The first IPO of an AI model developer is imminent. Anthropic, the company behind the increasingly successful Claude, has filed a confidential registration statement with the U.S. Securities and Exchange Commission (SEC). “Confidential” means that the document itself cannot be viewed prior to the IPO. OpenAI has not yet reached that stage, so it appears that Anthropic will beat the ChatGPT developer to the punch.
Anthropic’s valuation stood at $965 billion following its Series H funding round. It is not certain that this will also be the initial market value at the time of the IPO. It is also unknown how many shares will be offered for sale, though rumors suggest the figure could be around $75 billion. In other words, a public offering of roughly 5 to 10 percent of the AI company seems likely.
A confidential IPO filing allows a company to prepare its IPO plans without public exposure. Only once Anthropic definitively decides to go public will an S-1 registration statement follow, containing details on finances, legal matters, and the shareholder structure. The most recent funding round, from which Anthropic derived its current valuation, was co-led by Altimeter Capital, Dragoneer, Sequoia Capital, and Coatue, along with a range of institutional and strategic investors who were anticipating a potential IPO.
Busy IPO Season
The filing coincides with a remarkably busy IPO season. Not only that, but the market valuations involved are enormous. SpaceX is preparing for an IPO with a target valuation of $2 trillion. OpenAI is also taking steps toward a public listing, which it announced confidentially earlier this year. OpenAI raised $122 billion in a funding round back in March, at a valuation of $852 billion. Two of the world’s largest AI labs thus appear poised to face off against each other on the stock market.
Whoever goes public first may benefit significantly from it. The case of AI chipmaker Cerebras shows that a stunning first day on Wall Street offers far from any guarantees. For instance, that company’s value dropped by 30 percent in just half a month; this is a common occurrence with IPOs. The reason for this is as simple as it is difficult to anticipate in advance: a company with sufficient investment usually only goes public when its estimated value is at or near its peak. There may be other motivations, such as the current incentive for Anthropic and OpenAI to beat the other to the punch.
From underdog to top player
A notable detail is that Anthropic was founded in 2021 by former OpenAI employees and was long considered a smaller player in the AI landscape. That image has changed dramatically. Run-rate revenue recently surpassed $47 billion, compared to $9 billion at the end of 2025. Techzine previously reported that Anthropic is nearing profitability with an expected annual revenue of $10.9 billion.
At the same time, the company is still holding off on the broader rollout of the Mythos model, or an LLM comparable in performance. Anthropic presented the model in a very limited preview in April. The model has already discovered thousands of serious software vulnerabilities that must first be addressed. Anthropic is now granting the European cybersecurity agency ENISA access to Mythos.
Finally laying all cards on the table
Startups always hold a certain mystery for outsiders. Their revenue is only known if representatives choose to share it, while profitability is not the norm. In fact, companies that are already profitable as privately held firms with strong growth are sometimes assumed to be doing something wrong. Especially for AI companies, the current priority is above all to rapidly build out infrastructure and scale operations. This is partly to meet demand, but also to absorb the available hardware capacity. Amazon, Microsoft, Google, Oracle, Meta, OpenAI, and Anthropic account for the lion’s share of advanced computing power, partly through Nvidia, AMD, or Intel, but often also with their own chips. Investments to date have funded both research and this growth for Anthropic and OpenAI. An IPO could provide an additional cash injection. The downside is that the mysterious figures will become publicly shared results. That leaves the question of whether the business model will also work in the long term.