AI Giants Queue for IPOs: Will This Be the US Stock Market's "Last Hurrah"?

An IPO frenzy comparable to the peak of the internet bubble is taking shape. Three AI giants—OpenAI, Anthropic, and SpaceX—are racing toward public markets, each targeting valuations of $1 trillion. Their combined scale is sufficient to reshape the landscape of the US stock market. This unprecedented wave of listings serves as both the ultimate stress test for AI investment logic and the biggest key variable influencing the trajectory of risk assets this year.

On May 22, according to a Wallstreetcn article, OpenAI has prepared to secretly file for an IPO with regulators. It could list as early as September this year, targeting a valuation of over $1 trillion and aiming to raise approximately $60 billion. This would surpass the $25.6 billion IPO record set by Saudi Aramco in 2019 by more than double.

Meanwhile, competitor Anthropic is also advancing its listing plans, disclosing that Q2 revenue is expected to double quarter-on-quarter to $10.9 billion, potentially achieving quarterly operating profitability for the first time. Deutsche Bank pointed out in a research report that the execution of these two IPOs will "likely become a major swing factor for the direction of risk assets this year," making it a macro theme that must be closely monitored.

However, beneath the glossy valuations, the financial fundamentals of the two companies are starkly different. OpenAI generated $5.7 billion in revenue in Q1, but its adjusted operating margin was -122%, meaning it lost $1.22 for every $1 of revenue generated. Positive cash flow is not expected until 2029 or 2030 at the earliest. Anthropic, with $4.8 billion in revenue during the same period, expects Q2 revenue to jump to $10.9 billion and anticipates an operating profit of approximately $559 million, having already crossed the threshold to profitability.

Analysts point out that while the two companies are competing on the same stage, they present vastly different business logics, presenting public market investors with a rare dilemma.

The Largest IPO in History: How Shocking Are the Numbers?

Deutsche Bank noted in a research report that whether it is OpenAI or Anthropic, the size of a single IPO will exceed twice the amount raised by Saudi Aramco's 2019 IPO. Even after adjusting for inflation, it will easily become the largest IPO in history.

In another research report, Deutsche Bank stated that if OpenAI achieves its target valuation of over $1 trillion, it will become the 14th largest company by market capitalization globally, trailing only Berkshire Hathaway and surpassing Eli Lilly.

By comparison, Berkshire Hathaway reported revenue of over $370 billion and net profit of $67 billion last year; Eli Lilly had sales exceeding $65 billion and profits of $21 billion. OpenAI, however, is not yet profitable, with annualized revenue of approximately $30 billion and only a few thousand employees.

From a market capacity perspective, Deutsche Bank believes that the current total market capitalization of the US stock market is approximately $70 trillion, five times that of the peak of the internet bubble, indicating far stronger absorption capacity than in the late 1990s.

At that time, an average of nearly 500 companies went public annually, whereas the average for this decade is only about 120, with listed companies today generally being more mature.

Furthermore, a single IPO size of $60 billion is only slightly lower than the total US IPO fundraising for the entire years of 1999 and 2000 (both approximately $65 billion), equivalent to half of the record $119 billion raised in 2021.

The "Siphon Effect" of Giants and Major Shifts in Passive Capital

As these giants move toward public markets, their drain on US stock market liquidity has triggered high alert among Wall Street participants.

The clustered listings of SpaceX, OpenAI, and Anthropic, coupled with Nasdaq's newly introduced "fast-track index inclusion" mechanism, are brewing an unprecedented major shift in passive capital, namely the siphon effect of AI giants.

As mentioned in a Wallstreetcn article, JPMorgan estimates that if SpaceX reaches a target valuation of $2 trillion and ultimately has 50% of its shares in circulation, passive funds will be forced to sell approximately $95 billion worth of holdings in Wall Street's existing eight major tech stocks (NVIDIA, Apple, Microsoft, Amazon, Google, Broadcom, Meta, Tesla) to make room for new positions.

Todd Sohn, Chief ETF Strategist at Strategas, pointed out that since the initial float ratio of an IPO is typically only 5%, while ETFs track trillions of dollars in assets, this extreme supply-demand imbalance will cause the index inclusion process to be "slightly crazy," leaving passive investors with no choice but to buy in at high levels.

Valérie Noël, Head of Trading at Syz Group, stated that the market has already begun betting on downward pressure on existing large-cap stocks.

