Musk bundles xAI into the largest IPO in history

The biggest IPO ever is an AI bet in a rocket’s clothing

Wall Street is about to price the largest IPO in history, and most buyers will not fully grasp what they are buying. On June 11, SpaceX prices its public debut under the ticker SPCX, with trading set to begin June 12, targeting a $1.75 trillion valuation and a raise of up to $75 billion, per The Tech Marketer’s IPO breakdown. That raise would shatter Saudi Aramco’s 2019 record of $35.4 billion — more than doubling the largest offering ever recorded. But the headline number conceals the real transaction: a rocket-and-satellite company welded to a money-losing artificial intelligence lab, sold as a single, indivisible bet on Elon Musk.

The welding happened in February. Per CNBC’s account of the deal, SpaceX absorbed xAI — the company behind the Grok chatbot and the X social network — in an all-stock swap that valued the combined entity at $1.25 trillion, with SpaceX carrying $1 trillion and xAI $250 billion. Musk framed it in a blog post as building “the most ambitious, vertically-integrated innovation engine on (and off) Earth,” citing “orbital data centers” as the strategic glue. It was, at the time, the largest merger ever announced. Four months later, that merger is the load-bearing assumption beneath the largest IPO ever attempted.

The stakes are not abstract, because the AI half is bleeding. Per The Tech Marketer, the xAI/Grok segment posted an operating loss of roughly $2.5 billion in Q1 2026 on just $818 million of revenue — a unit that loses three dollars for every dollar it earns. The merger turned a profitable aerospace company into a consolidated loss-maker: SpaceX swung to a 2025 net loss of $4.9 billion, per TechTimes’ filing analysis. Public investors are being asked to underwrite that loss at the richest valuation any IPO has ever carried.

This matters for the AI industry far beyond Musk’s empire. For two years the public-market test of frontier AI has been deferred — labs raised in private at valuations no auditor scrutinized. SPCX changes that. It forces the first true price discovery on a frontier lab inside a fully reporting, SEC-supervised entity, even if Grok arrives smuggled inside a rocket company. As we noted when OpenAI filed confidentially for its own offering, the floodgates to public AI markets are opening. SPCX is the first one through, and the first to be marked to market.

The structure of the bet should unsettle anyone who values clean exposure. Buyers cannot own the launch business without the AI lab, cannot own Starlink without Grok’s regulatory exposure, and cannot vote against any of it. The IPO fuses three companies at radically different maturities — a cash-printing satellite utility, a capital-incinerating rocket program, and a litigation-magnet AI startup — into one ticker controlled by one man. The question SPCX poses is whether the market will pay a trillion-dollar premium for vertical integration, or discover that bundling a loss leader into a champion dilutes both.

Follow the segments, find the loss leader

Disaggregate the company and the valuation logic falls apart in plain sight. SpaceX is not one business but three, and only one of them makes money. Per The Tech Marketer, Starlink generated $3.26 billion in Q1 2026 revenue and $1.19 billion in operating profit across 10.3 million subscribers — the lone profitable division and the engine of the entire investment case. The space program and xAI both lose money, with the AI unit alone shedding $2.5 billion in the quarter. The math is brutal: the profitable utility is being asked to carry two cash furnaces at a $1.75 trillion price.

The multiples expose how much faith that price demands. At $1.75 trillion against 2025 revenue of $18.67 billion, SPCX is asking buyers to pay between 109x and 116x trailing sales, per The Tech Marketer. Even on optimistic projected 2026 revenue of $27–30 billion, the forward multiple compresses only to 58–65x — still richer than almost any large-cap technology stock trades. The one number bulls will cling to is adjusted EBITDA of $6.58 billion for 2025, which strips out the heaviest non-cash charges and makes the cash engine look healthier than the headline net loss suggests. But adjusted EBITDA is precisely the metric that flatters companies carrying large lease and financing burdens, and at these multiples it cannot bridge the gap. The takeaway: nearly every dollar of this valuation is priced on a future that has not yet been earned, much of it pinned on Grok and orbital data centers that do not yet generate profit.

