SpaceX dropped a bomb on Wednesday. After 24 years of keeping its books locked away, the company filed an S-1 with the SEC, and what it revealed wasn’t just financial details. It was a thesis about what SpaceX wants to become, and honestly, it’s audacious enough to make you wonder if Elon Musk’s visionary streak has tipped into something else entirely.
The headline number is impossible to ignore: $28.5 trillion. That’s SpaceX’s stated “total addressable market” across space, data, and AI. But here’s where it gets interesting. Only about $2 trillion of that pie comes from space and Starlink. The other $26.5 trillion? That’s supposed to come from AI, primarily enterprise applications.
“We believe we have identified the largest TAM in human history,” the company states in the filing. That’s not corporate understatement. That’s a claim so grand it demands scrutiny.
The Math That Almost Makes Sense
SpaceX’s case for orbital AI compute isn’t pulled from thin air. The filing cites projections from RAND Corporation and layers in internal assumptions about global data center compute demand, utilization rates, and power usage. The logic is seductive: if you can launch massive computing capacity into orbit, you’ve got a global infrastructure play that doesn’t depend on terrestrial real estate.
The company says it aims to launch 100 gigawatts of compute to space annually by the time it gets rolling. That’s not a typo. And they’re targeting a 2028 deployment date for the first orbital AI satellites. For context, SpaceX put up $18.67 billion in revenue in 2025, up from $14.02 billion in 2024. But it posted a $4.94 billion loss, largely because of AI development spending. The company is already bleeding capital to chase this vision.
“We believe we are the only company with a commercially viable path to building orbital AI compute at scale,” SpaceX states in the filing. That’s a bold assertion, but it’s not entirely unfounded. The company launches roughly 80 percent of all mass into orbit annually. It operates more satellites than the rest of the world combined. No other company has demonstrated the launch cadence or satellite manufacturing scale that SpaceX does.
Still, there’s a difference between having the best infrastructure and actually building a $26 trillion business on top of it.
The Elon Effect and Political Risk
Here’s something the filing inadvertently highlights: SpaceX’s future is tethered to forces beyond its control. Musk will retain 85.1 percent of combined voting power after the IPO closes, making him essentially immovable. But the company also acknowledges that shifts in political power and regulatory posture could materially alter its prospects. The filing notes that Musk serves as an advisor to President Trump, and the document gently alludes to how changes in administrations could reshape government spending priorities and contract allocation.
Translation: SpaceX’s government business, which has been lucrative, isn’t guaranteed to stay that way. The company knows it, and investors should too.
What’s almost comical is Musk’s salary: $54,080 in 2025, pegged to California’s minimum for exempt employees. Gwynne Shotwell, the president and COO, pulled in $1.08 million in salary but $85.8 million in total compensation when stock awards are factored in. Musk doesn’t need salary. He owns the voting structure. It’s a different game entirely.
The Starship Gamble
Beyond AI, SpaceX still has fundamental engineering challenges to solve with Starship. The company aims to get the cost per kilogram to orbit down to at least $185, which would be extraordinary if achieved. But the filing is honest about the work remaining. These systems involve “significant technological, engineering, and operational challenges,” the company admits, including developing habitable environments and performing complex in-orbit operations.
Translation: We don’t have all the answers yet, and it’s going to cost a lot to find them.
SpaceX plans to begin launching V3 Starlink satellites on Starship in the second half of this year, assuming a series of test flights resume soon from South Texas. The company is simultaneously betting on improving launch economics while building out a completely new business vertical in orbital compute.
The Bet Everything Strategy
What strikes you about reading this filing is how SpaceX is attempting to rebrand itself. It started as a Business dedicated to making rockets cheaper and more reliable. It became the world’s dominant launch provider. Now it’s trying to evolve into something bigger: an AI infrastructure company that happens to own rockets.
That’s not evolution. That’s transformation. And transformations fail more often than they succeed, regardless of how much capital you throw at them or how visionary your CEO claims to be.
The company’s argument is compelling from a first-principles angle. If you can launch cheaply and reliably, and you can manufacture satellites at scale, then orbital compute is a logical extension. But logical arguments and market success aren’t the same thing. Amazon had overwhelming competitive advantages when it entered cloud computing. SpaceX has advantages in launch and satellites. That’s not the same thing as dominance in AI infrastructure, which is already crowded with well-funded incumbents.
What Investors Are Actually Buying
The IPO isn’t happening until at least June 12, according to the filing. By then, investors will need to decide whether they’re buying into a proven space company with strong fundamentals or a speculative play on a future that depends on solving problems that don’t have easy answers yet.
SpaceX’s track record in space is stellar. Its track record in AI, social media (via the xAI acquisition), and other ventures beyond launch and satellites is essentially nonexistent. Yet a significant chunk of the company’s projected value rests on those unproven verticals.
The filing also reveals that SpaceX is hardly profitable relative to its revenue. A company pulling in $18.67 billion in revenue shouldn’t be losing $4.94 billion annually, even if some of that is R&D spending on future capabilities. It suggests the company is either massively inefficient or burning cash at a pace that assumes a bet-the-company success is just around the corner.
Maybe it is. Maybe SpaceX really does build an orbital compute empire that generates trillions in value. Or maybe it turns out that AI compute infrastructure is better served by fiber-connected terrestrial data centers, and the company ends up spending billions to prove it.
The question investors face isn’t whether SpaceX is a good space company. It clearly is. The question is whether you believe it can become something it’s never been before, and whether that belief is worth betting on against the skepticism of seasoned capital markets.
