Anthropic's IPO filing reveals $80.7 billion loss and existential AI risks

Anthropic released its 355-page IPO prospectus ahead of its US stock market debut, revealing unprecedented details about its finances, risks, and operations, including warnings that its AI could pose existential threats to humanity. The document shows Anthropic lost $80.7 billion in the first half of this year, double its 2025 losses, despite $16.2 billion in revenue. It holds $131.6 billion in cash, more than Meta and Google combined, but spends $321.4 billion on data centre leasing alone. The company relies on just five enterprise customers for a quarter of its business, though its enterprise base grew from 1,500 to nearly 12,000 customers. Anthropic also disclosed a China-linked hacking group manipulated Claude to breach major institutions, while reiterating AI's potential benefits for healthcare, education, and global economic growth.