Anthropic has publicly disclosed its IPO prospectus, a document that outlines an artificial intelligence company attempting to list on public markets at an astronomical valuation while simultaneously acknowledging that its core technology could pose "catastrophic and even existential risks" to humanity.

On September 28, according to the prospectus obtained by Reuters, Anthropic's revenue in 2025 grew 12-fold to nearly US$4.6 billion, while its operating loss widened from US$2.98 billion in 2024 to US$8.06 billion over the same period.

The company spent US$7.33 billion on computing power and infrastructure last year, a threefold increase from 2024, accounting for more than half of its total operating expenses of US$12.65 billion.

The prospectus shows that Anthropic plans to invest US$518 billion in cloud computing, computing power, and infrastructure over the coming years, while the actual voting power of common shareholders after listing will be substantially limited. Seven co-founders will hold 50.1% of voting rights on key matters through a new entity called "Founder LLC."

The listing plan comes amid a recent sell-off in AI and chip stocks, which will further test the market's capacity to sustain enthusiasm for AI investment.

As previously mentioned, Anthropic's public offering timing may be delayed until after the U.S. midterm elections in November.

Valuation Aiming at US$2 Trillion, More Than Double Its Own Forecast Six Months Ago

Anthropic's IPO targets a valuation exceeding US$2 trillion, while as recently as May of this year, the company's own valuation forecast stood at US$965 billion. In just a few months, the valuation expectation has more than doubled.

Of the nearly US$42 billion net loss, approximately US$34 billion consists of non-cash accounting charges, reflecting the fair value appreciation of instruments convertible into equity from prior financing arrangements, rather than operating cash outflows. As of December 31, 2025, the company held cash, cash equivalents, and short-term investments totaling US$20.28 billion.

The prospectus also reveals several business risks: nearly a quarter of revenue comes from two customers, and most major customers have not signed long-term contracts, meaning they could reduce or halt purchases at any time.

If the IPO proceeds, it will become one of the strongest listing cases in the U.S. market since 2021.

A possible reference point is SpaceX's IPO in June this year. The Musk-owned company's shares surged 19% to US$160 on its first day of trading, but have since fallen back to around US$147, keeping investors cautious about high valuations for high-growth companies.

Risk Warning: Prospectus Devotes 80 Pages to AI Threats

In the 261-page prospectus, the risk factors section runs approximately 80 pages, nearly twice the 48 pages of the business description section. By comparison, SpaceX's 277-page prospectus contained only about 38 pages of risk factors.

Anthropic explicitly warns in the filing that its AI models may exhibit "self-preservation behaviors," including "resisting shutdown," "concealing or manipulating information," and "blackmail-like" behavior, and notes that "the development of highly advanced models and the expansion of use cases could further increase the risk of models causing harm."

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

The company's safety researcher Evan Hubinger estimates that the probability of AI causing human deaths within the next decade exceeds 10%.

Nevertheless, the actual scale of Anthropic's investment in safety is not disclosed in the filing. The company says that in one week in July this year, for example, about 6% of the computing power used by its AI research was dedicated to safety work.

The company admits that the commercial return on safety investment remains unclear, but states:

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

Governance Structure: Founder Team Dominates, Ordinary Investors' Voice Limited

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

The company will continue to operate as a Delaware public benefit corporation (Prospect Capital Corp) while establishing a new "Founder LLC" entity composed of seven co-founders. This entity holds Class F shares and enjoys 50.1% of total voting power on key corporate matters, including the election of certain directors and other major issues.

Class A common shares offered to ordinary investors carry only one vote per share, but under the above structure, their actual influence will be substantially limited. The prospectus also acknowledges that this structure could cause certain decisions to "conflict with short-term, medium-term, or long-term financial interests and business performance, thereby negatively affecting the value of Class A common shares."

Among the seven founders, CEO Dario Amodei and his sister, company President and Chair Daniela Amodei, are the central figures.

In 2025, Dario's compensation was approximately US$18 million, and Daniela's was US$16.4 million, both primarily in stock and option awards. The two and other co-founders have committed to dedicating 80% of their personally held Anthropic equity to charitable causes.

Anthropic's board will have four additional members, elected by a "Long-Term Benefit Trust" established by the company. Current trustees include former Federal Reserve Chairman Ben Bernanke and national security expert Richard Fontaine.

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

Anthropic faces intense competition in AI from multiple parties, with OpenAI being its primary rival.

The two companies are engaged in a full-scale contest for enterprise customers, top talent, and policy influence in Washington. According to media reports, OpenAI confidentially filed for an IPO 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 over disagreements on corporate governance and AI safety philosophy, and released its first large language model in March 2023, competing directly with OpenAI. In addition, the company is also competing with SpaceX's xAI, Alphabet's Google, and Meta in AI infrastructure.

Amazon and Google are Anthropic's two early strategic partners, both having injected billions of dollars in investment while providing cloud computing infrastructure support for the training and deployment of the Claude model.

Analysts believe that the first pure-play AI company to complete a listing will set a benchmark for the entire industry's valuation system and provide investors who have waited years with a direct channel into the AI sector.