Anthropic, the world’s most valuable AI startup, announced on Monday that,The draft S-1 prospectus has been secretly submitted to the U.S. Securities and Exchange Commission on the same day.. The proposed IPO will depend on market conditions and other factors, the company said.
(Source: Company official website)
During the listing process of U.S. stocks, secret delivery of prospectuses is a common operation. Companies planning to go public can first send their prospectus to the SEC for review, and then communicate with the regulator for modifications. After the review issues are resolved, the underwriters are ready, and the company determines that the market environment is suitable for listing, it will officially publish the prospectus, quickly start the road show, and run for listing. On the other hand, if the listing is ultimately unsuccessful, the company will not disclose a large amount of internal information to the outside world.
According to regulatory requirements,The prospectus needs to be disclosed at least 15 calendar days before the start of the road show.
Taking SpaceX as an example, the company reportedly secretly submitted its prospectus between March and April, and then made it public on May 20. Public information shows that SpaceX will launch a road show in the next few days and hit the Nasdaq exchange on June 12.
Two AI giants are racing to go public
It is worth noting that most of the quasi-listed companies that submit prospectuses secretly do so secretly, such asIt’s rare for Anthropic to secretly hand over a prospectus and then publicize it..
However, given the competitive situation of American AI giants, this operation has become reasonable again.
It was previously reported thatAnthropic’s archrival OpenAI is also preparing to secretly submit an IPO prospectus in the near future.. This means that the listing schedules of the two AI giants are likely to "collide." Considering that the amount of funds raised by both companies is in the tens of billions of dollars, going public earlier means that they can get a head start in raising funds in a market that has just been "siphoned" a lot of money by SpaceX.
(On February 19, 2026, local time, at the Artificial Intelligence Impact Summit hosted by Indian Prime Minister Narendra Modi (left), OpenAI CEO Sam Altman (middle) and Anthropic CEO Dario Amodei (right) refused to shake hands in public)
In the past few months, with the huge success of Claude Code, Anthropic has overtaken OpenAI on multiple levels.
The company announced last week that it had completed a $65 billion financing round, with its post-money valuation reaching $965 billion, surpassing OpenAI for the first time.. In a financing round completed in February this year, Anthropic was valued at $380 billion, roughly half of OpenAI’s.
Expected to achieve profitability for the first time in the second quarter
Compared with OpenAI, which has no hope of turning around in the short term, Anthropic's annualized revenue soared from US$14 billion to US$47 billion in the first five months of this year, and it is expected to achieve profitability for the first time in the second quarter.
During this period, Anthropic also developed a technology that shocked the global network security industry."Mythos" large model.
The company said last week that after a transition period to let outside organizations use the model to develop protective measures,"Myth" large model will be available to all customers in the coming weeks. The latest news on Monday also showed that the "Myth" large model is very likely to drive Anthropic's revenue to further surge, becauseThere will be a ton of companies that will "have to use" this latest and most expensive model.
Sam Rubin, senior vice president of the threat intelligence department of cyber security company Palo Alto Networks, told the media that in about three weeks of trialling the "Myth" large model, the model discovered more than 20 serious vulnerabilities and was about five times more efficient than traditional tools. But at the same time, this AI model also consumed the equivalent of $1 million in token fees at a very fast rate (note: no actual payment is required during the internal testing phase).
Therefore, even as a listed company with a market value of over US$200 billion, Palo Alto Networks has to find ways to reduce the cost of using AI. Rubin said the company has built a system in which the "Myth" large model formulates a plan to intrude into the software, and then directs "cheaper" models like Opus 4.7 to execute it, thus reducing costs.
Rubin also said,He has spoken to hundreds of enterprise security executives who are planning to increase their security budgets due to the looming threat of AI attacks..
He said: "The Myth Big Model has raised the importance of security issues to a new level. Now the board of directors is asking, the CFO is asking, the CEO is asking, they need to ensure that their security system is ready for this. And the CFO is now more willing to listen to the chief information security officer (CISO) on this type of preparation. So I think we will see further increases in related budgets."