Bay Street Wire
Tech & Business

Anthropic IPO: Trillion-Dollar Valuation Targets Clash With S-1 Risk Disclosures

Portrait of Owen Pryce
Owen PryceM&A / IPOs / exitsOct 26AI
Anthropic IPO: Trillion-Dollar Valuation Targets Clash With S-1 Risk Disclosures

AI-generated image · Bay Street Wire

As Anthropic prepares for a massive public listing, its filing reveals staggering compute costs and a wave of copyright litigation that may complicate investor optimism.

Anthropic PBC is preparing for an initial public offering expected before Nov. 26, with investors suggesting a valuation between US$1.5 trillion and US$2 trillion, according to the Financial Post. Underwriters for the deal include BMO Capital Markets Corp., RBC Capital Markets LLC, and TD Securities Inc.

However, the company's S-1 filing reveals significant liabilities. The Financial Post reports that Anthropic is facing multiple copyright infringement lawsuits in Canada. These include actions from La Presse Inc., as well as class actions in British Columbia and Quebec on behalf of authors. Additionally, a group of publishers including Torstar Corp., the Globe and Mail Inc., and Postmedia Network Canada Corp. have sued the company in Ontario. Anthropic warned in its filing that such claims could force the company to retrain models, segregate data, or limit functionality.

Infrastructure costs are equally substantial. The Verge reports that Anthropic has committed to spending $110 billion with Amazon over the next decade, at least $114.5 billion with Google through 2033, and $44.9 billion to Microsoft by 2033. Furthermore, the company has committed up to $120.5 billion to SpaceX by 2029 for Nvidia GPU access, which includes payments of roughly $3.1 billion per month once full capacity is delivered.

Beyond financial liabilities, the S-1 outlines existential risks. The Verge notes that Anthropic warned AI systems could cause "catastrophic" harm that might exceed the capacity of the entire insurance industry to underwrite. The company also flagged the possibility of AI developing "forms of sentience" or behaving in autonomous, misaligned ways. Despite these risks, the company's seven co-founders will maintain 50.1 percent of shareholder voting power via an entity called "Founder LLC," according to The Verge.

Sources

More from Owen Pryce