The shift is happening in public: the first big frontier AI lab is going to market. On Monday, Oct. 26, Anthropic filed final paperwork with regulators for an IPO under the ticker ANT. Reports have said the offering could value it at $1.5 trillion to $2 trillion, which would put it alongside SpaceX, valued around $1.8 trillion when it listed in June. The prospectus does not say how many shares will be sold or at what price. The listing could come as soon as November.
Growth, and a very large loss
The numbers in the S-1 are brisk. Revenue for the first half of the year rose 1,300% to $16.3 billion. Adjusted operating income reached $1.6 billion, against a loss of $1.8 billion a year earlier. Moody’s Raj Joshi called that swing “a strong positive for the AI ecosystem”. Yet the company booked a loss of nearly $81 billion for January through June, up from $13 billion in the same months last year, and it has not yet turned an annual profit. Its last private round, in May, valued it at $965 billion. It had filed confidential papers with the SEC on June 1.
The prospectus gives 94 pages to risk factors, including a “catastrophic AI-related event” so large that insurance could not cover it. It also flags the danger that its models outrun the company’s ability to keep them safe, or are used to cause harm, along with political backlash over its views on policy and regulation. In a bullish passage, it says it believes AI will transform every sector of the economy by more than industrialization, electricity, semiconductors and the internet did.
A safety-first company in a jumpy market
What’s changing is who gets to tell the safety story to public investors. Anthropic began in January 2021, when seven former OpenAI staff, including siblings Dario and Daniela Amodei, left over worries about safe development. It is a public benefit corporation, and one technique behind its Claude models is constitutional AI, in which a model is trained to follow a written set of principles. Amodei has also pressed AI firms to “pace the frontier”.
The Pentagon fight is the sharpest test of that stance. Anthropic refused to drop contractual limits on mass domestic surveillance and fully autonomous weapons. The Defense Department answered by labeling it a “supply chain risk”, a tag once kept for entities tied to U.S. adversaries, and Anthropic has sued in two federal district courts. Those cases continue. Separately, the company has accused Chinese rivals of distilling Claude to train their own models.
Timing is the other story. Markets are jittery about AI. CoreWeave’s shares tumbled this month on doubts about financing its expansion, and Intel, Meta and Nvidia have also slid. Georgetown’s Jim Angel said going public brings in capital and lets investors cash out, but warned that “there’s concern that we may be in an AI bubble”. He still thinks it makes sense to “strike while the iron is hot”. SpaceX shares soared briefly after listing and have come back to around $150, where they began trading. OpenAI, meanwhile, has put its own IPO on ice, with Sam Altman calling a listing this year “ill-advised”. U.S. stock indexes rose Monday after Anthropic’s early announcement. Watch this space: Anthropic’s debut will show whether investors will pay for AI at this scale.
Sources
- Anthropic
- How Claude designed protein binders against scorpion neurotoxins
- AI giant Anthropic files for an IPO that could break records
Photo: Department for Science, Innovation and Technology (CC BY 2.0)
