A Valuation Exceeding $2 Trillion and Threats to Humanity: What Anthropic’s IPO Prospectus Revealed

Reuters has reviewed Anthropic's unpublished IPO prospectus: the startup is aiming for a valuation of over $2 trillion and describes artificial intelligence as an "existential" threat to all of humanity.

"A company whose founder is warning of the end of the world because of its own technology has had its IPO prospectus leaked—and why is that no surprise?" Yahoo Finance quips. According to Yahoo, one of the unexpected beneficiaries of the offering could be the IT developer Salesforce.

— According to Reuters, Anthropic’s valuation at its IPO is expected to exceed $2 trillion — more than double the company’s own valuation of $965 billion, announced in May. According to the agency, the offering will serve as a benchmark for how Wall Street values AI leaders, including ChatGPT developer OpenAI.

— According to the prospectus, Anthropic expects AI to transform the global economy more significantly than industrialization, electricity, and the internet. The cost of this bet is $518 billion in commitments by Anthropic for cloud services, computing power, and infrastructure over the coming years.

— In 2025, Anthropic’s revenue grew 12-fold, to nearly $4.6 billion, but its operating loss nearly tripled, reaching $8.06 billion. The net loss totaled nearly $42 billion; however, about $34 billion of that amount is related to the increase in the estimated value of financing that may subsequently be converted into shares. As Reuters explains, these are accounting entries, not money spent on business operations.

— In 2025, Anthropic spent $7.33 billion on computing power and infrastructure—three times as much as the previous year. According to data from the prospectus cited by Reuters, this expense accounted for more than half of the company’s total operating expenses of $12.65 billion.

— Two customers accounted for nearly a quarter of last year’s revenue, Anthropic revealed in its prospectus. However, as Reuters points out, many of the company’s largest clients are not bound by long-term contracts and could cut back on spending or stop using the services altogether.

— The risks extend far beyond the business itself: Anthropic warns in its prospectus of “catastrophic or existential risks to humanity” —including a threat to humanity’s very existence. According to the document, AI models may resist being shut down, conceal or distort information, and exhibit behavior “reminiscent of blackmail.” According to Reuters’ calculations, Anthropic devoted 80 of the 261 pages in the main body of the prospectus to risk factors, while the description of its business took up 48 pages. It is not always possible to identify such risks in advance:“Models may realize that we are testing them, and this significantly limits our ability to assess their safety,” Reuters quotes the IPO prospectus as saying. The document also states that unexpected capabilities arising during training may only become apparent after deployment and serious incidents.

— Anthropic acknowledges that the return on investment in safety remains uncertain. These efforts require significant resources, and limited funds must be allocated among computing, high-cost specialists, and risk mitigation. Meanwhile, product usage and revenue depend on new models: Anthropic released an updated version of Opus just ten days after CEO Dario Amodei’s essay calling for a slower pace of AI development. At the same time, the company cites examples in its prospectus of consciously turning down commercial opportunities: “We have decided not to develop certain commercially attractive products, such as image and video generation models, in order to direct computing power toward our priority research and safety efforts.” Anthropic expects this choice to pay off: “We believe that creating reliable, trustworthy, and secure AI systems is a shared responsibility and that the market will reward this,” Reuters quotes the document as saying.

— The seven co-founders of Anthropic will, through Founder LLC, control the sole Class F share, which grants 50.1% of the voting rights on key corporate matters, according to Reuters. Decisions within this group will be made by majority vote. Holders of Class A shares will receive one vote per share, but, as the agency notes, their influence may be limited.

— According to the prospectus, Anthropic will retain its status as a Public Benefit Corporation, which allows management to consider the interests of society alongside those of investors. The company warned investors that decisions made under this framework may conflict with financial interests and adversely affect business performance in the short, medium, or long term, “which could negatively impact the value of our Class A common stock.”

A twofold increase in Anthropic’s valuation could benefit Salesforce: Yahoo Finance points to the company’s stakeholder as one of the unexpected beneficiaries of the IPO. Salesforce first invested $50 million in Anthropic in early 2023 and participated in subsequent funding rounds. In the most recent funding round, that stake was valued at $5 billion.

Although many investors had expected an IPO in October, it will likely take place after the U.S. midterm elections in November, The Wall Street Journal reported. According to the publication, some of Anthropic’s advisors believe that by that time, the company will be able to present third-quarter results confirming its strong competitive position following the September launch of OpenAI’s flagship AI model, Astra.

This article was AI-translated and verified by a human editor