overvalued. So the stock fell 8% and analysts still think it costs too much. That is quite a position to be in.
The Dingo Take
You are supposed to believe that a company losing $541 million a quarter deserves a $1.5 trillion valuation because it might build data centers in space someday. That is the actual investment thesis on the table. SpaceX does genuinely extraordinary things — it builds reusable rockets, it runs the largest satellite constellation on the planet, it has materially changed what commercial spaceflight looks like. None of that is nothing. But spending 86 cents of every capital expenditure dollar on AI compute in a quarter when you are still posting nine-figure losses is not a sign of disciplined ambition. It is a sign of someone who got very excited about a trend and started writing checks.
The real story here is not that SpaceX had a bad day on the market. It's that the entire AI spending bubble is starting to make even true believers nervous. When your stock drops 8% on a quarter where you beat revenue expectations, the market is telling you something. It's telling you that "trust me, the AI will pay off" only works for so long when the bills keep doubling every three months.
Musk gets to stand on an earnings call and say SpaceX is building AI capacity faster than anyone else on earth. Maybe that's true. But faster than anyone else at spending money on something with an unclear return is not the competitive advantage it sounds like. The satellites are real. The rockets land themselves back on launchpads. Those are remarkable things. The Grok chatbot competing against fifteen other chatbots from better-funded pure-play AI companies is a considerably less remarkable thing, and right now it is eating the budget whole.
Comments