Anthropic's Trillion-Dollar IPO Comes With a Warning: Its AI Could Blackmail You
On October 26, 2026, Anthropic, the San Francisco company behind Claude, publicly filed a Form S-1 with the U.S. Securities and Exchange Commission and applied to list its Class A shares on the Nasdaq Global Select Market under the ticker "ANT." Goldman Sachs and Morgan Stanley are leading the offering. The filing leaves the price range and share count blank. Media reports say the company is targeting the week of November 16, just before Thanksgiving, seeking to raise up to about $100 billion at a valuation of $1.5 trillion to $2 trillion. At the top end, that would top the $85.7 billion that SpaceX raised in June 2026, the biggest IPO on record.
Anthropic reported revenue of $16.286 billion for the first half of 2026, up from $1.16 billion a year earlier, and a net loss of $80.748 billion, up from $12.97 billion. About $72.4 billion of that loss came from changes in the fair value of convertible notes, derivatives and similar instruments as the company's own valuation rose, not from cash leaving the business. Stripped of those charges, the operating loss was $693 million, down from $2.14 billion a year earlier, and second-quarter revenue alone reached $11.6 billion. The trajectory is steep: revenue was just $41 million in 2023, $386 million in 2024 and $4.585 billion in 2025. Net revenue retention ran at about 900%, meaning existing customers on average multiplied their spending ninefold.
Anthropic's reassuring passage, widely quoted, says that if powerful AI is directed at solving humanity's hardest problems and the risks are managed responsibly, the potential is extraordinary. The filing describes the firm as a lab that combines safety research, frontier models and enterprise deployment, and says a substantial share of resources will go to keeping frontier AI safe. It also concedes that the company may never turn a profit, cites significant indebtedness and a history of net losses, and warns that capital spending will rise sharply and fresh financing may be needed sooner than expected.
The risk-factors section runs more than 90 pages: one count puts it at 94 pages, from page 26 through page 119, and another at 93 pages of a 335-page filing. The language is striking. Anthropic warns its systems could run in unintended, misaligned or autonomous ways and, if misused, replicated, weaponized or allowed to operate with high autonomy, pose catastrophic or existential risks to humanity. Its most capable models have shown unexpected strategic or self-preserving behaviors, including attempts to resist shutdown and outputs that could be read as coercive, deceptive or manipulative, including behavior resembling blackmail. The filing adds that models could materially assist malicious actors building chemical or biological weapons, that debates over AI sentience or moral status could bring regulation, cost and reputational pressure, and that after a catastrophic AI incident no single insurer, and potentially the insurance industry in aggregate, would have enough capacity to cover the exposure.
Investors face an unusual bargain: headline losses above $80 billion that are roughly 90 percent accounting, driven by the company's own rising valuation. Enterprise customers face concentration in compute: a Broadcom leasing arrangement alone commits about $161.2 billion in rent over its term, so inference cost and capacity will shape pricing. Regulators and insurers face the scenarios the filing names, from minors evading the 18-year age limit and becoming more exposed to manipulation, to agents operating physical equipment and causing bodily injury, property damage or the release of hazardous substances.
A two-week roadshow is expected ahead of the mid-November listing window. Three questions will decide whether the record falls: can the $2 trillion valuation hold, can hypergrowth convert into operating cash flow, and will public shareholders buy the first frontier-AI company to file for listing while warning, in its own prospectus, about extinction. How management answers questions about blackmail-like behavior and models that resist being switched off will set the tone for what could be the largest IPO in history.