SpaceX IPO: The Numbers Behind a $1.75 Trillion Gamble

SpaceX Is Asking Investors for $1.75 Trillion. Here's What the Numbers Actually Show.

When a company files an S-1 prospectus with the SEC, the document is supposed to answer one question clearly: why is this business worth what we're claiming it's worth? SpaceX's IPO filing answers that question — just not in the way most investors might expect.

The company generated $18.7 billion in revenue in 2025. At a proposed valuation of $1.75 trillion, that's a price-to-sales ratio of roughly 94x. For context, Apple — one of the most profitable companies ever built — trades at around 8x revenue. Amazon, famous for years of razor-thin margins, rarely exceeded 4x. SpaceX is asking to be valued like a company that has already won markets it hasn't yet entered, on the basis of businesses that, in several cases, don't yet exist.

This is not a story about whether Elon Musk is visionary or reckless. It's a story about what the prospectus actually says — and what serious investors should make of it.

---

The Financial Reality Behind the Rocket Photographs

Let's start with the income statement, because it tells a story the 14 pages of rocket photography at the front of the filing do not.

In 2025, SpaceX posted a net loss of $4.94 billion. That figure is striking on its own. What makes it remarkable is the context: Starlink, the satellite internet division, generated $4.4 billion in operating profit. That means every other part of the business collectively burned through more than $9 billion in a single year.

Break it down by segment for Q1 2026 and the picture sharpens further:

• Space division (rockets, launch): -$662 million operating loss
• Connectivity division (Starlink): +$1.1 billion operating profit
• AI division: -$2.5 billion operating loss

One division makes money. Two don't. And the one that makes money is being used to fund the ones that don't, while management pitches investors on a valuation built around the losing segments.

Since its founding, SpaceX has accumulated over $37 billion in cumulative losses — more than any company that has ever gone public in history. Capital expenditure hit $20.7 billion in 2025, up fivefold from 2023, driven by Starlink expansion, Starship development, and AI infrastructure. To fund the operation, the company is carrying $29 billion in debt, including a $20 billion bridge loan taken out just weeks before the IPO filing. A meaningful share of the $75 billion being raised from public investors will go straight to repaying those lenders — not to building Mars colonies or orbital data centres.

Revenue is growing, to be fair — somewhere between 33% and 43% annually, which is genuinely strong. But that growth is almost entirely Starlink, and even Starlink's numbers carry a caveat: average revenue per user has dropped around 18% since 2023 as the company discounts aggressively to build subscriber numbers ahead of the listing.

---

SpaceX Claims It's an AI Company. The Data Says Otherwise.

The most consequential repositioning in the prospectus is this: SpaceX now describes itself primarily as an AI company. According to the filing, 93% of the company's total addressable market is attributed to artificial intelligence. Roughly 60% of all capital expenditure is now directed toward AI infrastructure.

The total addressable market claim is $28.5 trillion — larger than the entire US GDP, and built almost entirely on businesses that do not currently exist in any meaningful commercial form.

Grok, SpaceX's AI product through its XAI subsidiary, holds approximately 3.4% market share in the AI space, according to Bloomberg data. The filing itself concedes that the enterprise AI market is dominated by OpenAI, Anthropic, and Google. SpaceX's own engineers have reportedly been slow to adopt Grok for technical work because it underperforms competing tools. Musk acknowledged the problem publicly, stating the AI code needed to be rebuilt from scratch — shortly after completing a $250 billion acquisition of XAI by SpaceX.

The proposed solution: spend $60 billion to acquire Cursor, the AI coding tool SpaceX's own engineers were already using instead of Grok, with a $10 billion break fee if the deal collapses.

To summarise the AI thesis: the product representing 93% of claimed addressable market has single-digit market share, is avoided by the company's own technical staff, and the solution is to buy a competitor at a premium while carrying $29 billion in existing debt.

