Multiple investors have warned that if the three major IPOs of SpaceX, OpenAI, and Anthropic were to hit the market in succession within the same year and perform poorly, it could substantially dampen enthusiasm for AI investments. SpaceX aims to raise 75 billion US dollars, equivalent to nearly 10% of the daily average trading volume of the US stock market, raising questions about the market's capacity to absorb such a large offering.
On April 27, at The Information's AI Financing Summit held in New York, multiple investors and bankers issued warnings: SpaceX, OpenAI, and Anthropic might go public in the same year consecutively, a phenomenon that is extremely rare in the history of capital markets.
However, if these IPOs perform poorly, the impact on overall AI investment enthusiasm would be substantial.
Alexa von Tobel, Founder and Managing Partner of Inspired Capital, frankly stated at the conference, "These three could potentially be the largest IPOs in history, all possibly occurring within the same calendar year. Some of these IPOs might actually act as a reality check… There’s a scenario where they instead make surrounding capital more cautious."
SpaceX: A $75 Billion Stress Test
Ashley MacNeill, Co-Head of Capital Markets at Vista Equity Partners, was more direct in her phrasing. She stated at another sub-forum that SpaceX's IPO 'has countless ways to go wrong' and 'only a few ways to succeed.'
The crux of the issue lies in scale. SpaceX aims to raise $75 billion, far surpassing any previous IPO in history. MacNeill pointed out that this issuance size is equivalent to nearly 10% of the average daily stock trading volume in the U.S. market. Whether the market can absorb such a large quantity of new shares at once is itself an open question—let alone the selling pressure from the gradual release of a large number of shares into the market after the lock-up period ends.
Jon Redmond, Portfolio Manager at Discovery Capital, expressed optimism about the SpaceX IPO, stating he was "excited" about it and predicting that after the prospectus is released, "people will truly begin to realize the scale of opportunity this company represents." However, he also acknowledged that historically, ultra-large IPOs have often been accompanied by market corrections—investors need to sell existing holdings to free up funds for subscribing to new shares. "I think a very similar situation will occur, most likely led by a decline in tech stocks. If I had to guess, it would be the 'Magnificent Seven' leading the decline."
$300 Billion to $400 Billion Annually: A New Scale of AI Financing
Another key theme of the summit was the unprecedented demand placed on capital markets by AI infrastructure construction.
Anish Shah, Head of Global Debt Capital Markets at Morgan Stanley, stated that the market needs to finance $300 billion to $400 billion annually for AI-related expenditures, which would account for approximately 10% of the total global bond and equity capital markets. He emphasized that this type of financing barely existed a year ago but has now grown explosively.
Traditional construction financing or project financing models struggle to meet the demands of AI infrastructure — the funding scale is too large, and the pace of construction is too rapid. Shah cited the example of Meta’s collaboration with Blue Owl to build the Hyperion data center campus in Louisiana, referring to it as a "groundbreaking transaction": through a co-investment structure, the project secured financing via $27 billion in investment-grade debt, while allowing Meta to keep this debt off its balance sheet.
Brookfield Asset Management has taken a different approach: conducting an in-depth analysis of customer contracts within data centers to determine who the actual payers are, assess counterparty risks, and evaluate whether projects can rely on the borrower's own cash flow to cover debt. Hadley Peer Marshall, Brookfield’s Chief Financial Officer and Co-Head of Infrastructure Credit, pointed out that delays in construction are a risk lenders must confront: "Typically, contracts stipulate that payments are triggered only once the project is operational. If the project fails to go live on time, there may even be penalties — and the debtor does not wish to bear this exposure."
Is AI a bubble? Investors remain sharply divided.
Regarding the AI boom itself, opinions among participants vary significantly.
$CoreWeave (CRWV.US)$ Board member and billionaire investor Glenn Hutchins explicitly stated that AI is not a bubble but rather "one of the most profound transformations in the economic and organizational structure of humanity in history." He warned: "If you are not participating, neither your capital nor your time is on the right side... You face a real risk of being left behind."
However, Hutchins also predicted that the decline in enterprise software valuations will persist. He used a striking metaphor: "A massive tsunami is about to sweep across the global economy, and software companies are merely the first tourists on the beaches to be hit."
Alex Baker, a partner at PwC US in the Technology, Media, and Telecommunications Transactions practice, presented another perspective: as more companies shift to usage-based pricing models, AI agents become the new "users," potentially executing operations at a hundred times the daily rate of humans. "If you have a true moat and a valuable platform, your worth in this new environment could be even greater than before."
Von Tobel indicated that she is currently more focused on companies with relatively slower growth but strong customer stickiness and high technological barriers. "If a company says, 'We were founded a month ago, and our ARR skyrocketed from zero to $100 million,' that doesn’t appeal to us. We look for businesses that are hard to execute early on — those with deep insights or significant intellectual property, such as the many infrastructure, quantum computing, and photonics companies we’ve invested in, which have 42 patent applications pending. Competitors won’t find it easy to replicate these ideas."
Editor/KOKO