The capital structure tightens the squeeze. Per Whalesbook’s valuation analysis, auditor PwC reclassified roughly $20 billion of GPU lease arrangements as “failed sale-leasebacks” — converting them into secured loans — on top of about $9 billion in reclassified related-party debt, a burden built in part to fund xAI’s voracious compute appetite. A company carrying nearly $30 billion in such obligations while losing $4.9 billion a year is precisely the profile that public markets punish in a downturn. The IPO’s $75 billion raise is not growth capital so much as a pressure valve — fresh equity to refinance an AI bet that private markets were increasingly reluctant to keep funding alone.

Here is the proprietary stitch the press releases bury. Per The Tech Marketer, Anthropic is paying roughly $1.25 billion monthly — about $15 billion annually — for compute capacity tied to the SpaceX/xAI orbit. Set that against xAI’s own segment revenue of $818 million per quarter, or ~$3.3 billion annualized, and a startling ratio emerges: a single rival’s compute spend may exceed xAI’s entire product revenue by more than four to one, per the same figures. The AI unit’s most reliable cash flow could be selling infrastructure to competitors, not selling Grok to consumers. Investors buying “an AI champion” may actually be buying an AI landlord whose best tenant is the enemy — and whose own product line still operates deep in the red.

Then there is the control structure, which converts a financial bet into an act of faith. Per The Tech Marketer, Musk will command 85.1% of voting power through a dual-class share structure, and per TECHi’s S-1 watch, SpaceX is filing as a “controlled company” under Nasdaq rules — a designation that exempts it from key board-independence requirements. Public shareholders fund the enterprise but cannot direct it. They cannot block the next acquisition, replace the CEO, or veto a related-party deal. At a $1.75 trillion valuation, that is the most expensive minority stake ever sold, and the IPO’s success rests on buyers accepting that their capital buys upside, not authority.

The pricing mechanics explain the timing. Per TECHi, the deal prices around $135 per share across roughly 555.6 million shares to raise $75 billion — yet per BTCC’s free-float analysis, that represents only 3.75% of total equity. Selling a sliver of the company at a record price is how you crystallize a trillion-dollar valuation without ceding control or flooding the market with stock. A successful debut would also cement Musk’s status as the world’s wealthiest individual, anchoring his entire financial gravity from Tesla loans to xAI’s next raise. The IPO is as much a personal balance-sheet event as a corporate one, which is exactly why skeptics question whether the price was set by markets or by ambition.

The Nasdaq mechanics add a final, underappreciated tailwind. Per TECHi, under the exchange’s fast-entry rule a company of SPCX’s scale qualifies for the Nasdaq-100 roughly 15 trading days after listing, triggering mandatory buying from index-tracking ETFs regardless of valuation. Combine that with the deliberately thin 3.75% float, and the early price action may be driven by structural flows rather than fundamental judgment. The takeaway: the first weeks of SPCX trading could look like validation when they are merely plumbing — passive money forced to buy a stock that active money would scrutinize.

The ways this rocket could explode on the pad

The most credible warning comes from the people whose job is to price companies dispassionately. Per TradingKey’s coverage of Morningstar’s assessment, analyst Nicolas Owens pegs SpaceX’s fair value at $780 billion — roughly 56% below the IPO target — and states bluntly, “We believe the company is significantly overvalued, and investors will have the opportunity to buy in at a more attractive price following the IPO.” Morningstar values the launch and Starlink businesses at about $611 billion and the AI segment, on a probability-weighted basis, at only ~$170 billion. The gap between $780 billion and $1.75 trillion is not a rounding error; it is the entire bull case asking to be taken on faith.

