SpaceX prices its IPO on Friday. Demand, according to people familiar with the book, is approaching four times oversubscribed, with some institutional investors submitting orders for $10 billion blocks. The offering price is $135 per share. The implied valuation is approximately $1.77 trillion. And the specific number that shapes how all the other numbers should be read is $63. That is the fair value per share that Morningstar’s equity research team assigns to SpaceX, against an offering price of $135. The difference, $72 per share, is what Morningstar explicitly calls a call option on the company’s ability to deliver orbital data centers at the rate and capability Elon Musk has described. You are not paying $135 for what SpaceX is today. You are paying $63 for what it already is, and $72 for what Musk says it will become. Aswath Damodaran, the NYU finance professor whose corporate valuation work carries unusual market authority, assigns a value of approximately $1.2 trillion. Even his more charitable number is $570 billion below the offering valuation.
The orbital data center plan is the center of gravity for the call option. Musk, in a video published by SpaceX this week, described his expectation that the company could reach roughly one gigawatt per year of space AI compute by the end of next year, then eventually scale to a terawatt. The math behind one gigawatt requires producing approximately 6,666 AI satellites per year, or about 556 per month, based on Musk’s expected power delivery of 150 kilowatts per satellite. For comparison, SpaceX’s current reported Starlink satellite production rate is approximately 70 per week. The AI satellites are described as architecturally simpler, but the production facility to build them at that rate has not yet been built. Then there is Terafab, the chip foundry SpaceX intends to operate, which Musk acknowledges is one of the hardest modern industrial projects any company can undertake. And underpinning all of it is Starship – the vehicle without which none of the orbital infrastructure goes to orbit. Those three interlocking engineering challenges are where the $72 premium lives, and they are the challenges NewsTrackerToday follows the money on as the actual bet public investors are making.
Daniel Wu places the ambition in historical perspective: “The combination of a heavy-lift reusable rocket, a domestic chip foundry, and an orbital data center network has never been attempted by a single company simultaneously. The closest analogue in industrial history is probably the integrated vertical model that AT&T ran from manufacturing through Bell Labs through network operation during the mid-twentieth century, and even that didn’t require building the physical infrastructure in orbit. What Musk is proposing is genuinely unprecedented in scope, which makes it genuinely difficult to value. The people paying $135 are pricing in a version of this that mostly works. The people paying $63 in Morningstar’s model are pricing in a version that is mostly worth what it already earns.”
The S-1’s market sizing is the number that earns the most skeptical reading in the prospectus. SpaceX describes its largest opportunity in enterprise AI, framing the total addressable market at $22.7 trillion, against $2.4 trillion for AI infrastructure and just under $2 trillion for space. That sequence, where the AI enterprise opportunity is listed as nearly ten times the company’s actual core space business, reflects the xAI integration and Musk’s stated intention to compete with OpenAI and Anthropic in model-building. But this sits alongside existing contracts to sell compute to both Anthropic and Google, at $1.25 billion and $920 million per month respectively. Selling compute to your stated AI competitors is not out of place for Musk, but it raises a specific question about where SpaceX expects value to accrue in the AI stack, and it is the question that NewsTrackerToday puts plainly because the S-1 does not answer it: is SpaceX building a frontier model business or a compute provider business, and can it credibly be both?
Starship’s current status adds a physical constraint to the financial ambiguity. The most recent test flight went well enough, but the booster stage failed to make a controlled reentry as planned, and SpaceX is still undergoing an FAA mishap investigation. The company has not publicly stated when the vehicle will fly again. NASA, which has a contract worth nearly $4 billion to use Starship as a Moon lander for Artemis missions, is not ready to commit to a crewed test with the vehicle before late 2027 at the earliest. SpaceX has said it expects to begin launching Starlink satellites with Starship by the end of this year. Both of those timelines, alongside the satellite and chip production targets, are where the orbital data center schedule gets compressed into a very tight window, and it is the window NewsTrackerToday returns to: if Musk’s annualized one-gigawatt target requires the production and launch rate he described, and Starship’s current reusability timeline is murkier than the company’s public communications suggest, the $72 call option depends on at least three concurrent engineering miracles arriving on schedule.
The uncomfortable conclusion Friday’s pricing forces is not that Musk cannot pull this off. He has pulled off improbable things before, and the space launch business he has already built would alone justify a valuation substantially above zero. The uncomfortable conclusion is that the $135 price is asking investors to finance the orbital data center, Terafab, and Starship reusability simultaneously, on a timeline that the company’s own engineering constraints suggest is optimistic, in a market environment where competitive AI model development is moving faster than any single company’s plans. News Tracker Today names the specific dynamic that the four-times oversubscription obscures: when a book is that oversubscribed, investors are no longer pricing the company. They are pricing the fear of being the one who didn’t own it.
