Anthropic Files Confidential IPO with SEC, Pulls Ahead of OpenAI in the Race for a Trillion
The maker of Claude submitted a draft S-1 less than a week after a $65 billion Series H led by Altimeter and Sequoia. The annualised run rate surged to $47 billion.
On Monday (1), Anthropic submitted a confidential draft prospectus to the Securities and Exchange Commission, a move that positions the maker of Claude ahead of OpenAI in the race for the first IPO of a frontier AI lab. The announcement was made via a statement on the company’s own website. "This gives us the option to go public after the SEC completes its review. The proposed initial public offering will depend on market conditions and other factors," the company stated. The number of shares and offering price have yet to be determined.
This filing comes less than a week after the $65 billion Series H led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, which elevated Anthropic's market valuation to $965 billion. This figure stands $113 billion above the $852 billion valuation that OpenAI received in its March round, and is just shy of the trillion-dollar milestone crossed by only three publicly traded companies: Nvidia, Apple, and Microsoft.
Rapid Revenue Growth
The contrast with the bubble theory that has been repeated since 2024 lies in the revenue. Anthropic informed investors that the annualised run rate, a metric that multiplies the most recent month's revenue by twelve, increased from around $10 billion in June 2025 to nearly $47 billion by May 2026. Internal projections presented during the Series H indicate $50 billion by the end of July. Over a twelve-month window, the company quintupled its annualised revenue base.
Revenue is coming from two reinforcing fronts. The business API operates on Anthropic, AWS, and Google Cloud, and has formed global partnerships such as the integration of Claude on KPMG's Digital Gateway platform, announced on May 19, which projects the model across the firm's 276,000 consultants in 138 countries. Claude Code, a paid tool for engineering teams, has become a significant revenue stream following the delivery of parallel subagents in May. This combination is what sustained the leap from a run rate of $9 billion at the end of 2025 to the current level.
The IPO Race
OpenAI is also preparing to file for an IPO with Goldman Sachs and Morgan Stanley, according to Reuters, targeting a debut in September. The difference is that Anthropic got there first and did so confidentially, taking advantage of provisions under the Jumpstart Our Business Startups Act that keeps the S-1 hidden from the public until the company decides to launch a roadshow. Complete financial numbers will only emerge if and when the offering progresses.
The simultaneous presence of two offerings approaching a trillion tests institutional appetite. In March, Anthropic raised $30 billion in three hours during a secondary round, and the Series H was oversubscribed by a wide margin. Sell-side analysts warn of absorption effects: a primary offering exceeding $30 billion consumes capital from pension and sovereign funds, which is typically distributed over months. This does not mean that the offering fails; rather, it means that other IPOs in the quarter may experience delays.
How the Market Interprets This Outside the US
For the European Union, the IPO transforms what was a private startup with opaque governance into a publicly-listed entity subject to quarterly disclosures. Regulators in Brussels, who will enforce the AI Act starting August 2, gain a wealth of data on contracted computing power, revenue by model type, and internal governance that they currently can only obtain through direct requests. The AI Office in Brussels has already indicated intentions to use market disclosures as inputs for investigations.
In India, where TCS and Infosys have reduced hiring in 2026 under pressure from AI automation, the outlook is different. The publicly listed Anthropic will need to meet quarterly targets. It has two obvious levers: raise prices, which is unlikely in the short term due to competition with GPT-5.5, or expand captive use via Claude Code, replacing outsourced human squads. Indian delivery centres that currently sell man-hours for tasks run on Claude will need to renegotiate contracts regarding who captures the generated savings.
The open question this week is not whether Anthropic will go to market. It is whether OpenAI can maintain its pace of private fundraising with a competitor already listed and operating under SEC regulations.