The AI valuation is the wobbliest leg, and Morningstar quantifies the fragility. Per TradingKey, the firm modeled three outcomes for xAI: an optimistic case worth $1.3 trillion but assigned just 7% probability, a base case soaked in “high scientific and economic uncertainty,” and a 43%-probability “shelved” scenario in which the AI bet destroys $81 billion in value outright. Read plainly, the analysts judge it more likely that xAI is abandoned than that it pays off spectacularly. For an IPO whose premium over fair value rests largely on AI optionality, a base rate that tilts toward failure is the single most dangerous fact in the prospectus.

Grok’s legal exposure is no longer theoretical, and it strikes at the AI thesis directly. Per Engadget, UK Member of Parliament Jess Asato sued xAI over sexualized deepfake images the chatbot generated, a case filed at the High Court in England and detailed by Dataconomy. It is one of at least six major suits across the US and UK, layered atop investigations by UK regulator Ofcom and the EU’s Digital Services Act apparatus, plus pressure from dozens of US state attorneys general. xAI has reportedly set aside a $500 million litigation reserve. A product facing bans and probes across its largest Western markets cannot easily compound the revenue the valuation assumes.

The float and lockup dynamics threaten the price the moment momentum fades. Per TradingKey, the deliberately minimal float that props up early trading also sets a trap: as private-investor lockups expire over the following months, accumulated selling pressure could overwhelm thin demand and reset the stock far below its debut. The structural buying from index inclusion is a one-time event; the lockup unwind is a recurring tide. Early gains driven by scarcity are not the same as gains driven by conviction, and buyers who mistake the former for the latter may be the ones left holding shares when the float widens and the real price emerges.

Governance is the risk that compounds every other risk. Per TradingKey, Morningstar flagged Musk’s concentration of voting control alongside the non-arm’s-length xAI acquisition as “significant hidden risks,” and Danish pension fund AkademikerPension called the structure “disastrous.” When one person serves as CEO, CTO, and board chair while controlling 85% of the vote, every conflict — the price paid for xAI, the compute deals with related parties, the Mars-vesting share grant — resolves in his favor by default. The non-arm’s-length nature of the merger means public investors must simply trust that $250 billion was a fair price for xAI, with no independent check. That is governance by faith, sold at a record premium.

The human capital warning is quieter but telling. The post-merger reorganization saw key xAI co-founders depart, thinning the founding research bench just as the lab was folded into SpaceX. In frontier AI, talent is the moat; a research lab shedding founders before its parent goes public is a lab whose ability to keep pace with OpenAI, Anthropic, and Google is in doubt. Concerns over related-party debt and the entanglement of three businesses also clouded the debut, per Whalesbook’s valuation analysis. The bull case needs xAI to win a race it may be quietly losing on its most important input.

The strongest rebuttal, in fairness, is that betting against Musk has bankrupted skeptics before. Starlink alone — 10.3 million subscribers, $1.19 billion in quarterly operating profit, a reusable-rocket monopoly that no competitor approaches — is a genuinely extraordinary asset, and reusable launch could make orbital data centers a real category that justifies the vertical integration. If Grok climbs into the frontier tier and orbital compute becomes the cheapest way to train models, the $1.75 trillion price could look cheap in hindsight. The counter-counterpoint: that scenario requires the 7%-probability outcome to land, and pricing the best case as the base case is exactly how record IPOs become cautionary tales.

What to watch when SPCX rings the bell

Where this leads depends on a question the IPO cannot answer on day one: is SPCX a space company that happens to own an AI lab, or an AI bet that happens to own rockets? Per BTCC, the offering lists June 12 at a $1.75 trillion valuation on a float of just 3.75%, and its early performance will be shaped as much by index mechanics and that thin float as by fundamentals. The honest read is that the first weeks will tell you almost nothing about the durable value; the lockup expiries, the next Grok ruling, and the first post-IPO earnings report will tell you everything. This is a position to understand before it is a position to take.

