A company that lost $4.94 billion last year is asking public markets to value it at $1.77 trillion.
That is not a typo. Space Exploration Technologies Corp. will sell 555.6 million shares at $135 apiece, raising up to $75 billion — the largest IPO in history, dwarfing the $26 billion Saudi Aramco raised in 2019. At that price only six companies in the S&P 500 are worth more.
So what, exactly, are buyers paying 107 times sales for?
Last week we walked through SpaceX's first public financials — the revenue-versus-profits tension nobody had seen before. This week the actual S-1 has landed, the valuation target is set, and the sector is already bleeding. The prospectus answers the question. It just doesn't answer it the way the mission statement wants you to think.
Start with the arithmetic nobody disputes. In 2025 SpaceX booked roughly $18.7 billion in revenue and a $4.94 billion net loss — a hard swing from $791 million in profit the year before, reportedly driven by the xAI merger and infrastructure buildout. Priced against that top line, the IPO carries a price-to-sales ratio of about 107.
Now do the subtraction that matters. Strip out the AI narrative and value the two businesses everyone can actually price — Starlink and launch — and even the most optimistic estimates top out near $300 billion. That is under 20 percent of the $1.75 trillion target.
Which means more than 80 percent of the largest IPO ever rests on two things that don't yet exist: orbital AI data centers and a fully reusable Starship.
More than 80 percent of the largest IPO ever rests on two things that don't yet exist: orbital AI data centers and a fully reusable Starship.
The launch business is real, and it is dominant. Since 2023, SpaceX has carried more than 80 percent of the world's orbital payload mass, at a mission success rate above 99 percent. Falcon 9 booster 1067 just flew its 35th mission. That pedigree is genuinely hard to argue with.
Starlink is real too — reportedly more than 10,000 active satellites, over 10 million subscribers, coverage across 164 countries. It is a profitable, cash-generating business that now dwarfs the launch operation that spawned it.
But that is the $300 billion. Everything above it is a promise.
The boldest promise is orbital compute: a constellation of up to one million satellites running AI workloads on solar power, cooled by the vacuum of space. Days before the IPO, Musk unveiled the first detailed design of the prototype, the AI1 satellite. SpaceX plans to begin deploying orbital AI computing satellites from 2028.
That is the pitch. It is engineering that hasn't happened. As Caleb Henry of Quilty Space puts it, SpaceX went from launch to satellite internet, and "this next evolution is to become an AI company" — but even he adds, "I can't tell you if it scales efficiently to an orbital data center."
The second promise is Starship itself — the fully reusable vehicle Musk is counting on to reach Mars, and the one NASA needs for a crewed lunar landing reportedly targeted for September 2028. Today Falcon 9 recovers its booster and fairings but throws away the upper stage, which puts a hard floor under how cheap a launch can get. Starship is meant to break that floor. It hasn't yet.
Look at where the money is actually going and the reclassification becomes explicit. In the second quarter, AI ate roughly 86 percent of total capital expenditure of $18.37 billion. The AI segment reportedly brought in $2.56 billion in revenue in Q2 2026, up 247 percent. This is not a rocket company that dabbles in compute. It is a compute company strapped to a rocket company's balance sheet.
Wall Street sees it clearly, and it doesn't hide the ball. Morgan Stanley keeps its official target at $300 — but frames the whole range around AI: a $75 bear case if AI monetization and Starship both slow, a $300 base, and a $600 bull case.
What hangs the bull case? Not rockets. It is SpaceX's pending $60 billion acquisition of Anysphere, maker of the AI coding platform Cursor, which Morgan Stanley expects to reach $8 billion in annual recurring revenue by the end of 2026 and about $33 billion by 2030. The upside lives in a coding startup, not a launch pad.
Here is where the story turns from a valuation debate into a sector event. SpaceX's gravity is not lifting the field. It is draining it.
On Monday, June 1, space stocks that had rallied on the IPO news gapped down and closed lower for a second straight day. Rocket Lab fell more than 15 percent. Intuitive Machines dropped about 13 percent. AST SpaceMobile fell nearly 7 percent, tripping a volatility halt intraday. The Procure Space ETF gave back more than 10 percent from its 52-week high.
Two triggers, per Interactive Brokers' Steve Sosnick. First, Blue Origin's New Glenn exploded during a static-fire test on May 28, likely wrecking the roughly $1 billion LC-36 pad and setting Bezos's program back a year. "Today was a tough day," Bezos posted. Sosnick called it "a cold shower on an already very hot, even overheated, sector."
Second, the report that SpaceX had trimmed its own IPO target from above $2 trillion to $1.8 trillion. Musk denied it — but as one strategist read it, even industry insiders behaving as if valuations had "approached the ceiling" was itself the signal.
Then comes the mechanical part, the wrecking ball. Analysts expect investors to sell existing space names to free up cash for SpaceX once it lists — a "siphoning effect," in Eric Diton's phrase, hitting an already-tired sector a second time.
This isn't hypothetical. Goldman's John Flood notes that ahead of every major US IPO in recent decades, equity mutual funds have raised cash levels — and large mutual and passive funds had already begun hoarding cash and preparing to trim big-cap holdings to make room for SpaceX and OpenAI. The raise pulls capital out before a single share trades.
Which brings us to the tell — the disclosures about the dream itself.
The mission statement on the prospectus's opening pages is pure Musk: "to build the systems and technologies necessary to make life multiplanetary… and to extend the light of consciousness to the stars." Lunar bases. Mars settlements. Asteroid mining. Space "real estate."
And yet, in the actual S-1, all of it appears only as brief, hedged, lawyer-scrubbed language — space tourism, on-orbit manufacturing, cargo to the Moon and Mars mentioned in passing, the property-rights question skirted entirely. The 1967 Outer Space Treaty bars nations from claiming celestial bodies; whether private firms can is unsettled, and SpaceX's counsel plainly chose not to plant a flag on it.
Read the document as a document and the priorities invert. The colonization vision — the thing that animates the mission, the thing retail investors think they're buying — is treated as risk, not asset. The AI capex is the concrete, recurring, dollar-denominated ask.
That is the danger in a sentence. The Mars dream is the marketing. The orbital data center is the bill. And the gap between them is now wide enough that merely announcing the raise has started pulling money out of the very sector SpaceX was supposed to validate.
The launch dominance is real and profitable. Whether it is worth $1.77 trillion depends entirely on engineering that, as of the filing, has not yet happened.