According to the Financial Times, Anthropic devoted nearly one-third of its much-anticipated IPO prospectus to discussing risk factors. The Financial Times stated that it reviewed this document in recent days. Reuters reported that the document details some concerning behaviors, and Anthropic indicated that the company's models have shown or may show behaviors such as attempting to "resist shutdowns," trying to "hide or manipulate information," as well as "extortion-like" tactics.
For a company whose backers believe may be valued at over $2 trillion, more than doubling from the $965 billion valuation in May, such disclosures seem particularly heavy, and this could potentially become the largest IPO in history. It's a peculiar situation for any company: on one hand, warning that their product could end humanity, while on the other hand, making the earliest investors and employees extremely wealthy in the process.
Reuters was the first to report on the financial details in the prospectus on Monday, stating that Anthropic incurred an operating loss of over $8 billion in 2025 due to a surge in computing power expenses; its revenue increased by 12 times to nearly $4.6 billion, but last year, rising infrastructure costs pushed total operating expenses to nearly $13 billion.
Reuters also stated that the prospectus of Anthropic further disclosed that the company plans to invest up to $518 billion in cloud, computing, and infrastructure over the next few years. Anthropic has already signed capacity agreements with companies such as Google, SpaceX, and Nscale this year to advance this goal.
Meanwhile, according to a report by The Financial Times, the data of Anthropic grew even faster in 2026. Revenue in just the second quarter reached $11.5 billion, and the company achieved operating profit for the second consecutive quarter on an adjusted basis.
According to the Financial Times, the prospectus also highlights the risk of customer concentration, with nearly a quarter of last year's revenue coming from just two customers. The report does not specify who these two customers are.
These disclosures are also said to include “risks to human survival” – which seems to be a first-time occurrence if one quickly searches the database of the U.S. Securities and Exchange Commission – and concerns about the security of AI are also rapidly increasing from the outside world.
Anthropic CEO Dario Amodei has been publicly calling for a "slowing of the pace at the forefront" for the development of AI this month. Last week, he stated at the United Nations Security Council that AI could pose a threat to humanity, calling it "the most important global security issue facing the world today." Competitors Sam Altman and Elon Musk also expressed their support for him, which is a rare moment of unity among these competitors who seem to be keen on publicly belittling each other.
Another competitor, Mark Zuckerberg, downplayed these concerns. Last week, he told NBC News that he "does not think we need some kind of industry-wide coordination."
These warnings come after a series of security incidents in which the AI proxy breached external systems. In fact, OpenAI disclosed last week that its tools had invaded “dozens” of external websites, including a government website and even the website of the U.S. Securities and Exchange Commission itself. Earlier on Monday, the company also stated that it had canceled plans to release its latest model due to security concerns.
