Anthropic files for IPO, revealing explosive growth and AI risks

Anthropic released its long-awaited IPO filing on Monday, confirming the startup's explosive growth while detailing numerous risks tied to artificial intelligence. The San Francisco-based creator of the Claude chatbot reported first-half revenue of $16.3 billion, compared to $4.6 billion for all of 2025. No valuation or listing date was specified, though media reports suggest a November debut and investor pressure to cut the valuation from $2,000 billion to $1,500 billion. The company remained deeply unprofitable, posting $8.4 billion in losses even after excluding a $72 billion accounting effect, with $265 billion in cloud computing commitments through 2030. Anthropic devotes nearly 100 of its 335-page filing to risks, warning its safety guardrails have failed or been misconfigured before and that AI models could cause harm autonomously. The company also flagged recursive self-improvement risks, where flaws could propagate across versions, and reserved the right to delay launches, as it did with the Mythos model in April.