The reason isn’t simply that SpaceX will be fighting the law of large numbers by starting life as a public company as an extremely expensive stock. Put simply, as the prospectus highlights, Elon Musk’s creation has essentially re-invented itself from a commercial space pioneer facing relatively mild competition, to an AI-centric player that’s vying for the same dollars and customers, as the hyperscaler crowd led by Microsoft, Google, OpenAI, Coreweave, and sundry smaller but still formidable participants.

To win in that crowded and hot sector, SpaceX will need to go super-big on capex for data centers and R&D that hatches fresh enterprise products. As the prospectus displays, those already-huge expenditures are already accelerating, and they’ll keep ramping over the next few years. Yet garnering major profits from AI may take a lot longer. The S-1 makes that point as well. The original space businesses may prove highly successful, but it’s likely not big enough to do most of the work. It’s clear that Musk’s ambitions, and the investors’ hopes as reflected in the valuation, are heavily tilted to a knockout performance in AI.

To handicap SpaceX’s prospects, it’s crucial to ignore the Wall Street buzz and Musk hype about colonies on the moon and examine, via the prospectus, how much money SpaceX now pours into fashioning the AI franchise, and what it’s reaping from the space ventures to support it.

An excellent new report from David Trainer, CEO of financial research firm New Constructs, identifies several weaknesses threaten SpaceX’s prospects. They include a lopsided governance structure where the funds and individual will own almost 60% of shares but get almost no voting power. Instead, Elon Musk will exercise virtually total control; the founder and CEO can’t be removed from office by a shareholder vote and is free to name a board dominated by insiders. (This reporter addressed these issues in a previous story https://fortune.com/2026/05/22/space-x-stock-ipo-price-elon-musk-shareholders/.) Trainer also notes that SpaceX will make its public debut as the most unprofitable player in all of its main businesses.

Another red flag: Trainer dug into the S-1 to find that the lion’s share of the projected IPO proceeds is already spoken for. So the question arise, where will all this money needed for capex come from? The potential share issuance and borrowing needed to the AI march could prove a big negative for investors.

Of the two non-AI sectors, rockets are spouting losses while the satellite side’s thriving

As the S-1 shows, SpaceX stands on three main legs, Space, Connectivity and AI. All told, the consolidated enterprise posted $18.7 billion revenue and booked an operating loss of $2.6 billion. AI is the biggest drag, highlighting the challenge ahead. Space comprises the rocket lineup that the company manufactures in-house, and deploys to launch its own satellites. It also sells rockets to NASA and performs launches as a contractor for the agency, as well as hosting special orbital trips for high-paying VIPs.

By contrast, the Connectivity segment’s a big money-spinner that boasts an outstanding runway. It’s SpaceX’s only profitable franchise, and accounts for almost two-thirds of total sales. The Starlink segment runs a galaxy of 9,600 satellites, three-quarters of the total fleet in orbit. Over ten-million users pay subscriptions for mobile and broadband service. The business is well protected, since it’s far the biggest global player in commercial satellites, and the gigantic investment and tech wizardry needed to challenge its dominance provides a deep moat. It’s also annuity-style steady.

In the 12 months ended in Q1, Starlink doubled its roster of subscribers to reach that 10 million mark. The downside: The new customers it’s adding are getting less profitable. Its revenue per sub, a key metric, has fallen from $99 in 2023 to $66 in Q1 of this year. As a result, revenue growth lagged, rising 50% in 2025 vs last year to $11.4 billion. So far, the segment’s extremely profitable, returning $4.4 billion in operating income last year for a margin of 30%. Still, it’s facing strong pricing competition from terrestrial networks, operated by everyone from T-Mobile to Google fiber that users often find more reliable than satellite service. The overriding issue: Connectivity’s subscription machine, even if it keeps growing fast, isn’t nearly big enough to drive the kind of revenues and profits SpaceX will need to reward shareholders.

A big concern for shareholders: The IPO proceeds won’t be available from funding most of its AI capex. So where will the the tens of billions a year come from?

Instead, that burden falls to AI. It’s just recently that SpaceX recast itself as a giant in the hottest tech arena of this century. The shift came in February, when it merged with Musk-controlled xAI, a combo that reportedly boosted SpaceX’s private valuation by $250 billion. AI’s also responsible for the lion’s share of the deficits: In the past five quarters, it’s collected $4 billion in revenues and suffered over twice that amount, $8.9 billion, in operating losses.

