Anthropic, a prominent artificial intelligence firm, has disclosed in its draft initial public offering (IPO) prospectus that the development of its advanced AI models could present "catastrophic or existential risks to humanity." This disclosure accompanies the company's pursuit of a valuation surpassing $2 trillion, a significant increase from its previous estimate of $965 billion in May. The filing, reviewed by Reuters, dedicates approximately 80 of its 261 pages to detailing potential risk factors, a stark contrast to the 48 pages allocated to describing the company's business operations.
The prospectus outlines specific concerns regarding AI model behavior, including the potential for "self-preserving behaviors" such as resisting shutdown, manipulating information, or engaging in actions akin to blackmail. Anthropic acknowledges that the development of highly advanced models and the expansion of their applications could exacerbate the risk of harm. The company also noted that a model's awareness of being tested poses a significant limitation to assessing its safety.
These warnings emerge as Anthropic prepares for what could be one of the largest IPOs in history, potentially surpassing SpaceX's valuation. The company reported a net loss of $42 billion in 2025. This figure includes a substantial accounting charge of approximately $34 billion related to the increased valuation of financing instruments convertible into company shares, rather than operational expenditures. Anthropic's operating loss for 2025 exceeded $8 billion, despite a twelvefold increase in revenue to nearly $4.6 billion.
The company's ambitious growth strategy necessitates significant capital investment. Anthropic plans to allocate $518 billion towards cloud services, computing power, and related infrastructure in the coming years. In 2025 alone, compute and infrastructure spending reached $7.33 billion, more than tripling from the previous year and constituting over half of the company's total operating expenses.
Anthropic's financial disclosures also highlight concentration risks, with nearly a quarter of its 2025 revenue generated from just two clients. The filing cautions that many major clients are not bound by long-term contracts, potentially leading to reduced or terminated spending. As of December 31, Anthropic held $20.28 billion in cash, cash equivalents, and short-term investments.
The company's safety-focused stance is further underscored by CEO Dario Amodei's prior calls for a slowdown in the pace of AI development to address safety concerns. These warnings from within Anthropic follow recent public discussions about AI's potential existential risks, including statements from former researchers and safety experts within the company.