The near-term AI revenue picture rests substantially on a $15 billion annual contract to rent compute capacity to Anthropic — one of Grok's direct competitors. That contract, worth roughly 40% of projected near-term AI revenues, can be cancelled by Anthropic with 90 days' notice. SpaceX's AI business model, at its core, is renting GPU capacity to rivals on terminable terms. As one analyst framing put it, SpaceX is essentially a space-sector CoreWeave — an infrastructure landlord being valued as a software platform.

If Anthropic's quarterly revenue is already roughly double SpaceX's entire AI segment revenue, and Anthropic is reportedly seeking a $1 trillion IPO valuation, the internal logic of SpaceX's $1.75 trillion ask becomes difficult to follow.

---

Related-Party Transactions and Governance Concerns

The prospectus contains several related-party disclosures that warrant close reading.

The Cybertruck problem: SpaceX purchased $650 million in goods from Tesla last year, including $131 million in Cybertrucks at full retail price — no volume discount. That's roughly 1,500 vehicles. At one point in Q4 2025, SpaceX alone accounted for 18% of all Cybertruck registrations in the United States. Using capital from one company you control to support the sales figures of another company you control, immediately before both face significant financial events, raises questions that shareholders and regulators will likely revisit. There is also a $77 million discrepancy between what SpaceX says it paid Tesla and what Tesla's own filings report receiving.

The board-member lease problem: Antonio Gracias, a long-term Musk ally and sitting SpaceX board member, runs a private investment firm called Valor Equity Partners. Per the prospectus, SpaceX and XAI are carrying more than $20 billion in AI infrastructure lease obligations tied to entities connected to Valor. Several of these are structured as failed sale-leaseback transactions — meaning the assets remain on SpaceX's balance sheet and the obligations are recorded as debt. A board member whose firm holds $20 billion in financial claims over the company he oversees is a governance arrangement most institutional investors would flag immediately.

The XAI acquisition parallel: In 2016, Musk used Tesla shareholder money to acquire SolarCity, a company he also controlled, in a deal later scrutinised by a Delaware court as "deeply flawed." The court found the deal fair on the grounds that 85% of shareholders approved it and the stock subsequently rose — not exactly a ringing endorsement of the process. The SpaceX acquisition of XAI follows a structurally similar pattern, this time executed in Texas, where corporate law sets a 3% ownership threshold (approximately $52 billion at current valuation) to bring a derivative lawsuit, requires proof of fraud or intentional misconduct — not merely a breach of fiduciary duty — and explicitly excludes emails and text messages from shareholder document requests.

The lesson drawn from the SolarCity episode appears to be about jurisdiction, not process.

---

Starlink Is Real. Everything Else Is a Roadmap.

It's worth separating signal from noise here, because burying Starlink's genuine achievements under the AI repositioning does investors a disservice.

Starlink is a legitimate, scaled, profitable business. It generates real cash flow, serves real customers across more than 100 countries, and has demonstrated a defensible competitive moat through low-Earth orbit satellite density that is extraordinarily difficult and expensive to replicate. The $4.4 billion in operating profit it generated in 2025 is not a rounding error — it's the only reason the rest of the company exists.

The problem is valuation. At 94x revenue, investors are not buying Starlink. They're buying the promise of point-to-point rocket travel (still no commercial product, with a 2018 decade-deadline expiring shortly), in-orbit manufacturing, Mars colonisation, asteroid mining, and an AI platform currently losing billions per quarter to competitors with larger model libraries, more enterprise customers, and faster product iteration cycles.

Each of those future business lines may, in some form, eventually become real. The question investors must ask is at what price that optionality is worth owning — and whether $1.75 trillion is that price, or whether it prices in futures that may never materialise on a useful timeline.

---

What the IPO Structure Tells You About Who Benefits

The clearest way to understand who this IPO is designed to serve is to follow the capital stack backward.

Morgan Stanley and Goldman Sachs are co-leading the deal — Morgan Stanley reportedly so exercised about appearing second in the tombstone (purely for alphabetical reasons, the term sheet had to clarify) that SpaceX issued an explicit clarification. Twenty-three banks are involved in total. That is not an unusual number for a deal this size, but it is a reminder that the ecosystem surrounding this offering is vast and highly incentivised.