As Trainer observes, the losses from Space, and especially AI–despite Starlink’s success––place SpaceX as a whole dead last among all of its peers in broadband and mobile and in AI, measured by both profitability and return on invested capital. It scored -7% and -3% respectively in those categories last year, far behind Comcast (12% and 6%), AT&T (17% and 4%), Amazon (11% and 14%), and even Coreweave (10% and 1%).

That company-wide deficit in profitability could potentially raises questions about it will fund its gigantic requirement for the capex required to deliver on its promise make its AI side a supreme winner.

SpaceX is rapidly expanding at its portfolio of data centers newly acquired from xAI. The flagships are the aptly-named Colossus I and II facilities in Memphis covering a total of 2 million square feet. It’s also spending $20 billion on a new hyper-scale center in Mississippi. Since the start of 2025, the AI side has lavished $20.4 billion on infrastructure, two-thirds of the SpaceX’s total capes over that span. In Q1, the prized segment’s bill came to a staggering $7.7 billion. Last year, AI absorbed the equivalent of 60% of the overall enterprise’s R&D, and the number for Q1 hit $3.5 billion, twice the dollars spent a year ago.

The math suggests that Musk is marshaling his genius for hype, as well as legendary reputation as a visionary, to create a highly-overvalued stock. Hence, the IPO’s generated such seldom-seen excitement that it’s expected to raise $80 billion, yet will require SpaceX to sell only around 5% of its shares. It would appear, then, that SpaceX will be amass warchest sizable enough to fund a few years of AI investment on its down.

But as Trainer spotlights, that’s not the case. As he points out, SpaceX has pre-pledged $62.8 billion, 78% of the expected proceeds, for payments to third parties. Almost exactly one-third each will go to Valor Equity Partners, a major early investor, Musk’s X Corp. and xAI creditors holders for repayment of debt, and Echostar for the “Spectrum Acquisition Closing.” That leaves less than $18 billion to devote towards growing SpaceX, principally by financing the promised explosion in AI compute capacity. Keep in mind that the AI area devoured more than amount just in the past five quarters.

The problem: $18 billion won’t last long, given the escalation in the AI spend, not just on capex, but operating expenses and R&D. It’s clear from the S-1 that the free cash flow from the rest of the company can fund just a piddling portion of what’s needed. By Fortune‘s estimate, the rest of SpaceX outside of AI last year generated just $1 billion free cash flow that could potentially support what Musk sees as its principal engine. The prospectus also talks about floating more shares and issuing debt to keep data center build rushing forward. But those moves will cost shareholders in dilution and rising interest expense.

Perhaps the most important statement in the entire, roughly 400 page document comes on page 53, where SpaceX details the immense expense and long timeline needed to mine the super-rich AI lode. The prospectus states, “We expect to allocate substantial capital to expand our compute infrastructure, and we expect a multi-year investment horizon before these deployments translate into sustained positive AI Segment Adjusted EBITDA. During this investment period, our capital expenditures will scale as quickly.” Keep in mind that’s “scaling up” from capex that in Q1 alone, reached a towering $7.7 billion.

The effusive document also reveals how heavily SpaceX is leaning on its new AI arm to achieve the explosive growth in revenues and profits needed to reward investors buying in at a $1. 5 trillion valuation. Page 11 displays one of the S-1 chief set-pieces, a chart showing the total addressable markets for each of its three segments. SpaceX projects its total TAM at a staggering $28.5 trillion. The corker: Of that total, AI accounts for $26.5 trillion or 93%. As Trainer writes, “Large TAMs provide strong growth potential. They also invite competition.” SpaceX is moving from relying on a satellite business it dominates to a field that’s lured a pantheon of the world’s most successful enterprises, from Microsoft to Google. The pie will expand fast, but sundry rivals will be vying for the pieces––a scenario that’s sure to pressure prices and margins.

Trainer’s a master of using discount models to forecast how much companies must earn in the future to justify huge market caps today. His analysis posits that to deliver decent returns to shareholders at a $1.5 trillion valuation, SpaceX would need to be booking $189 billion in annual profits by 2035. At $1.75 trillion, the bogey rises to $245 billion. For 2025, no U.S. company came close to even the lower number. As Trainer avows, Musk is counting on “Out of this World Profits” to ring the bell. And keep in mind, SpaceX is debuting as an enterprise that’s in the red.

If investors should focus on one item in the S-1, it’s not the gauzy stuff about solar powered data centers in orbit, but Musk’s frank admission that making money in AI will cost a ton and take a long time. Folks and funds may be right to back him. But though the spirits over the most eagerly awaited IPO are super-high, the risks are just as big.