A big contributor to the soaring spirits surrounding the pending SpaceX IPO is near-certainty that the offering will raise $80 billion or more in fresh capital. But the S-1 for the Space, Connectivity and AI enterprise released the evening of May 20th contains a surprising revelation. SpaceX has already pledged the largest part by far of what’s expected to be the largest sum ever raised in an IPO to third parties. That arrangement could prove a major negative for the folks and funds clamoring to get shares. The reason: CEO Elon Musk is essentially betting the future on gigantic growth in the new AI franchise. Of the total addressable market of $28.5 trillion he forsees for the company in the S-1, $26.5 trillion of that is in AI. That’s 13x his estimated TAM for what were until recently its two core businesses, mobile and broadband service provided by its constellation of satellites, and making and selling rockets, as well as providing flights for NASA and other customers.
SpaceX entered AI in a big way when it combined with Musk-controlled xAI in February. Overnight, it basically reinvented itself as a hyperscaler focused on rapidly building compute capacity. Indeed, the shift opened a gigantic new market to SpaceX. But it also came at a heavy price: The requirement for capital expenditures to build out such mega-projects as its Colossus I and II data center in Memphis covering two million square feet. In the past five quarters, the AI side has devoured over $20 billion in cash to fund its buildout, two-thirds the total for all of SpaceX. In Q1 alone, the number doubled from a year ago to $7.7 billion. The S-1 states that the AI investment budget will ramp rapidly from there.
At first glance, it appeared that the huge take from the IPO would comfortably fund those requirements for at least a couple of years. Not so as it turns out. As David Trainer of research firm New Constructs notes, the S-1 reveals that $62.8 billion or 78% of the forecasted $80 billion is already spoken for by insiders and vendors. Specifically, SpaceX pledges to pay that more than three quarters of the proceeds to third-parties, Valor Equity Partners (a large shareholder), Musk X Corp. and xAI investors for repayment of debt, and Echostar for “the Spectrum Acquisition Closing.”
That leaves less than $18 billion to fuel the AI express. Although the Starlink mobile and broadband subscription sector mints money, SpaceX as a whole, even outside of AI, is posting losses. I estimate that its other two businesses contribute just $1 billion or so in free cash flow, a mere dribble of what the AI engine requires. The $18 billion in IPO proceeds plus the $1 billion in free cash flow from the other two businesses combined wouldn’t even cover the $20 billion the AI side spent on plant, property and equipment last year, and that figure is waxing fast.
As a result, SpaceX will need to look elsewhere for cash. The S-1 states that it will fund expansion by floating new, post-IPO shares and raising debt. That’s a drag for shareholders. The need for fresh, outside funding will dilute the investors buying at the debut, and raise interest costs, curbing profitability.
As Trainer points out, huge TAMs offer big growth opportunities. But they also attract competition. SpaceX will be vying for slices of a fast-expanding pie with the Microsofts, Googles and other titans that constitutes perhaps the toughest roster of rivals on the planet. In the S-1, Musk acknowledges that SpaceX will make years of huge investments in AI before the franchise turns highly profitable. SpaceX would potentially become the huge money-maker Musk promises a lot quicker were it free to spend the entire $80 billion that’s generated so much buzz on its epic AI campaign. The revelation that the funds are mainly pledged before they’re collected makes an IPO that already looked dangerous before those details emerged the diciest of bets.
