After 24 years of operating in shadows, SpaceX finally opened its books. The nearly 400-page S-1 filing submitted to the SEC ahead of an anticipated June 12 IPO is a fascinating read, not because it confirms what we already suspected about the company’s rocket business, but because it reveals something far more ambitious: SpaceX is betting its future on becoming an AI company.
The numbers alone demand attention. SpaceX reported $18.67 billion in revenue for 2025, a healthy jump from $14.02 billion the year before. But here’s where things get interesting. The company lost $4.94 billion in 2025, largely due to artificial intelligence development spending. This isn’t the profitable space launch company Wall Street thought it knew.
The TAM That’s Hard to Believe
On page 171 of the filing, SpaceX makes a claim that reads almost like science fiction: “We believe we have identified the largest TAM in human history.” That total addressable market? $28.5 trillion.
To put that in perspective, roughly $2 trillion of that comes from space and Starlink operations. Everything else, $26.5 trillion, supposedly flows from AI. The company’s thesis here is that orbital AI compute represents the next trillion-dollar market, powered by its rockets and satellites.
It’s a seductive narrative. SpaceX has the launch capacity. It manufactures satellites. Elon Musk founded xAI. Why not connect all the dots into a constellation of orbital data centers?
But let’s pump the brakes. SpaceX’s own filing admits the estimates rely on projections from the RAND Corporation, combined with internal assumptions about how much global compute capacity will be allocated to AI workloads, power usage, and utilization rates. In other words, there’s a lot of guesswork embedded in that $26.5 trillion number. This is scenario planning dressed up as market analysis.
The risk isn’t that the number is too high. It’s that investors might believe it without understanding the gap between potential and execution.
From Space Company to Something Else
Here’s what makes this filing genuinely revealing: SpaceX is no longer primarily a launch company trying to expand into other sectors. It’s now a company with its fingers in launch, satellite internet, aerospace, and artificial intelligence, all ostensibly working in service of a much larger AI ambition.
The company states it aims to launch 100 gigawatts of compute to space annually, with orbital AI deployments beginning as early as 2028. That’s an extraordinary engineering challenge. It’s also a bet that the future of computing isn’t on Earth.
The filing acknowledges significant technological hurdles remain. Starship, the vehicle that would make this whole vision economically viable, still needs to complete a series of test flights and prove full reusability. The company wants to reduce Starship’s cost per kilogram to orbit to $185, but we don’t yet know if that’s achievable or when.
Who’s Really Running This
The governance structure tells its own story. After the IPO, Musk will retain 85.1 percent of combined voting power. He’ll serve as CEO and chairman. Removing him won’t be easy.
His salary in 2025 was $54,080, pegged to California’s minimum for exempt employees. Meanwhile, Gwynne Shotwell, the company’s president and COO, pulled in $1.08 million in salary but $85.8 million in total compensation when stock awards are factored in. The disparity is worth noting, if only because it underscores who’s really driving strategy at SpaceX.
The filing also hints at something less discussed: political risk. SpaceX notes that shifts in congressional composition or presidential administration could materially affect its future, given its dependence on government contracts and regulatory favor. Musk’s advisory role to President Trump adds another layer to that equation.
The Space Launch Business Nobody Talks About
Surprisingly, the filing reveals little detail about the actual economics of SpaceX’s core business. There’s no breakdown of the Falcon 9’s internal cost per launch, though industry estimates peg it around $15 million versus the $74 million public price. That margin has made SpaceX formidable, but the filing doesn’t dwell on it.
For Starship, the ambition is clearer but the timeline more uncertain. The company intends to launch V3 Starlink satellites in the second half of this year, assuming upcoming test flights go smoothly. But Starship still faces “significant technological, engineering, and operational challenges,” the filing admits, particularly in developing habitable environments and performing complex in-orbit operations for lunar and Martian missions.
The Leap From Here to There
SpaceX is making a specific claim: it’s the only company with a “commercially viable path to building orbital AI compute at scale.” That’s a bold assertion, and it might be true. But it’s also a bet on a market that doesn’t yet exist, using technology that hasn’t yet been proven, to solve problems that haven’t yet been fully articulated.
The company went from trying to land a small rocket to putting 80 percent of all mass into orbit annually. That track record commands respect. But pivoting from space launch to AI infrastructure is a different animal entirely.
Investors will have months and years to decide whether this is visionary or overconfident. What’s clear from the S-1 is that SpaceX’s leadership is no longer thinking like a launch company. The real question is whether the market will follow.