According to information disclosed on March 28 this year, OpenAI's public listing will be a substantive referendum on the entire AI investment logic. The information shows that OpenAI's revenue reached $13.1 billion in 2025, but the projected net loss for 2026 will reach $14 billion.

Meanwhile, OpenAI has committed to investing approximately $1.4 trillion in infrastructure construction by 2033. If S&P Global, FTSE Russell, and Nasdaq adopt fast-track inclusion rules, it could force passive funds to buy approximately $24 billion to $48 billion worth of shares immediately after listing.

Faced with such massive capital restructuring, ordinary investors' portfolios will be passively reshaped along with the changes in rules, regardless of whether they are active or passive investors.

Deutsche Bank pointed out in a research report that the execution of these IPOs will be a major swing factor for the direction of risk assets this year. PitchBook's analysis is even more blunt:

A "systemic quality inversion" has appeared in the private market—companies with the highest valuations score lowest on business quality metrics that are truly priced in the public market.

For ordinary investors holding index funds or ETFs, it is difficult to stay out of this game: regardless of whether they are active or passive, their portfolios will be passively reshaped along with changes in index rules.

For active investors, when the S-1 filings are made public and all financial secrets are laid bare in the sunlight, the market will face a clear choice: believe in a company that has already found a profitable model, or a giant asking the market for several more years and hundreds of billions of dollars to explore the possibility of profitability?

The answer will determine whether this carnival is the starting point of a new cycle or the last dance before the party ends.

A Tale of Two Extremes: Anthropic's Profitability vs. OpenAI's Massive Losses

Although valuations for both have soared, the financial situations of the two AI leaders present starkly different pictures. Anthropic has already begun to profit, breaking the traditional notion that huge expenditures by AI companies would drag down near-term profitability.

As written in a Wallstreetcn article, on Wednesday local time, The Wall Street Journal reported that Anthropic's Q2 revenue is expected to more than double to $10.9 billion, achieving an operating profit of approximately $559 million.

Anthropic's gross margin has jumped from 38% to over 70%. Its CEO, Dario Amodei, once joked that revenue growth has become "too difficult to handle."

The company's success is mainly attributed to explosive demand for its coding tools from enterprise clients. About 85% of its revenue comes from corporate and developer customers, a model with clear willingness to pay and lower service costs.

In contrast, OpenAI is still losing money.

As mentioned in a Wallstreetcn article, data shows that OpenAI's Q1 revenue was $5.7 billion, but its adjusted operating margin was -122%, meaning it lost $1.22 for every $1 earned.

Approximately 85% of OpenAI's revenue is related to ChatGPT consumer subscriptions. Despite having 55 million paying users, it is backed by over 900 million weekly active users. The vast pool of free users brings a huge black hole of inference costs.

OpenAI expects to achieve positive cash flow only in 2029 or 2030. Its CEO Sam Altman and Application Business CEO Fidji Simo are attempting to shift focus toward commercial clients that can generate direct revenue.

In terms of IPO narratives, the two companies are telling completely different stories. Anthropic holds verified quarterly profit data, and its story can be benchmarked against Salesforce or ServiceNow, following the logic of an enterprise software company.

OpenAI needs to convince the market that AI agents, image generation, and even advertising businesses will eventually convert massive consumer traffic into profits.

In Sam Altman's plan, ChatGPT's advertising business could generate approximately $102 billion in revenue by 2030, but this requires time. Time is precisely the scarcest resource for OpenAI as it trades losses for growth.

AI Giants Cluster for IPOs: Essentially Passing the "Hot Potato" to Retail Investors?

As written in a Wallstreetcn article, Joachim Klement, Managing Director at Panmure Liberum, views this wave of AI giant IPOs as essentially a "risk transfer," a cash-out action that transfers early-stage investment risks on a large scale to retail investors, pension funds, and other institutions.

He believes that companies like OpenAI and Anthropic are accelerating their listings amidst high investor sentiment, intending to cash out at high valuations before the hype fades. Early institutional investors can exit the public market safely, while retail investors and pension funds taking over the positions will face the risk of financial logic finally returning to reality.

He directly characterized this process as "an action to transfer investment risk from current holders on a large scale to those willing to pay for the story."

Klement cited Alan Greenspan's 1996 warning of "irrational exuberance" as a reference—three years before the bubble burst. He judges that AI hype may continue in 2026, and it is unlikely that hyperscale cloud providers will cut investments; however, the "impossible math" will eventually return to reality, "perhaps not in 2026, but possibly in 2027 or 2028."