SpaceX, Anthropic, and OpenAI are preparing the largest IPO wave ever attempted. Behind the historic valuations, the rewritten index rules, and the stripped governance rights lies a single question of who is really paying for the AI era?
Three of the most expensive private companies in the world are about to go public, all at once. SpaceX, Anthropic, and OpenAI are preparing to list on the stock market at a combined target value of roughly USD 3.6 trillion. To put that in perspective, that figure matches the entire yearly economic output of France, the world’s seventh-largest economy.
The biggest US-listed IPO in history was Alibaba in 2014, which raised USD 21.8 billion when it listed. SpaceX alone is targeting a raise of USD 75 billion, more than three times that record. This is not simply a big week on Wall Street. It represents a fundamental change in how risk is priced, who bears it, and who profits.
Valuations with No Comparison
For any company about to list on the stock market, investors need a way to decide whether the asking price is fair. The standard method is to look at how much revenue a company earns, and compare the listing price to that figure.
This gives you a price-to-sales multiple. A profitable, well-run software company typically trades at seven to eight times its annual sales. Even fast-growing tech startups at their peak rarely exceed 40 times sales.
SpaceX is asking to be valued at roughly 94 to 107 times its annual sales. OpenAI is asking for over 75 times. These numbers have no real precedent among large companies. The justification offered by the investment banks shepherding these listings is the speed at which revenues are growing.
OpenAI, for instance, went from USD 2 billion in annual recurring revenue at the end of 2023 to USD 25 billion by early 2026, a 12-fold increase in just over two years, a pace several times faster than Google or Amazon ever grew.
But here is the problem. Unlike a traditional software company, where serving one more customer costs almost nothing, these AI companies must pay a real, significant cost every single time someone uses their product. Every prompt typed into ChatGPT requires processing on expensive computer chips. The more users ChatGPT gets, the larger that bill becomes. This is structurally much closer to manufacturing than software.
Three Very Different Companies, Three Very Different Problems
Anthropic, the company behind the Claude AI system, is the most financially coherent of the three. It has built its business around large corporations rather than individual users. Its revenue has been growing rapidly, from roughly USD 10 billion annually at the end of 2025 to a reported USD 47 billion run rate by May 2026.
More than a thousand large companies now spend over a million dollars a year on Claude. Critically, Anthropic has been improving the ratio of computing costs to revenue. For every dollar of revenue earned, the compute cost has fallen from 71 cents to 56 cents in recent months. The company is projecting its first profitable quarter for June 2026.
None of this comes cheaply. Anthropic has raised tens of billions in private funding over the past two years, culminating in a USD 65 billion round in May 2026 that valued the company at USD 965 billion. At 21 times its annualised revenue, Anthropic’s asking price is aggressive, but not obviously unreasonable if the profitability trend continues.
OpenAI is in a far more precarious position. The company behind ChatGPT has 900 million weekly users, and revenue growing quickly. But it is burning cash on a scale that is genuinely alarming. For 2026, OpenAI is projected to lose somewhere between USD 14 billion and USD 26 billion depending on how the accounting is done, the difference being whether you count stock-based compensation and other standard costs.
The company’s computing bills alone are expected to hit USD 14 billion this year, and its gross margin is only 33 cents on the dollar. Under conventional accounting, the company will not break even until 2029 or 2030, and the total cash it will burn between now and then is projected at USD 665 billion.
In March 2026, OpenAI raised USD 122 billion in a private round. Just six weeks later, the company’s own chief financial officer privately acknowledged that this runway might not be enough to reach the IPO. The company is going public not as a triumphant success story, but because it has to. If it does not raise capital in public markets, it may run out of money by mid-2027.
SpaceX is the most complex of the three. Its core rocket and satellite internet business, the Starlink division with 10 million subscribers, is genuinely profitable, generating billions in cash each year. But SpaceX merged with Elon Musk’s AI company xAI in February 2026, absorbing a business that was haemorrhaging money, and had taken on USD 16 billion in high-cost debt to fund its AI computer infrastructure.
SpaceX refinanced that debt at a cheaper rate using a USD 20 billion loan, but the effect was to move xAI’s debt onto SpaceX’s balance sheet, bringing its total debt to roughly USD 29 billion. In just the first three months of 2026, SpaceX spent over USD 10 billion in capital expenditure, with more than three-quarters of that going toward AI infrastructure and chip procurement. Its cash reserves dropped by nearly USD 9 billion in a single quarter. The company posted a net loss of USD 4.28 billion in Q1 2026 alone.
Nicolas Owens, a lead equity analyst at Morningstar who published a valuation note on SpaceX ahead of its listing, put it plainly: “We think the company has been significantly overvalued, and investors will have opportunities to buy the stock at more attractive levels after the IPO.”
Morningstar’s discounted cash flow model values SpaceX at USD 780 billion, roughly 55% below its USD 1.75 trillion target.
Contracts That May End Up Looking Different
To justify SpaceX’s valuation to investors, the company’s IPO filing presents a series of large, long-term revenue contracts. In one deal, Anthropic agreed to pay SpaceX USD 1.25 billion per month to use its AI computing facilities. In another, Google agreed to pay USD 920 million per month from late 2026 through mid-2029. Together, these deals would represent tens of billions in committed future revenue.
