SpaceX Files for the Largest IPO in History, Revealing Starlink's Dominance and a $6 Billion xAI Bet

When SpaceX filed its S-1 prospectus with the Securities and Exchange Commission on May 20, 2026, it did something no one had seen in a generation: it turned the most valuable private company in the world into an open book. The document running to hundreds of pages of financial tables, risk factors, and business descriptions is the first comprehensive look investors have had at the economics of the Musk empire — and the picture it reveals is more complicated than either the bulls or the bears had fully anticipated.

The headline numbers first: SpaceX plans to list under the ticker SPCX on the Nasdaq, targeting a valuation of $1.75 trillion and raising between $75 and $80 billion in fresh capital. The roadshow begins June 5, with pricing targeted for the June 18-30 window. Morgan Stanley serves as left-lead underwriter, with Goldman Sachs, JPMorgan Chase, Bank of America, and Citigroup co-leading. If the deal prices at the targeted valuation, it would be the largest IPO in the history of global capital markets — larger than Saudi Aramco’s $25 billion record set in 2019.

Three Companies in One Filing

The most important thing to understand about the SpaceX S-1 is that it is not just one company. In February 2026, SpaceX completed its acquisition of xAI — Elon Musk’s artificial intelligence startup, which controls both the Grok model family and the X social media platform. The S-1 consolidates all three business lines: Starlink satellite connectivity, the rocket/aerospace segment, and the xAI/X segment. Reading the financials without holding that structure clearly in mind leads to fundamental misinterpretations of the company’s profitability.

The Connectivity Segment (Starlink) is the business that makes SpaceX worth taking seriously as a public company. In 2025, Starlink generated $11.39 billion in revenue — 61% of SpaceX’s total — and posted an operating profit of $4.42 billion, an 86% increase in adjusted EBITDA from 2024. In Q1 2026, Starlink reached $3.26 billion in quarterly revenue (69% of total consolidated revenue), as subscribers more than doubled year-over-year from 5.0 million in Q1 2025 to 10.3 million subscribers across 164 countries as of March 31, 2026. The network now operates approximately 10,000 satellites in low Earth orbit.

The subscriber doubling is the most important data point in the filing. It tells investors that Starlink’s penetration is still in relatively early innings across its existing markets, that the product is retaining customers, and that the recurring subscription revenue model is working. Starlink is, by a substantial margin, the fastest-growing satellite internet business in history, and its competitive moat — the cost and complexity of replicating a 10,000-satellite constellation — is genuinely formidable.

The Space Segment (rocket launches, Dragon spacecraft, Starship development) generated $4.1 billion in revenue in 2025 but posted an operating loss of $657 million, largely because Starship development consumed approximately $3 billion in R&D spending. The aerospace business is operationally important as the launch infrastructure that enables Starlink expansion, but it is not yet a standalone profit center at scale.

The AI Segment (xAI/X) is where the S-1 gets genuinely provocative. Combined revenue from xAI’s Grok model business and X’s advertising platform reached $3.2 billion in 2025, up 22% year-over-year. The operating loss, however, was $6.4 billion — driven by $5 billion in R&D spending, a 331% increase that reflects the extraordinary compute demands of training and deploying frontier AI models. In Q1 2026, the AI segment burned another $2.5 billion.

The Anthropic Deal Changes the Calculus

Buried in the S-1 but critical to understanding the AI segment’s strategic logic is a single contract: SpaceX’s Colossus 1 data center in Memphis has secured a deal with Anthropic worth $1.25 billion per month through May 2029. Over its lifetime, that contract is worth approximately $40 billion.

The deal reframes what Colossus represents. Built at a scale that made it the world’s largest AI training cluster when it opened, Colossus was initially presented as infrastructure for xAI’s own model training. It is now also commercial AI compute infrastructure. The Anthropic contract — running Claude inference and potentially training workloads on SpaceX-owned hardware — turns the AI capex losses into partially recoverable infrastructure investment. It also creates an interesting competitive dynamic: Anthropic, the company that just closed a $30 billion round at a $900 billion valuation, is now also one of SpaceX’s largest B2B customers.

The Musk Governance Question

No S-1 discussion of SpaceX is complete without addressing governance. Elon Musk controls SpaceX through a combination of economic ownership and supervoting share classes. The filing reveals he holds approximately 40% economic ownership but control over the company’s direction is not structurally constrained by that figure. Public investors in SPCX will be buying a claim on the economic value of Starlink, aerospace, and xAI without meaningful governance influence over how those businesses are run.

The S-1 risk factors section is unusually candid about this. It dedicates substantial space to what it calls “key person risk” — explicitly naming Musk and acknowledging that his attention is simultaneously committed to Tesla, xAI/X, SpaceX, and his role in U.S. government policy. The filing also notes that a five-for-one stock split was completed this month, adjusting the per-share reference price that retail investors will likely anchor to.

Why Institutional Investors Are Paying Attention Anyway

The structural governance concerns are real, but they have not dampened institutional interest. The IPO is reported to be significantly oversubscribed at current price ranges, for reasons that boil down to two arguments.

First, there is no other way to own Starlink exposure at scale. A decade of private market appreciation means that most of the value creation in SpaceX has already happened off public markets. But Starlink at 10.3 million subscribers and growing is not a mature business — it is arguably still in the first third of its potential subscriber curve, with opportunities in maritime, enterprise, and government connectivity that residential accounts have barely scratched. For large institutional allocators who need to own the satellite internet theme in a public portfolio, SPCX is the only vehicle.

Second, the orbital AI compute thesis is genuinely novel. SpaceX’s S-1 mentions, briefly but significantly, that the company expects to begin deploying orbital AI compute satellites as early as 2028 — meaning satellites designed not just for communications but for running AI inference workloads in space. The Colossus deal with Anthropic is the terrestrial precursor to that strategy. Investors willing to take a five-to-ten year view on what SpaceX builds on top of its orbital infrastructure are making a bet that no amount of detailed short-term financial modeling can properly capture.

The Numbers That Give Pause

Consolidated SpaceX generated $18.7 billion in total revenue in 2025 but reported an operating loss of $2.6 billion. The profitability picture is entirely dependent on Starlink continuing to grow subscribers and maintaining its ARPU (average revenue per user), while the xAI loss either narrows dramatically or begins producing revenue at scale through commercial AI contracts like the Anthropic deal.

The Q1 2026 quarterly revenue of $4.7 billion against an operating loss of $1.9 billion suggests that trajectory is not yet inflecting positively. Bulls will note that the xAI R&D spend should taper as models reach deployment readiness; bears will note that AI R&D costs have a way of expanding to match ambition rather than contracting to match revenue.

At a $1.75 trillion target valuation, SpaceX is being priced as though Starlink’s subscriber count will quadruple, the xAI losses will eventually resolve into profitability, and orbital compute will become a real business. None of those outcomes is implausible. All of them require time and things to go right. The June pricing will reveal whether enough institutional capital agrees.