SpaceX’s results were spectacular, but its next-quarter building budget could exceed twice its latest revenue. Is that visionary or reckless?
KEY TAKEAWAYS
SpaceX beat Wall Street’s expectations across every major financial measure, reporting $14.58 billion in revenue and $4.4 billion in net income. Its first quarterly profit was far stronger than analysts anticipated.
“Super Intelligence” has become SpaceX’s largest business, generating $7.56 billion, or 52% of total revenue. The company now earns far more from renting AI computing capacity than from launching rockets.
SpaceX spent $19.6 billion on capital expenditures and burned approximately $23 billion in free cash flow during the quarter. Management expects next-quarter capital spending to climb to between $35 billion and $40 billion.
Musk believes an enduring shortage of AI computing capacity will support high rental prices and rapid investment paybacks. But scarce chips, power equipment, and cooling systems also give SpaceX’s suppliers considerable leverage.
Investors should distinguish the CFO’s formal revenue guidance from Musk’s vision of $500 billion to $1 trillion in annual recurring revenue. The company’s opportunity may be enormous, but its execution and financing risks are just as extraordinary.
MY HOT TAKES
SpaceX’s spending is not automatically reckless, but calling it visionary does not make the risks disappear. The investment case depends heavily on today’s unusually favorable AI rental prices lasting long enough to recover the upfront cost.
SpaceX may be an AI landlord to its customers, but it remains a tenant to the companies supplying chips, electricity, and cooling infrastructure. The businesses controlling those bottlenecks may capture more of the economics than investors currently expect.
Musk’s trillion-dollar revenue range is an aspiration, not actionable financial guidance. It may communicate the scale of his ambition, but it does not belong in an investor’s valuation model.
The third quarter’s headline profit deserves scrutiny because one-time benefits and stock-based compensation affected the underlying economics. Investors need to see whether profitability survives when those favorable factors disappear.
The defining question for SpaceX–and the wider AI trade–is whether revenue can grow on the same timetable as spending. A great long-term opportunity can still become a painful investment when the bills arrive years before the payoff.
You can quote me: “Guidance is a promise management expects to be held to. Aspiration is a dream with a microphone.”
Bigger boat. I have watched the film Jaws more times than I care to admit. I remember the first time quite clearly–the summer of 1975. I was probably too young to be watching a man-eating shark on the big screen, and I most likely have my older cousin, who was summering with us that year, to thank (or blame) for getting me in the door. More than fifty years later, I still wait for the same moment every single time. Chief Brody–Roy Scheider at his finest–is on the stern of the Orca, ladling chum (the bucket of fish guts shark hunters toss overboard as bait) and grumbling that this job is beneath him. Then the great white rises out of the water, right behind his ladle. Brody backs slowly into the cabin, cigarette still dangling, and delivers the most famous understatement in movie history to Quint (the weathered shark-hunter captain): ”You’re gonna need a bigger boat.”
Here's what I love about that scene. There was nothing wrong with the boat. The Orca was a perfectly good boat–right up until the moment everyone on board saw what they were actually dealing with. The boat didn't change. The size of the problem did.
I thought about Brody yesterday afternoon as I watched SpaceX report its Q3 results. The company that went public in June just posted its first-ever quarterly profit–and a big one. Then, later in the call, the CFO told investors that next quarter's capital spending will be $35 billion to $40 billion. That's roughly DOUBLE what they spent in the quarter that just ended. My friends, this was SpaceX looking over the stern, seeing what it was dealing with, and ordering…a bigger boat.
Let me give credit where it's due, because the quarter was genuinely impressive. Revenue came in at $14.58 billion versus the $12.88 billion Wall Street expected, up 161% from a year ago. Earnings were $0.35 per share against a $0.13 estimate. Net income was $4.4 billion–more than double the consensus forecast of $1.9 billion–compared with a loss of $1.1 billion a year ago. EBITDA (earnings before interest, taxes, depreciation, and amortization–a rough, heuristic gauge of operating cash power) hit $8.8 billion. SpaceX beat on every major line. Every. Single. One. The stock closed at $147.79 on Tuesday and was up about 9% in premarket trading this morning. But the most important number isn't the profit. It's what kind of company produced it.
SpaceX, believe it or not, is no longer primarily a rocket company. It isn't even primarily a Starlink company. Its ”Super Intelligence” segment–the AI business, which mostly rents computing power to other companies (think of it as a landlord for data centers)–brought in $7.56 billion! That's 52% of total revenue, up roughly $5 billion from the prior quarter alone. A year ago, that segment generated $813 million. Starlink, the satellite internet business everyone associates with the name, contributed $5.25 billion. Launching rockets–the thing Elon Musk built the company to do–was just $1.77 billion. The rocket company is now an AI landlord that happens to own rockets.
Cue the huge mechanical shark!
SpaceX spent $19.6 billion on CAPEX (Wall Street throws this abbreviation around a lot, so in case no one ever told you, it stands for: capital expenditures–money spent on long-lived physical assets like data centers, chips, satellites, and launch pads 😉) in the third quarter. That works out to roughly $1.34 of spending for every $1 of revenue. According to my friends at Bloomberg Intelligence, the company burned about $23 billion of free cash flow in the quarter. Twenty-three billion dollars–in THREE months!
And the fourth-quarter guide of $35 billion to $40 billion means the spending is about to outrun the revenue by an even wider margin, at least for a while. Management guided first-quarter 2027 revenue to $33 billion to $37 billion. Read that again. 👈👀 The building budget for THIS quarter is roughly the same size as the revenue they expect to collect in the quarter AFTER it–and more than DOUBLE the revenue they just reported.