But both deals contain important catches. Elon Musk himself publicly clarified that the Anthropic arrangement is structured as a 180-day lease, followed by a 90-day cancellation window. Either party can walk away with 90 days’ notice. This means the deal that SpaceX’s filing presents as worth USD 44 billion could in practice be terminated after delivering just USD 7.5 billion.
The Google contract, which the search engine giant described publicly as a short-term bridging arrangement, is listed in SpaceX’s filing as stable recurring revenue. Google also holds a 5% stake in SpaceX worth over USD 100 billion after the IPO, giving it a direct financial interest in making SpaceX’s revenue look as large as possible before the listing.
The gap between how these contracts are described in the filing and what they actually guarantee is a significant problem for anyone trying to assess what SpaceX is genuinely worth.
The Index Trap
Here is where things become uncomfortable for ordinary savers and retirees. Over USD 30 trillion in retirement savings and pension funds around the world is invested in passive index funds – funds that simply buy and hold every company in a major stock market index like the S&P 500 or the Nasdaq-100.
These funds do not make judgements about whether a stock is overvalued. They are legally and contractually required to own whatever is in the index.
Index providers have historically made companies wait before admitting them. The S&P 500 required one year of trading and four consecutive quarters of profit. Nasdaq required three months. But in the months preceding these IPOs, the major index providers quietly changed their rules. Nasdaq cut its waiting period to 15 trading days. FTSE Russell and CRSP cut theirs to five trading days. MSCI reduced it to 10. The S&P 500 is now debating dropping its profitability requirement entirely.
These changes mean that within days of listing, SpaceX, OpenAI, and Anthropic could be added to the world’s major indexes. Once they are, every passive retirement fund that tracks those indexes will be forced to buy their shares, regardless of price, regardless of losses, regardless of governance structure. Bloomberg Intelligence estimates that passive S&P 500 funds alone may be required to absorb nearly 20% of SpaceX’s public float within six months.
Selling Into the Wave
SpaceX shares are expected to start trading on the stock market on June 12.
But the company is floating only 5% of its total shares, around USD 75 billion worth. That means a very small number of shares will be available to buy. If passive index funds are forced to purchase hundreds of billions of dollars’ worth of a stock with limited supply, the price will go up mechanically, not because the company is worth more, but simply because the demand is forced.
SpaceX’s pre-IPO investors, venture capital firms, early backers, and insiders who bought in when the company was worth a fraction of its current price, have arranged a sophisticated schedule for selling their shares into exactly this wave of forced buying.
Rather than the traditional six-month lockup, which prevents insiders from selling immediately after an IPO, SpaceX has set up a rolling system that releases insider shares in tranches timed to coincide with the peaks of passive fund buying. The net effect is a direct compelled retirement savings buy in at the top, as private investors cash out.
Chad Anderson, founder of the venture capital firm Space Capital and an early SpaceX investor, gave a candid indication of intent when asked on CNBC whether his firm would sell shares at the first lockup expiration. “We’ve been invested for almost 10 years,” he said, before adding, “it’s our business to return capital to investors.”
Who Controls What
Even after buying shares, public investors will have almost no say in how these companies are run. SpaceX plans to issue two classes of shares. Ordinary public investors receive one vote per share. Elon Musk and a small circle of insiders will hold shares carrying 10 votes each. The result is that Musk, despite owning roughly 42% of the company’s equity, will control between 79% and 85% of the votes. He can only be removed as CEO by himself.
Under Nasdaq rules, this concentration of control qualifies SpaceX as a ‘controlled company’, which legally exempts it from requirements to have an independent board, independent compensation committee, or independent director nominations.
Professors Lucian Bebchuk of Harvard Law School and Kobi Kastiel of Tel Aviv University, writing on the Harvard Law School Forum on Corporate Governance in May 2026, acknowledged the commercial appeal while flagging the structural risk to ordinary shareholders: “SpaceX could well have assets with very high value and exceptional growth prospects, and investor enthusiasm about it might thus be fully understandable. However, SpaceX also has poor governance arrangements, which would have considerable adverse effects on public investors.”
Additionally, SpaceX’s charter requires disputes to go to private arbitration rather than the courts, making class-action shareholder lawsuits impossible. Under Texas corporate law, where SpaceX is now incorporated, any shareholder wishing to bring a lawsuit on behalf of the company must first own at least 3% of its voting shares. At a USD 1.75 trillion valuation, 3% of SpaceX is worth over USD 50 billion. No pension fund in the world holds that much. In practice, only Musk himself could sue Musk.
The Transfer of Risk
What is being marketed as the democratisation of access to transformative technology is, in financial reality, the offloading of venture capital risk onto public pension funds. The private investors who funded these companies over the past decade face a capital treadmill they can no longer sustain on their own. AI infrastructure is too expensive, the losses too large. The public markets, equipped with trillions in forced buying power, are the exit ramp.
For patient investors willing to wait out the post-IPO frenzy, Morningstar’s Owens offered a measured note of possibility in his published analysis. “We think long-term investors eager to participate in SpaceX’s future endeavours and potential success will have opportunities to do so with more margin of safety than the initial offering is likely to provide.” The best entry point may come precisely when insiders are selling, and the forced buying has run its course.
The rules have been changed to speed up their entry. The governance protections have been stripped to prevent accountability. The contracts have been packaged to look more durable than they are. What ordinary savers are being offered is the bill, dressed up as an opportunity.