Houston, we have a valuation
SpaceX has pulled off the largest stockmarket flotation in history. Living up to the price will be harder than landing a rocket.
For a company whose founder spent the better part of a decade insisting it would never go public, the manner of its arrival was fitting. On June 12th Space Exploration Technologies began trading on the Nasdaq under the ticker SPCX. It had priced the night before at $135 a share, a $1.75trn valuation, raising $75bn—the largest initial public offering ever recorded, roughly triple the next-biggest. The stock then closed its first day at $161, up 19%, vaulting the implied value past $2trn and, with it, Elon Musk's personal fortune into territory no individual has occupied. Senator Elizabeth Warren greeted the debut by renewing her call for a wealth tax. The symbolism was complete: the most consequential private company of the century had become a public referendum on one man.
The sum involved is worth dwelling on, because superlatives have lost their force in this market. SpaceX sold 556.6m shares to raise $75bn—a figure that comfortably eclipses Saudi Aramco's 2019 record. It did so, moreover, while losing money. The temptation is to treat the flotation as a verdict on space exploration. That is the wrong lens. The firm that listed is no longer a rocket company in any meaningful financial sense. It is a satellite-broadband monopoly with two expensive science projects attached.
What you are actually buying
Strip away the iconography and SpaceX reports in three parts. The first, Space, is the original business: Falcon 9, Falcon Heavy and the Dragon crew capsule, which has flown astronauts since 2020. It generated about $4.1bn in 2025 but grew just 8%, drawn mostly from the Pentagon and NASA; of 165 Falcon 9 launches that year, only 43 carried outside customers—nearly three-quarters were flying SpaceX's own satellites. It is, in other words, increasingly an internal logistics arm, and it loses money: the segment spent close to $3bn on Starship development, against its $4bn of revenue.
The second, Connectivity, is the engine. Starlink booked $11.4bn in 2025, 61% of group revenue, with $4.4bn of operating profit—the only consistently profitable part of the company. It threw off roughly $7bn of segment EBITDA at 63% margins; that cash funds everything else in the prospectus that incinerates capital. Subscribers more than doubled in 2025, from 4.4m to 8.9m, and reached 10.3m by the end of March 2026 across some 160-odd countries. About three-quarters of all active manoeuvrable satellites in low-Earth orbit now belong to SpaceX; there is no real second-placed competitor.
The third part is the newest and the reason the valuation has detached from the launch pad. In February 2026 SpaceX acquired Musk's artificial-intelligence venture xAI in an all-stock deal valuing xAI at about $250bn and the combined group at roughly $1.25trn. xAI—which already owned the social platform X, and makes the Grok chatbot—became a wholly owned subsidiary; in May Musk announced plans to fold it into a sub-brand called "SpaceXAI". The result is a conglomerate yoking rockets, satellites, a frontier AI lab and a social network under a single, founder-controlled roof. The AI unit booked $3.2bn of revenue in 2025—and an operating loss of $6.35bn.
Mind the gap
Here the bull and bear cases collide. On a consolidated basis SpaceX reported about $18bn of revenue in 2025, a net loss of $4.9bn, and EBITDA of $6.58bn—the kind of split that should make any investor read the footnotes. The chasm between positive EBITDA and a GAAP net loss is driven by stock-based compensation, depreciation on the Starlink constellation, and AI infrastructure spending; the accumulated deficit now stands at $41.3bn. The bleeding accelerated in early 2026: a $4.28bn net loss in the first quarter alone, with the AI segment burning $2.5bn in three months. In plain terms, Starlink's profits are subsidising xAI's expenditures, and the AI losses are what tip the whole group into the red.
Even the jewel shows strain. Average revenue per user has slid from about $99 a month in 2023 to the mid-$60s by March 2026, as SpaceX traded price for global volume. In May 2026 it reversed course, lifting some Starlink plans by up to $10 a month—a signal that the land-grab phase is ending and the squeeze-the-installed-base phase beginning. At the offer price the company traded at more than 90 times trailing revenue; after the first-day pop, around 110 times—a multiple that assumes not just growth but perfection.
Why the rocket went up
So why did investors pay it? Four forces, none of them entirely about spaceflight.
The first is scarcity. SpaceX was the most coveted private asset on earth, accessible only through tender offers and a handful of crossover funds. Two decades of pent-up demand met a single liquidity event. The second is the deliberate courting of small investors: SpaceX reserved up to 30% of the offering for retail buyers, roughly three times the usual allocation, with Musk framing retail participation as a feature of the deal. A founder with a devoted following engineered a flotation his fans could pile into—and they did.
The third, and the proximate cause of the trillion-dollar premium, is the AI story. A week before listing, Musk laid out plans for a constellation of orbital data-centre satellites, arguing SpaceX already possesses most of the technology and that its experience operating constellations at scale is a moat no rival can match. The pitch reframes Starlink not as a broadband utility but as the distribution backbone of off-planet compute. Whether that is visionary or a valuation device is the central question for shareholders; the market, for now, has chosen to believe.