In a prior world that might come across as being reckless. Is it? Not necessarily. SpaceX ended the quarter with roughly $90 billion in cash and marketable securities, a $63 billion backlog of contracted business, and CFO Bret Johnsen insists that at current rental prices, new computing capacity pays for itself in less than a year (try to remember that for the next few earnings announcements 😉). Musk went even further, suggesting a six-month payback next year was ”not impossible” (that quote is the work of a genius wordsmith committing to absolutely nothing 🤣). If those numbers do hold, today's spending is tomorrow's cash machine. That's a big ”if,” and it brings us to the strangest moment on the call.
Musk explained that SpaceX tried to rent extra computing capacity from other cloud providers to free up its own machines for training its AI models. He compared the experience to the Monty Python cheese shop sketch–the one where a customer asks for cheese after cheese after cheese, and the shop has none of them. What about Nvidia's GB300s? None. GB200s? None. Google's TPUs? Amazon's Trainium chips? None and none. Even H100s–chips that are a few years old now–couldn't be found. Musk's conclusion: a shortage of AI compute is coming, and it will ”remain for a long time.” He even claimed older chips are renting for more today than when they were new. As you probably suspect, I seized on that instantly!
Think about what that means. On one hand, it's the single best argument for SpaceX's spending spree: if you literally cannot buy computing power at any price, the company that owns it has enormous pricing power. Management said rental prices are tracking above the high end of the range it discussed just last quarter. On the other hand, a market where buyers can't find supply is a market where the SELLERS of chips, power equipment, and cooling call the shots. SpaceX is a landlord to its customers–and a tenant to its suppliers.
Then came the line that will be quoted for months. After walking through the gigawatts and the GPUs, Musk said SpaceX's annual recurring revenue (ARR–the revenue run rate if you annualize the most recent period) could be ”somewhere between $500 billion and $1 trillion” by the end of 2027. To his credit, he immediately quipped that he didn't want to give his CFO a heart attack. Gwynne Shotwell, the company's president, picked it right up: ”On that crazy note.”
I'd encourage investors to separate two very different numbers here. One is guidance: the CFO's $33 billion to $37 billion for the first quarter of 2027. If you annualize that, you get roughly $132 billion to $148 billion. The other is aspiration: the CEO's $500 billion to $1 trillion. The low end of the aspiration is more than triple the high end of the guidance. Guidance is a promise management expects to be held to. Aspiration is a dream with a microphone. Both have their place. Pro tip: only one belongs in your spreadsheet.
There were also a few quieter details worth noting. Management warned that fourth-quarter gross margins will look more like the second quarter's, because the third quarter benefited from one-time items–including tariff refunds that added roughly three percentage points to Starlink's margin–and a favorable customer mix. The AI segment's operating income was just $431 million on $7.56 billion of revenue, though management points out it would have been $1.6 billion without stock-based compensation tied to the Cursor acquisition. And when an analyst asked how all this gets funded, the answer included operating cash flow, debt markets, and partnering with suppliers and customers ”across the value chain.” Anyone who did deals during the telecom boom, as I did, gets a little twitch when they hear about vendor and customer financing in the same breath as record CAPEX.
So what should investors take from all this? I won't tell you whether to buy SpaceX–but I will share the questions I'm asking myself. Is revenue growing faster than capital spending, or is the spending growing faster than the revenue? Does the third quarter's profitability survive once the one-time benefits roll off? Does the ”payback in under a year” math still work if rental prices cool, even a little? And if power and cooling–not money–are the real bottlenecks, who owns the bottleneck? Those questions don't just apply to SpaceX. Meta reports after the bell tonight, along with Tesla and IBM, and I'll be asking Meta the very same ones. The question for AI spenders is no longer whether they'll spend. It's whether the revenue shows up on the same schedule as the bills. Get used to that common theme. It isn’t going anywhere.
Which brings me back to the Orca. Spoiler alert for anyone who skipped 1975: the bigger boat never arrives. The Orca sinks. Brody wins anyway–clinging to a sinking mast with one rifle, his last bullet, and an absurd amount of luck. It makes for a terrific movie. It makes for a terrible investment plan.
SpaceX has a bigger boat on order, a $90 billion cushion, and a CEO who sees a trillion-dollar shark. I'm not betting against the boat. I'm just watching very carefully to see whether it gets delivered before the shark gets hungry. Musk will no doubt read the moment a bit differently. I suspect his updated version of Brody's line goes something like this: ”We're gonna need a bigger power plant.” 🦈😉
YESTERDAY’S MARKETS
Stocks finished mostly higher yesterday as strong third-quarter earnings and a pullback in oil offset weakness in tech shares. The Dow Jones Industrial Average jumped 530 points, or 1%, to 52,496, the S&P 500 rose 0.3%, the Nasdaq Composite slipped less than 0.1%, and the small-cap Russell 2000 gained 1.7%. The 10-year Treasury yield fell about 3 basis points to 5.12% after September housing starts dropped 5.2% to an annualized 1.221 million units, below expectations. WTI crude settled down 1.8% at $101.79 a barrel and Brent fell 0.7% to $108.79, as traders weighed diplomatic efforts to ease the conflict.
NEXT UP
No major economic releases today, but plenty of important earnings. Be on the watch for: AT&T, NextEra, Abbott Labs, Danaher, Elevance Health, Thermo Fisher Scientific, WABCO, Philip Morris International, IBM, CSX, Knight-Swift Transportation, TI, Southwest Airlines, Lam Research, Tesla, United Rentals, Kinder Morgan, Las Vegas Sands, and Meta.