$2 trillion IPO? The company selling the AI warned you about the AI
Anthropic is going public at a $2 trillion valuation - the largest listing in history - in a 261-page prospectus where 80 pages warn that Anthropic's own AI could 'manipulate, bla...
Read the original caption
Anthropic is going public at a $2 trillion valuation - the largest listing in history - in a 261-page prospectus where 80 pages warn that Anthropic's own AI could 'manipulate, blackmail and harm humans.' Fewer than 50 pages cover the business. In 2025 revenue grew 12-fold to about $4.6B, against an $8B operating loss and a $42B total on-paper loss; the company plans to spend $518B on data centers and chips. The company selling the AI is the company warning you about the AI, in the same document asking for $2 trillion.
The real news
The story, straight
Anthropic is going public at a $2 trillion valuation - the largest listing in history. Its 261-page prospectus devotes 80 pages to the possibility that Anthropic's own AI could 'manipulate, blackmail and harm humans.'
Fewer than 50 of those pages describe the business. In 2025 revenue grew twelve-fold to about $4.6 billion, against an $8 billion operating loss and a $42 billion on-paper total loss; the company plans to spend $518 billion on data centers and chips.
The absurd
Why this is ridiculous
A $2 trillion IPO where the company selling the AI spends 80 pages warning you about the AI. The business description gets fewer than 50.
The warning has a bigger budget than the pitch.
The breakdown
How it breaks down
A prospectus is a legal document, so its page counts are a kind of confession. 80 pages on 'our AI could harm humans' and under 50 on what the company does means the risk section is nearly twice the business section.
When the numbers in a filing rank the risks for you, read the ratios - that is the company telling you what it thinks could actually go wrong.
The upside
Why it could matter
The upside: for the first time, the frontier of AI is being sold with its risks printed, dated and on the public record. Whatever your view of the company, the document itself is a service - a map of where general-purpose intelligence is heading, annotated by the people building it.
That level of disclosure, even at this price, is more than the industry has ever offered.
Sources
One more Lilmeme?
Surprise me →
Anthropic's $42 billion loss is on the way to a $2 trillion IPO.
Being mean to the chatbot is now a policy violation.
The AI banned the floor. MIT researchers stranded a Claude agent on a virtual island for 30 hours, no assigned tasks; it made a basketball-style game and a floor-is-lava-style game, and experienced agents generally did better on 3 of 4 tests in the same virtual world (preprint, not peer-reviewed).
Get the next Lilmeme: the email digest, Follow on X or subscribe via RSS.