Anthropic Files Confidentially for IPO, Moving Ahead of OpenAI in Race for Capital

Anthropic has confidentially filed draft paperwork for an initial public offering, putting it ahead of OpenAI and positioning the company for a potential listing as soon as this fall. Once seen as a smaller startup that entered the market later than OpenAI, Anthropic has rapidly narrowed the gap as its coding-focused AI products and cybersecurity model Mithos win strong reviews.

Bloomberg and CNBC reported on June 2 that Anthropic confidentially submitted a draft registration statement for an IPO on June 1. Anthropic said in a statement that the filing gives it the option to proceed with an offering after the U.S. Securities and Exchange Commission completes its review. In a blog post, the company said the number of shares to be sold and the price range have not yet been determined.

Anthropic raised $6.5 billion in its latest funding round last week at a $96.5 billion valuation. That surpassed OpenAI, which was valued at $85 billion in a funding round at the end of March, though the two raises were about two months apart.

Its AI agent Claude Code has become highly popular with developers. Its cybersecurity-focused Claude Mithos Preview model also drew strong reactions on Wall Street and within the U.S. government, while attracting broad interest from financial institutions.

The company said in May that its annualized revenue had surged to $4.7 billion from $1 billion a year earlier.

Bloomberg News reported that Anthropic's second-quarter revenue is expected to reach $1.09 billion and that the company is on track to post its first profitable quarter this year. Anthropic told investors that annual revenue would exceed $5 billion as of the end of June. Its annual revenue stood at $400 million as of July last year.

Anthropic derives an overwhelming share of its revenue from developer API plans and enterprise subscriptions rather than consumer products. That business mix has helped fuel expectations that it can reach profitability earlier than other AI companies.

By contrast, The Information reported last week that OpenAI's annualized revenue had topped about $1 billion as of the end of 2025. First-quarter revenue rose sharply from a year earlier to $570 million, but still lagged Anthropic.

The bigger problem is profitability. OpenAI posted a non-GAAP adjusted operating margin of negative 122% in the first quarter, meaning it lost $1.22 for every $1 of revenue it generated.

Monthly active users of ChatGPT, the company's chatbot, have surpassed 350 million this year. But paying users account for only 5% to 8% of that total even including enterprise customers, meaning losses grow as usage rises.

OpenAI has budgeted $14.1 billion this year for the computing infrastructure needed for AI inference alone. Its losses are being driven by spending on Nvidia chips for advanced model development, cloud server rentals from Microsoft and others, and competition for top talent. Even if the company meets its target of $2 billion in monthly revenue this year, losses would still exceed revenue under its current structure.

That has raised fresh doubts about growth as reports suggest the company has internally missed revenue and user targets.

OpenAI Chief Executive Officer Sam Altman rejected the idea that the company was racing Anthropic to an IPO in a June 1 interview with CNBC. He said OpenAI would go public when it judged the timing appropriate.

By moving ahead with listing preparations before OpenAI, Anthropic has built a platform to attract investor attention and fresh capital.

Anthropic was founded in 2021 by seven researchers, including siblings Dario Amodei and Daniela Amodei, after they left OpenAI. Their departure was widely linked to internal concerns that Sam Altman's push toward commercialization was weakening safeguards meant to ensure AI development would not harm humanity.

Investors and management have taken comfort from the popularity of Anthropic Mithos. Earlier this year, defense contractors canceled contracts with the company after the U.S. Department of Defense blacklisted Anthropic's model following a dispute. Private-sector adoption of Anthropic's AI coding tools rose in the meantime, however, helping its commercial business continue to expand.

Anthropic has sued the Trump administration to overturn the blacklist designation, and the case is still ongoing. President Donald Trump said in an April interview with CNBC that a settlement between Anthropic and the Defense Department was possible.

Anthropic is also signing infrastructure deals to expand capacity as its business grows. Last month, it struck a deal with SpaceX, also a rival in AI, to use available computing resources at the Colossus 1 data center in Memphis, Tennessee.

As part of that agreement, Anthropic will pay SpaceX $1.25 billion a month through May 2029, SpaceX said in a business document.

Kim Jung-a, guest reporter, Hankyung.com, kja@hankyung.com