The broader signal for AI is the one operators should not miss. SPCX drags frontier-lab economics into the daylight of public disclosure, and the numbers are sobering — a leading lab losing $2.5 billion a quarter, leaning on a rival’s $15-billion-a-year compute bill, and facing litigation across its core markets. Whatever the stock does, the prospectus is a rare audited window into what frontier AI actually costs to run. As Anthropic and OpenAI line up their own offerings, SPCX is the canary: how the market prices Grok’s losses inside a rocket company is a preview of how it will price standalone labs whose losses have nowhere to hide. The era of unaccountable private AI valuations is ending, and this is the first invoice.

For operators, investors, and builders weighing exposure, a disciplined checklist matters more than the headline:

• Separate the segments before you separate from your money. Value Starlink ($1.19B quarterly operating profit) on its own merits, then decide whether xAI’s $2.5B quarterly loss is a price you want bundled in. You cannot buy one without the other.
• Anchor to fair value, not the ask. Morningstar’s $780 billion estimate sits 56% below the IPO target; treat the gap as the market’s optimism tax and size positions for the possibility that the stock reprices downward after lockups expire.
• Watch the lockup calendar, not the launch-day pop. Index-inclusion buying and a thin float can inflate early prices; the real test is whether demand holds as private holders are freed to sell over the following months.
• Price the AI segment as optionality, not a sure thing. With a 43% modeled probability that the AI bet is shelved, treat any Grok upside as a call option you are paying a steep premium for — not as embedded value.
• Track Grok’s legal docket as a revenue input. Bans, DSA and Ofcom probes, and a $500M litigation reserve are not side issues; they cap the addressable market the valuation assumes.
• Read the governance as a permanent discount. With 85.1% voting control and a non-arm’s-length merger baked in, assume related-party decisions will favor the founder and demand a margin of safety for the loss of shareholder voice.
• Use SPCX as a comp, not just a trade. However it prices, the disclosures reset the public benchmark for what a frontier AI lab is worth — calibrate your read of the coming OpenAI and Anthropic listings against it.

The deepest lesson of SPCX is that the AI boom has reached the stage where its costs can no longer hide in private markets. Musk’s genius was to bundle a fragile, litigated, cash-hungry AI lab into the one asset whose mythology could carry it — a reusable-rocket monopoly and a satellite cash machine. Whether that bundling is financial alchemy or financial engineering will be settled not on June 12, but in the quarters after, when the float widens, the lockups lift, and the market finally has to price Grok on its own terms.

In other news

Anthropic confidentially files for a landmark IPO at a $965B valuation — Anthropic filed confidentially with the SEC on June 1, targeting an October listing on Nasdaq after closing a $65 billion Series H that vaulted it to a $965 billion valuation — eclipsing OpenAI’s $852 billion mark for the first time. The Claude maker’s revenue run-rate reportedly hit $47 billion, up from $10 billion a year earlier, and it expects its first profitable quarter in Q2 (Fortune).

OpenAI targets a September IPO at up to $850B — Days after Anthropic’s filing, OpenAI moved toward its own offering, with reports pointing to a listing as soon as September at a valuation of up to $850 billion. Two of the three largest AI labs are now racing to Wall Street within months of each other, a clustering that analysts warn could either open the floodgates or test investor appetite all at once (TechTimes).

The Pentagon races to replace Anthropic’s Claude — over safety guardrails — The Defense Department is now testing OpenAI, Google, and xAI’s Grok across classified systems after Anthropic refused to drop guardrails barring mass surveillance and fully autonomous weapons. Claude had been the primary AI provider inside the Maven Smart System; Anthropic is contesting the removal in court, arguing it could cost billions in revenue (TechTimes).

Orbital Industries raises $50M to discover exotic new materials with AI — The materials-discovery startup closed a $50 million Series B to scale its AI platform for designing novel compounds, joining a wave of capital flowing into AI-for-science applications. The round underscores how investors are increasingly funding AI bets tied to physical R&D rather than pure software (Fortune).