The more consequential beneficiaries are the venture capitalists and lenders who financed Musk's $44 billion acquisition of Twitter in 2022. That deal was widely regarded as chaotic, and several banks were left holding underwater debt they couldn't easily syndicate. The XAI-SpaceX merger converted a meaningful portion of that exposure into SpaceX equity — conveniently timed just before the largest IPO in US history would provide liquidity at the company's stated $1.75 trillion valuation.

Public market investors, by contrast, are buying in at the top of that capital structure, at a valuation that assumes the AI business works, the rocket economy materialises, and Grok somehow closes a 96.6% market share gap against OpenAI and Google. That's not impossible. But it is a very specific set of bets to be making at 94x revenue.

---

What Investors Should Actually Watch

If you're evaluating SpaceX as a potential investment rather than a cultural event, here are the metrics that matter most:

• Starlink ARPU trend: Average revenue per user falling 18% signals aggressive discounting. Whether that stabilises post-IPO is critical to cash flow projections.
• Anthropic contract renewal: A 90-day cancellation window on 40% of AI revenue is a material risk. Watch for whether that contract is renegotiated on longer terms.
• Grok adoption rate: Enterprise AI revenue requires enterprise adoption. Third-party usage data from sources like Bloomberg Second Measure or SimilarWeb will matter more than TAM claims in the prospectus.
• Capital expenditure trajectory: $20.7 billion in capex on $18.7 billion in revenue is unsustainable without either a step-change in profitability or continuous equity raises. The pace of Starship development and AI infrastructure build-out will determine how long the cash burn continues.
• Debt repayment schedule: The $20 billion bridge loan matures on terms that aren't fully disclosed. Understanding when and how it rolls over matters for liquidity risk.
• Texas litigation environment: In the absence of the governance protections Delaware shareholders enjoyed, institutional investors may need to price in reduced oversight and fewer legal remedies.

None of this means SpaceX fails. Starlink alone is a business worth serious investor attention. But $1.75 trillion is a number that requires every ambitious bet in the prospectus to land — and the filing is unusually candid, in its own way, about how many of them haven't yet left the launch pad.

---

Frequently Asked Questions

How much money has SpaceX lost since it was founded?

According to SpaceX's IPO prospectus, the company has accumulated over $37 billion in cumulative net losses since its founding — more than any company that has previously gone public in US history.

Is Starlink profitable?

Yes. Starlink generated $4.4 billion in operating profit in 2025, making it the only profitable division within SpaceX. The space and AI divisions both posted significant operating losses during the same period.

What is SpaceX's AI product and how large is its market share?

SpaceX's AI product is Grok, developed through its XAI subsidiary, which was acquired by SpaceX for $250 billion. Grok holds approximately 3.4% market share in the AI sector, according to Bloomberg. The enterprise AI market is dominated by OpenAI, Anthropic, and Google, a fact the prospectus itself concedes.

What are the main governance concerns raised by the SpaceX IPO filing?

The prospectus discloses several related-party transactions that raise governance questions: SpaceX purchased $131 million in Cybertrucks from Tesla at full retail price; a sitting board member's firm holds over $20 billion in lease obligations with SpaceX; and the $250 billion XAI acquisition was structured in Texas, a state with significantly less shareholder litigation protection than Delaware, where a similar SolarCity transaction faced intense court scrutiny.

Why does SpaceX's valuation appear so high relative to its revenue?

At $1.75 trillion on $18.7 billion in revenue, SpaceX is priced at roughly 94x sales. That multiple reflects investor expectations for future businesses — AI, in-orbit manufacturing, Mars transport, asteroid mining — rather than current financial performance. The Financial Times has noted this may be the largest and most speculative stock the market has ever been asked to price, given that 93% of the claimed $28.5 trillion total addressable market consists of markets that do not yet exist in commercial form.

---

This article is for informational purposes only and does not constitute financial advice. Always consult a qualified financial professional before making investment decisions.