SpaceX and OpenAI IPOs expose massive AI losses, reigniting bubble concerns

Two of the world's most ambitious technology companies are preparing to go public with valuations that dwarf their current earnings—and the numbers reveal a stark contradiction that is forcing investors and analysts to confront an uncomfortable question: Is artificial intelligence the next great bubble?

SpaceX, Elon Musk's aerospace and AI conglomerate, is seeking to raise $75 billion in what would be one of the largest initial public offerings in history, with a total company valuation around $2 trillion. Yet buried in the company's filing documents is a troubling reality: its AI division, which includes the xAI startup that Musk acquired and integrated in February 2025, lost $6.4 billion in 2025 alone. The division generated $3.2 billion in revenue against total costs of $9.6 billion—a gap that tells the story of an operation burning cash at an unsustainable rate. Research and development consumed $5.1 billion of that spending, nearly 59 percent of the entire group's R&D budget.

Over the past three years, SpaceX's AI business has accumulated $11.9 billion in operating losses while bringing in only $8.8 billion in cumulative revenue. The bleeding continued into the first quarter of 2026, when the AI segment generated $818 million in revenue against an operating loss of $2.5 billion. SpaceX frames this as strategic investment, arguing that xAI has become a leader in building and scaling the computational infrastructure that AI systems require. The company points to its massive data center clusters—COLOSSUS and COLOSSUS II—as proof that it is constructing the world's largest AI training infrastructure. In May, it closed a deal with Anthropic to deepen this position. Yet the company's own numbers show that growth and investment are consuming far more capital than the business is generating.

OpenAI's situation is, if anything, more dramatic. The company controlled by Sam Altman is preparing to file for an IPO expected to close in September, with plans to raise approximately $60 billion. If successful, this single offering would exceed the historic 2019 IPO of Saudi Aramco and would nearly match the total capital raised by all U.S. companies during the entire dot-com boom. OpenAI's valuation at debut is expected to hover around $1 trillion, though it already exceeded $850 billion in a March funding round backed by SoftBank, Amazon, and Nvidia.

At that valuation, OpenAI would rank as the fourteenth largest company in the world by market capitalization—ahead of pharmaceutical giant Eli Lilly, which generated $65 billion in annual revenue and $20.6 billion in profit last year, and behind Berkshire Hathaway, which reported $371 billion in revenue and nearly $67 billion in net profit. Yet OpenAI has not yet published its financial results. Internal documents reviewed by U.S. financial media outlets project losses of $14 billion for 2026 against revenue of approximately $18 billion. The company is expected to burn through $27 billion in cash this year alone. Analysts do not expect OpenAI to achieve positive free cash flow until 2029 at the earliest. The company's annualized revenue this month may reach $30 billion, but that figure masks the reality that it is losing money on nearly every dollar of sales.

These two IPOs—and potentially a third from Anthropic—are raising an old specter in technology markets: the bubble. The concern is not that AI is worthless or that these companies lack genuine technological achievement. Rather, it is that the astronomical valuations being assigned to them rest on the assumption that massive losses will eventually transform into massive profits, a conversion that may take years or may never occur. SpaceX's AI division is spending nearly three dollars to generate every dollar of revenue. OpenAI is spending roughly $1.50 for every dollar it brings in. Neither company has demonstrated a clear path to profitability at scale.

What makes this moment distinctive is the sheer magnitude of the capital being deployed. SpaceX and OpenAI are not small startups betting on a distant future. They are companies with real revenue, real customers, and real technology. Yet they are also companies whose current operations cannot justify their current valuations by any conventional measure. The market is being asked to believe that the losses will shrink, that revenue will accelerate, and that the infrastructure being built today will generate returns tomorrow. History suggests that sometimes this faith is rewarded. It also suggests that sometimes it is not.