SpaceX is set to hit the market on Friday, and investors can’t contain their excitement. The $75 billion stock offering is reportedly significantly oversubscribed, with some institutional investors snapping up $10 billion blocks of Elon Musk’s empire.
There are plenty of reasons to be skeptical about this investment — big IPOs tend to sink, companies are losing money, and Mr. Musk’s erratic online behavior would be scary to other tech CEOs — but no one seems to be holding back. Tech investors have learned to never bet against Elon, no matter what the business logic tells us.
But a sober look at SpaceX’s financial plans reveals a lot about what they’re betting on. It’s an orbital data center-centered business that emerged over the past 18 months as Mr. Musk sought a vision to unite the conglomerate ahead of an IPO.
In true Musk fashion, it’s a bold plan, requiring at least three near-impossible feats of engineering. It’s reusable rockets, a state-of-the-art American chip manufacturing plant, and a sprint to build satellites faster than ever before.
This type of business plan can be difficult to score. This week, two analyzes sought to offer a more sober assessment of SpaceX’s plans. One by financial research firm Morningstar and the other by Aswath Damodaran, a New York University finance professor with a special interest in corporate valuations. Both exercises found SpaceX’s value to be significantly lower than the roughly $1.8 trillion valuation offered by the company’s bankers. Morningstar assigns a value of about $825 billion, while Damodaran suggests the company is worth $1.2 trillion.
This big difference is, in many ways, the result of bolting a world-dominant space monopoly into the much riskier AI business. Morningstar analysts characterize the difference between their valuation of $63 per share fair value and SpaceX’s offering price of $135 as a $72 call option on the company’s ability to deliver an orbital data center at the speed and capacity Musk believes possible.
Both analyzes show that the high margins of the company’s space launch business and satellite internet network are its most attractive features, while its AI business is the most uncertain.
To cloud or not to cloud?
Part of the question is: What is SpaceX’s AI business? The company’s S-1 market analysis frames the biggest opportunities in enterprise AI. The company’s model is that it will augment coding tools built by teams acquired from Cursor, or the company’s Macrohard project, which aims to give digital agents the ability to perform white-collar labor. SpaceX valued the total market for its business at $22.7 trillion. By comparison, AI infrastructure is worth $2.4 trillion, and the company’s space efforts are just under $2 trillion.
But this contradicts the company’s recent deals to sell large amounts of computing to Anthropic and Google, its ostensible competitors in the modeling business. That wouldn’t be out of place for Musk’s company. SpaceX frequently launches satellites operated by competitors onto its Starlink network. It’s usually just doing it from a place of power rather than playing catch up.
Working like a neocloud may be good business in the short term, but it begs the question of where the value lies in the AI technology stack. If you can’t be both, is it better to be a compute provider or a model builder?
The scaling logic that governs the AI business requires serious frontier labs to constantly train new, more powerful models (or, as Musk admitted in his recent lawsuit against Sam Altman, by extracting features from other companies’ models). Competitors that don’t move forward are likely to fall behind, but the increasing capabilities of cheaper open source models could undermine that dynamic.
Space data centers are one way to square the circle, providing so much computing that SpaceX can effectively do both.
Musk’s space data center architecture
In a video interview released by SpaceX this week, Musk explained the logic behind why SpaceX is best positioned to serve data centers. The crux of the argument was that SpaceX is the only company that can cheaply put a lot of mass into orbit, make a lot of solar panels, and make a lot of chips. Industry experts generally believe that large-scale space data centers are about a decade away, but Musk insisted (with a number of caveats) that they are much closer.
“This is not a promise of what we will do,” Musk said in the video. “This is what we’re trying to do, and we think we can probably do, in terms of space AI computing, is to reach a rate of about 1 gigawatt per year by the end of next year.”
Based on his expected maximum power delivery of 150 kW per satellite, that would be a production rate of 6,666 satellites per year, or about 556 satellites per month. This is approximately twice the reported current production rate of Starlink satellites (just 70 per week). Musk has said that the architecture of AI satellites is simpler, but that’s a lot to ask of production facilities that haven’t yet been built. The company is currently proceeding with the construction of a solar panel production facility.
That’s before you even get to Terafab, the company’s much-discussed chip foundry, which Musk sees as a late-stage product launch as the company looks to scale annual computing production to terawatts. Chip fabs are among the most difficult modern industrial projects, typically costing billions of dollars and taking as long as 10 years to construct.
Now comes the most important question. What will happen to Starship, which is key to SpaceX’s ability to economically get all these chips into orbit?
Recent test flights went well enough, but they didn’t suggest that rapid reuse is around the corner. SpaceX may initially end up reusing only boosters, which would increase the cost of deploying space data centers. Currently, the company is still undergoing an FAA accident investigation to determine why the booster stage failed to perform a controlled reentry as planned. SpaceX did not answer questions about when the rocket would fly again, but said it plans to begin launching Starlink satellites by the end of this year.
However, please take that with a grain of salt. Consider that NASA has a nearly $4 billion contract with SpaceX to use Starship as a lunar lander, but it’s not yet ready to tackle the spacecraft’s test mission, scheduled for late 2027.
Buyer beware
If retail investors bought SpaceX stock, they would have a near-monopoly bet on U.S. and European access to space, global communications networks and the most ambitious infrastructure projects of the AI era.
These projects rely on SpaceX creating something never seen before: a fully reusable rocket. The company will also need to build a high-speed production facility for AI satellites, which will be completed in 18 months instead of the 10 years it took for Starlink’s manufacturing development. Finally, there is a need to build chip foundries in the United States, something even silicon specialist companies are reluctant to undertake. Musk is right that SpaceX is the only company capable of building these anytime soon, but that says as much about the challenge as the company’s chances of achieving it.
Mr. Musk has always said he would not take SpaceX public until it reaches Mars because fickle investors could lose confidence along the way. Those plans may have been postponed, but the plans he laid out ahead of the company’s IPO could be just as difficult.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
