More than 90 percent of the growth SpaceX outlined in its IPO prospectus comes from AI, not from rockets or satellites.
In the summer of 2026, SpaceX stepped into public markets carrying a $2 trillion valuation and a quiet confession buried in its prospectus: more than nine-tenths of its projected growth rests not on rockets, but on artificial intelligence. Goldman Sachs has given that wager a number — $322 billion in AI revenue by 2030, a hundredfold rise from today — and the arithmetic, if it holds, would transform a half-million-dollar stake into $1.6 million. Yet the distance between a compelling forecast and a delivered future is precisely where the most consequential human stories tend to unfold.
- SpaceX's IPO revealed that its entire growth thesis is concentrated in an AI division that barely existed at launch, making the company's future unusually dependent on a single, unproven bet.
- Goldman Sachs projects a 100-fold revenue leap to $322 billion by 2030, a figure so large it strains credibility and has already drawn sharp public skepticism from veteran investors like Jeremy Grantham.
- The company is deploying IPO proceeds and $25 billion in bonds into AI compute centers, chip manufacturing, and orbital data facilities — a capital-intensive buildout with no guarantee of return.
- SpaceX remains unprofitable, faces near-certain future share dilution, and must execute an infrastructure challenge — orbital data centers, chip fabrication, space operations — that no company has attempted at this scale.
- The projection lands as a conditional promise: if AI demand holds, if competitors don't surge ahead, and if markets value SpaceX like Nvidia, the math works — but each 'if' carries the full weight of the bet.
SpaceX entered public markets in 2026 at a $2 trillion valuation, but the number that quietly defines the company's future isn't found in its rocket launch manifests — it's buried in the AI projections that account for more than 90 percent of its outlined growth. The space business is almost incidental to the story SpaceX is now telling investors.
To fund that story, the company raised capital through its IPO and a subsequent $25 billion bond offering, with management directing most of it toward AI infrastructure: compute facilities, chip manufacturing, and orbital data centers. The underlying logic is that building the hardware backbone for large-scale artificial intelligence could generate returns that dwarf anything the launch business ever could.
Goldman Sachs, which underwrote the IPO, recently attached numbers to that logic. The bank forecasts SpaceX's AI revenue growing from $3.2 billion to $322 billion by 2030 — a hundredfold increase in four years. Applying Nvidia-style valuation multiples to that figure produces a projected company worth of $6.3 trillion, and a theoretical tripling of a $500,000 investment by decade's end.
The calculation travels fast in investor circles, but the caveats travel more slowly. SpaceX is not yet profitable and will likely need to issue more shares, diluting existing holders. Critics including GMO co-founder Jeremy Grantham have argued the prospectus promises are simply undeliverable. The AI sector itself may be overvalued, and the operational challenge of building orbital data centers while running a space launch company simultaneously has no precedent.
What Goldman Sachs is ultimately offering is a map of the best possible outcome — one where AI demand sustains, competition doesn't accelerate, and markets remain generous. Whether the territory matches that map is the only question that will matter.
SpaceX went public in 2026 with a $2 trillion valuation, and the company's future hinges almost entirely on a bet that few investors fully understand: that its artificial intelligence division can grow into something genuinely transformative. More than 90 percent of the growth SpaceX outlined in its IPO prospectus comes from AI, not from rockets or satellites. That's a remarkable concentration of hope in a single business line that barely exists yet.
The company raised capital through both its initial public offering and a subsequent $25 billion bond offering, and management has signaled it will deploy most of that money into AI infrastructure—compute centers, chip manufacturing, orbital data facilities. The logic is straightforward: if SpaceX can build the hardware and software backbone for large-scale artificial intelligence, the returns could dwarf anything the space business generates. But the question hanging over all of this is whether the math actually works.
Goldman Sachs, which served as an underwriter for SpaceX's IPO, recently published a forecast that gives some shape to what "works" might look like. The bank projects that SpaceX's AI revenue, which totaled $3.2 billion last year, could grow to $322 billion by 2030. That's a hundredfold increase in four years. If that happens, and if the market values SpaceX's AI business the way it currently values Nvidia—at roughly 19.7 times sales—the company's total worth could reach $6.3 trillion. A $500,000 investment made today would theoretically be worth $1.6 million by the end of the decade.
It's a seductive calculation, and it's easy to see why it circulates among investors hungry for the next big thing. But the forecast comes with caveats that deserve serious weight. SpaceX is not yet profitable. The company will almost certainly need to raise more capital in the coming years, which means issuing additional shares and diluting existing shareholders. The market may not value SpaceX the way it values Nvidia—the businesses are fundamentally different, and some observers believe the entire AI sector is in a bubble, with valuations disconnected from reality. Jeremy Grantham, the co-founder of GMO LLC, has been blunt in his skepticism, arguing that SpaceX will "fail to deliver anything like its promises in the prospectus."
There is also the matter of execution. Building orbital data centers and manufacturing chips at scale while simultaneously running a space launch business is not a solved problem. The company has never done it before. Timing matters too—if AI demand cools, or if competitors emerge faster than expected, the entire projection collapses. What Goldman Sachs is really saying is: if everything goes right, and if the market cooperates, SpaceX could be worth vastly more than it is today. But everything going right is not the same as everything going right.
Notable Quotes
SpaceX will fail to deliver anything like its promises in the prospectus— Jeremy Grantham, co-founder of GMO LLC
The Hearth Conversation Another angle on the story
So Goldman Sachs is saying SpaceX's AI revenue could grow a hundredfold in four years. Does that number feel real to you, or is it the kind of forecast that sounds impressive in a spreadsheet?
It's both. The company is starting from $3.2 billion, which is real money but still small relative to the total valuation. Getting to $322 billion would require SpaceX to capture a meaningful slice of the AI infrastructure market—the chips, the compute, the data centers. That's not impossible. But it requires flawless execution in a space where competition is intensifying.
And the $1.6 million figure—that assumes the market will value SpaceX like Nvidia. Why would it?
That's the real gamble. Nvidia is a pure-play semiconductor company with decades of execution history and consistent profitability. SpaceX is a space company that's trying to become an AI company while still losing money overall. The market might reward that ambition, or it might demand a discount. There's no way to know until it happens.
What's the thing that keeps you up at night about this forecast?
Dilution. SpaceX will need capital to build all this infrastructure. That means more stock offerings, which means existing shareholders get a smaller piece of a larger pie. The math in Goldman's forecast doesn't really account for how much new equity the company might have to issue.
So even if the AI business grows exactly as projected, shareholders might not see the gains they expect?
Exactly. The business could succeed and the stock could still disappoint, because the gains get spread across more shares. It's a subtle but crucial distinction that gets lost in the headline numbers.