The fourth is the monopoly itself—genuine, and the soundest plank in the bull case. SpaceX's dominance of low-Earth orbit is real, its launch cadence unmatched, and Starlink's economics, in isolation, excellent. The error the price embeds is extrapolating that one proven advantage across two unproven ones.
Whose company is it anyway?
If the financials are precarious, the governance is the part the cheerleading has airbrushed out, and it deserves blunt treatment. In its prospectus SpaceX disclosed that Musk controls 85.1% of the voting power and warned investors plainly that they will not enjoy the protections afforded shareholders of fully governance-compliant companies. The mechanism is a dual-class structure—Class B shares carry ten votes, Class A just one—giving Musk roughly 85% of the vote on an estimated 42-43% economic stake, held through his revocable trust.
This is not Tesla redux; it is worse. Where Tesla has a single share class and Musk's 2018 plan merely sought to lift his vote toward 25% over a decade, SpaceX locks in 85.1% from day one, using Texas incorporation, mandatory arbitration and high thresholds for shareholder proposals to insulate management from the activism Tesla has faced. An explicit charter clause states that Musk is not legally bound to prioritise SpaceX—remarkable in a man running several capital-hungry ventures simultaneously. Governance scholars have not minced words. A Harvard analysis warns that a small-minority controller can extract private benefits at the expense of all shareholders—through related-party deals with Musk-affiliated entities, the allocation of opportunities between them, and the design of his own pay. One Morningstar analyst put it crisply: the structure "effectively entrenches Musk as a CEO who can't be deposed". Buyers of SPCX own cash-flow exposure to Musk's ambitions and almost no say over them.
The next frontiers
Where, then, is it heading? On three trajectories at once, each of which the valuation treats as a near-certainty.
The first is making Starship work. Flight 12, on 22 May 2026, was the debut of the upgraded Version 3 vehicle and was broadly successful—it reached space, survived re-entry and splashed down in the Indian Ocean, albeit with engine anomalies and a break-up at the surface. Yet SpaceX has still never placed Starship in orbit or flown a real payload; recent missions have carried only dummy satellites. The rocket is the keystone for everything else: cheap Starlink deployment, the Moon and Mars. NASA, having awarded SpaceX a $2.89bn contract in 2021 to build the Artemis lunar lander, quietly rewrote Artemis III in February 2026—the mission now stays in low-Earth orbit as a docking rehearsal, with the actual landing slipped to Artemis IV in 2028. Orbital propellant transfer—the gating technical demonstration—was internally targeted for June 2026. Until Starship flies to orbit and refuels reliably, the Moon and Mars timelines are aspiration, not schedule.
The second is the satellite roadmap. The next-generation Starlink V3 satellites, slated to begin launching in 2026, are designed to deploy 60 at a time on Starship, each launch adding some 60 terabits per second of capacity—more than 20 times current launches—with one-terabit laser links. This both deepens the broadband moat and, SpaceX argues, lays the groundwork for direct-to-mobile service and edge AI delivered from orbit.
The third—and the speculative heart of the $2trn case—is space-based compute. On 8 June Musk unveiled "AI1", SpaceX's first orbital data-centre satellite: 150kW of peak compute, a 70-metre wingspan wider than a Boeing 747-8, and a 110-square-metre radiator to shed heat into the vacuum. Two prototypes are planned for early 2027, to be built at a vast new "Gigasat" factory in Bastrop, Texas, as part of a filing for a constellation of up to one million satellites. SpaceX's own filing claims that "within a few years, the lowest cost to generate AI compute will be in space", on the logic that constant solar power and passive cooling beat any terrestrial grid. The model is to lease that capacity to AI and cloud tenants; reports name large anchor customers, though the specific figures circulating are unverified and, against $3.2bn of disclosed AI revenue, implausibly large. The competition is not standing still: Amazon's Project Kuiper, having launched its first satellites in 2025, plans commercial service from mid-2026 and intends to fuse it with AWS, while Microsoft's Azure Space links to Starlink today.
What of Mars, the founding purpose? It survives chiefly as motivation—and as compensation. Musk's pay package is, strikingly, tied to milestones including a colony on Mars. It is the tell of the whole enterprise: a public company whose chief executive's incentives point at another planet, run by a man legally unbound to prioritise it, valued as though every frontier will be conquered on schedule.
This newspaper has long held that genuine technological monopolies deserve a premium and that betting against Mr Musk's engineers has been a losing trade. Both remain true. But a $2trn valuation is not a bet on Starlink, which would justify perhaps a fraction of it; it is a bet that an unproven rocket, an unbuilt orbital cloud and a cash-burning AI lab will all come good, under a shareholder who answers to no one. The rocket, at least, has a flight plan. The valuation is flying on faith.