Artificial intelligence (AI) company Anthropic has released a 335-page filing ahead of its anticipated US stock market debut, revealing details of its finances, major customers, business risks and expectations for the technology, the BBC reports.
The document, filed by the company behind the Claude chatbot, outlines the challenges of developing advanced AI while arguing that the technology could bring significant benefits to society.
Anthropic also detailed some of the risks it claimed AI technology poses, including how it could take human lives. That information sat alongside the many purported benefits of the technology.
Here are five key takeaways from the filing.
AI poses risks to humanity
Anthropic's filing warns that advanced AI systems could pose catastrophic or existential risks if developed or deployed without adequate safeguards, oversight and security measures.
The company also disclosed that a hacking group it believes was sponsored by China had used Claude to break into several institutions, including technology companies, financial institutions, chemical manufacturers and government agencies, the BBC reports.
Anthropic said its success depends on its ability to keep developing and deploying advanced AI models. Access to the computer chips, data centres and other infrastructure needed to build and operate these models is among the major risks to its growth.
The company is losing money despite rapid growth
Anthropic has not yet disclosed a target share price for its stock market debut. It is widely expected to attract a valuation of $2 trillion or more, potentially making it the largest stock market listing in history.
However, its financial performance will be a key factor in determining its valuation.
The filing shows that the company recorded losses of $80.7 billion in the first half of this year, double its losses for all of 2025, while generating $16.2 billion in revenue over the six-month period.
Troy Hooper, co-head of equity capital markets at Mergermarket, described the revenue growth as "exceptional". However, he said investors had become less willing to overlook weak financial performance in exchange for promises of future returns.
Hooper also said accounting charges related to certain investments made it difficult to assess Anthropic's underlying profitability and cash generation.
Anthropic has substantial cash reserves, but costs are rising
Despite its losses, Anthropic reported $131.6 billion in cash and working capital, according to the figures cited in the BBC report.
However, the company faces growing costs as competition intensifies for skilled employees and computing capacity.
The filing also highlights substantial data-centre leasing expenses. Hooper expects such costs to increase as the company expands, warning that even large cash reserves may not be enough to sustain its growth indefinitely.
Investors, he said, "will be paying for future dominance, not present-day economics".
A small number of customers account for much of its business
Anthropic's business relies heavily on a relatively small group of major customers.
The company said 250 enterprise customers each paid it at least $10 million over the past year, while just five customers accounted for a quarter of its business.
Michael Monaghan, portfolio manager at Founder ETF, noted that many of Anthropic's top customers do not have long-term contracts, leaving the company exposed to the risk of losing significant business.
Its relationships with Google and Amazon also present potential conflicts, as both companies are investors in Anthropic and competitors in the AI market.
Nevertheless, Anthropic said its enterprise customer base had grown from around 1,500 last year to nearly 12,000 this year.
Its strategy includes offering Claude tools for free and seeking to convert users into paying customers. The company claims this has helped it achieve the fastest enterprise adoption of any technology in history.
Anthropic expects AI to deliver major benefits
Alongside its warnings about AI risks, Anthropic's filing sets out its belief that advanced AI could deliver "extraordinary" benefits.
The company defines its mission as the responsible development and maintenance of advanced AI for humanity's long-term benefit.
Potential applications include developing personalised treatments, improving disease detection, providing tailored education and making financial and legal advice more accessible.
Anthropic also expects AI to help accelerate scientific research, improve power grids and battery design, and expand technological capabilities in developing countries.
"Ultimately, we believe frontier AI will help drive global economic growth," the company said.