new York
Anthropic’s debut on the public markets will be unlike any before, in a way that amounts to either a once-in-a-generation opportunity or a massive financial gamble, depending on who you ask.
When the startup behind the Claude AI assistant begins trading next week, there will be no analog to measure its performance. The market has simply never seen a dedicated AI model developer go public. It is rare for a company to IPO as a public benefit corporation and, with the exception of SpaceX, it is rare for a company to IPO with a potential target valuation greater than $1 trillion.
Anthropic’s IPO, which is expected to begin marketing to investors this week, is unconventional in every way. The company promised in its pre-IPO regulatory filing that its technology would “bring about an era of miracles” while warning that it could doom humanity. Yet the promise of a unique technological revolution (even with 90 pages of warnings and caveats in its 300-plus page filing) could make Anthropic the largest IPO in history.
But there are other superlatives and financial quirks — not to mention the continuing coolness toward AI on Wall Street — that will keep many investors away.
Despite rapid revenue growth, the five-year-old company has made only a quarter of its operating profit and will have to pay nearly $900 billion over the next few years – money it has pledged to spend on IT power and data center infrastructure.
Anthropic fans view its sales growth as a positive indicator of its financial health: The company had $17 billion in revenue in the most recent quarter, up 1,127% from the same period a year earlier. And in the first nine months of this year, revenue reached $33.2 billion, compared to just $4.6 billion for all of 2025.
But according to Anthropic itself, this type of growth is not sustainable in the long term.
“We do not expect our revenues to continue to grow at comparable rates, and we cannot assure you that they will continue to grow,” the company said in its pre-IPO prospectus.
At the same time, Anthropic relied this year on only five clients for a quarter of its turnover. The company did not disclose these five companies, but noted that some of them are also competitors.
This speaks to a precarious situation in which a handful of companies have outsized control over Anthropic’s future cash flows. Each of these companies, which may build their own AI models to compete with Claude, can decide to cut their own spending or withdraw completely at any time. Ideally, investors would like to see a stronger customer base.
Even optimistic investors say so-called concentration risk is worth watching.
Paul Marino, chief revenue officer at Themes ETFs, told CNN that the relatively small number of major clients “concerns me a little bit” because their needs could change on a dime. “It’s something to keep an eye on, but it doesn’t worry me too much because I’m one of those people who believes AI is the future.”
Normally, when a company goes public, insiders can’t just dump their shares on day one and cash them out: They have to hold on to them, usually for at least three to six months, to keep the stock price from falling.
Anthropic’s lock-up schedule is perhaps the most complicated of any IPO in recent memory. The company has opted for a complex, staggered schedule that allows certain “non-founding” insiders to sell 10% of their stakes on day one. The prospectus then lists some 23 dates during the first year when non-founding employees can offload more shares.
Critics see the unorthodox timeline as another red flag, given the roughly $130 billion Wall Street insiders and venture capitalists who have already invested in the company.
“This has maxed out private investors, who now need the company to tap public markets to get their paychecks,” financial research firm New Constructs wrote in a blog post.
David Trainer, founder and CEO of New Constructs, added on a call with CNN that he considers the insider payments to be “a horrible misallocation of capital.”
“We’re going to make a lot of people richer…that’s value that’s being taken away from public investors who are actually giving that to them out of their own pockets, out of their retirement accounts.”
Concentration risk is not limited to Anthropic’s professional clients. Critics also point to the concentration of power in the company’s peculiar ownership structure, which gives its seven founders the majority of shareholder voting rights. This new approach has become more common among tech companies in recent years, but it still raises concerns among some government officials.
“We’re seeing a very worrying trend of shareholder disempowerment and a shift in the entire balance of power in favor of management,” New York City Comptroller Mark Levine told Bloomberg News.
Anthropic’s investor tour this week could hardly come at a worse time for AI sentiment on Wall Street and beyond. Last week, AI-related stocks fell amid investor concerns that Democratic victories in Congress and state governments would slow data center construction. Americans’ feelings about AI safety have fallen into the gutter. And on Monday, Nvidia shares fell 5% after CEO Jensen Huang admitted to having an affair as his company announced it would investigate him on foreign intelligence grounds.
Anthropic, which declined to comment to CNN on its IPO plans, appears to be moving forward despite the turmoil. The Wall Street Journal reported Monday that the company had scheduled meetings with key investors this week and was considering listing its shares on Thursday or Friday of next week, citing people familiar with the matter.
Gn headline
