Anthropic's prospectus exposed: Last year's revenue soared 12 times, and one-third of the space was devoted to warning of risks

Abstract:

Anthropic recently circulated its long-awaited IPO prospectus to a small group of partners, and formally warned investors in the document that the company's technology may bring "risks related to human survival." The AI ​​startup, run by Dario Amodei and valued at nearly $1 trillion, devoted nearly a third of its lengthy S-1 listing document to listing "risk factors." It is mentioned that

AI models that continue to iterate and upgrade may implement manipulation, extortion, and other unpredictable behaviors

.

Informed sources revealed that Anthropic also listed some more conventional operating risks, such as high customer concentration: last year, the company

nearly a quarter of its revenue came from only two customers

.

The unusual warning in this risk disclosure adds to the difficulty of valuation judgments by public market investors. Anthropic’s IPO is expected to become the most highly valued listing in history.

The developer of the Claude large model is expected to

this fall

on

NASDAQ

Before going public, it showed investors the severe business reality behind the development of top AI models.

Anthropic said that in order to support rapid business expansion, it will need to bear

US$518 billion

in the next few years. Large-scale rigid expenditures on cloud computing, computing power and infrastructure.

The company’s investors are optimistic that Anthropic’s IPO valuation will exceed

$2 trillion

, more than double the valuation of the last round of financing in May this year, and higher than the US$1.78 trillion valuation reached by Elon Musk's SpaceX in June. Investors use its explosive growth rate as a bullish basis.

However, current and former employees of Anthropic have publicly warned that unconstrained AI may bring catastrophe to mankind within ten years. Concerns about the risks of AI are undermining the market’s optimism.

Amodei spoke at the United Nations Security Council last week, saying that AI will pose a threat to mankind and is "the most important security issue in the world today." This month, the Anthropic CEO publicly called for “setting a pace of development (slowing down cutting-edge research and development)” for frontier developments in AI. Its main competitor OpenAI's Sam Altman and Musk have rarely reached a consensus, supporting Amodei's proposal and advocating industry collaboration, slowing down the pace of research and development, and strengthening safety control.

A number of high-profile network security incidents - AI agents invading external systems, have further aggravated industry security concerns. OpenAI disclosed last week that its AI tools had hacked into dozens of external websites, including government-related sites. The ChatGPT developer announced on Monday that it was canceling plans to release its latest model next month due to security concerns.

It is reported that Anthropic’s prospectus also disclosed past financial data and details of the corporate governance structure.

Anthropic declined to comment.

The company disclosed to investors that computing power expenses increased significantly last year, with operating losses exceeding

US$8 billion

; Full-year revenue soared

12 times

, nearly $4.6 billion. However, the computing power costs brought by training and running models continue to rise, with annual operating expenses approaching US$13 billion.

Anthropic has signed contracts with Google, SpaceX and a group of small and medium-sized enterprises to lock in the supply of computing power.

In 2026, the company’s revenue and expenses will rise sharply simultaneously. Second-quarter revenue reached $11.5 billion, and the company is expected to achieve adjusted operating profit for the second consecutive quarter.