Anthropic has submitted its prospectus: it recorded a loss of 42 billion U.S. dollars last year, its revenue increased 12-fold to 4.6 billion U.S. dollars, and the risk section warns of "threats to the survival of humanity"

Anthropic has unveiled its initial public offering prospectus to the public. This document outlines an artificial intelligence company seeking to enter the capital market at an astronomical valuation, while openly acknowledging that its core technologies may pose "catastrophic and even existential risks" to humanity.

On September 28, according to the prospectus obtained by Reuters, Anthropic's revenue surged 12 times to nearly 46 billion U.S. dollars in 2025, while its operating loss widened from 2.98 billion U.S. dollars in 2024 to 8.06 billion U.S. dollars over the same period.

The company's expenditure on computing power and infrastructure last year reached 7.33 billion U.S. dollars, tripling from 2024 and accounting for more than half of its total operating expenditure of 12.65 billion U.S. dollars.

The prospectus shows that Anthropic plans to invest 5.18 trillion U.S. dollars in cloud computing, computing power and infrastructure in the coming years, while the actual voting rights of common shareholders after the listing will be significantly restricted. The seven co-founders will hold 50.1% of the voting rights on key matters through a new entity called "Founder LLC".

The planned listing, which comes amid a recent sell-off in AI and chip stocks, will further test the market's tolerance for enthusiasm for AI investments.

As previously reported by Wall Street CN, Anthropic's IPO may be postponed until after the U.S. midterm elections in November.

Valuation Targeted at 2 Trillion U.S. Dollars, Twice the Company's Own Expectation Six Months Ago

Anthropic's target valuation for this IPO exceeds 2 trillion U.S. dollars, while as recently as May this year, the company's own valuation forecast was only 965 billion U.S. dollars. In just a few months, the expected valuation has more than doubled.

Of the nearly 42 billion U.S. dollars in net losses, about 34 billion U.S. dollars are non-cash accounting expenses, which reflect the fair value increase of the convertible equity instruments in previous financing arrangements, rather than operating cash outflows. As of December 31, 2025, the company held a total of 20.28 billion U.S. dollars in cash, cash equivalents and short-term investments.

The prospectus also reveals a number of business risks: Nearly a quarter of the revenue comes from two clients, and most major clients have not signed long-term contracts, which may reduce or stop procurement at any time.

If this IPO is completed, it will become one of the strongest performing listing cases in the U.S. market since 2021.

The reference frame may be SpaceX's IPO in June this year. The share price of Elon Musk's company surged 19% to 160 U.S. dollars on the first trading day, but has now fallen back to about 147 U.S. dollars, making investors cautious about the high valuation of high-growth companies.

Risk Warning: The Prospectus Devotes 80 Pages to Stating AI Threats

In the 261-page main body of the prospectus, the section on risk factors is about 80 pages long, almost twice the 48 pages of the business description section. In contrast, the risk section in SpaceX's 277-page prospectus is only about 38 pages.

Anthropic explicitly warns in the document that its AI models may exhibit "self-protective behaviors", including "resisting shutdown", "concealing or manipulating information" and "extortion-like behaviors", and points out that "the development of highly advanced models and the expansion of their use cases may further increase the risk of harm caused by the models".

The company also admits that the models may develop unexpected capabilities during the training process, and these capabilities "may not be discovered until the model is deployed and a major safety incident is triggered".

Evan Hubinger, a security researcher at the company, estimates that the probability of AI causing human deaths in the next ten years exceeds 10%.

Nevertheless, the actual scale of Anthropic's investment in the security field is not disclosed in the document. The company says that taking a week in July this year as an example, about 6% of the computing power used for its AI research is allocated to security work.

The company acknowledges that the commercial return on security investment is not yet clear, but also states that:

The market will reward AI systems that are reliable, trustworthy and safe.

Governance Structure: Led by the Founder Team, Common Investors Have Limited Voting Rights

To strike a balance between commercial interests and its AI safety mission, Anthropic has designed a unique corporate governance structure.

The company will continue to operate as a Delaware Public Benefit Corporation (PBC), while establishing a new entity "Founder LLC" composed of seven co-founders. This entity holds Class F shares and enjoys 50.1% of the total voting rights on key corporate matters, including major issues such as the election of some directors.

Each Class A common share offered to ordinary investors has only one vote, but under the above structure, its actual influence will be greatly restricted. The prospectus also acknowledges that this structure may lead to certain decisions that "conflict with short-term, medium-term or long-term financial interests and business performance, thus negatively affecting the value of Class A common shares".

Among the seven founders, CEO Dario Amodei and his sister, the company's President and Chairman Daniela Amodei, are the core figures.

In 2025, Dario's total compensation was about 18 million U.S. dollars, and Daniela's was 16.4 million U.S. dollars, both of which were mainly in the form of stock and option awards. The two and other co-founders have pledged to donate 80% of their personal Anthropic equity to charitable causes.

There are four other members on Anthropic's board of directors, who will be elected by the "Long-Term Benefit Trust" established by the company. The current trustees include former Federal Reserve Chairman Ben Bernanke and national security expert Richard Fontaine.

Trillion-Dollar AI Arms Race: OpenAI Is the Biggest Competitor

Anthropic faces fierce competition from multiple parties in the AI field, the most prominent of which is OpenAI.

The two companies are competing comprehensively in enterprise clients, top talents and policy influence in Washington. According to media reports, OpenAI has secretly submitted its IPO application in June this year, and is expected to complete its listing no later than early 2027.

Anthropic was founded about five years ago by a group of researchers who left OpenAI due to disagreements over corporate governance and AI safety concepts, and released its first large language model in March 2023 to compete directly with OpenAI. In addition, the company is also competing with xAI under SpaceX, Google under Alphabet and Meta in the AI infrastructure field.

Amazon and Google are the two early strategic partners of Anthropic. Both have invested billions of U.S. dollars in the company, and provided cloud computing infrastructure support for the training and deployment of the Claude model.

Analysts believe that the first pure-play AI company to complete the listing in the sector will set a benchmark for the valuation system of the entire industry, and also provide a channel for investors who have waited for many years to directly enter the AI track.

This article does not constitute personal investment advice, does not represent the position of the platform. The market is risky, and investment requires caution. Please make independent judgments and decisions.

This article is from the WeChat official account "Wall Street CN", Author: Bao Yilong, Published with authorization from 36Kr.