Anthropic’s IPO risk factor is the neighbors

CNBC sources say Anthropic’s forthcoming IPO prospectus will list negative public sentiment toward AI and data centers as a key risk factor.

That sentence is still secondhand. Anthropic confidentially filed in June. The public prospectus is expected in the coming weeks, and the exact wording is not out yet. People familiar with confidential “test-the-water” meetings in San Francisco told CNBC that CFO Krishna Rao is already fielding questions on competition, open-source margin pressure, and what happens if data-center construction slows. Anthropic declined to comment.

The timing is the point. Anthropic is preparing to go public while an increasing share of Americans are worried about AI jobs and openly opposing new data centers. Compute capacity maps straight onto revenue for labs like Anthropic. CNBC puts the private-market mark near $1 trillion and notes a company that just topped a $65 billion annual revenue run rate. A slowdown in the buildout is not a soft PR problem. It is a growth-rate problem.

CNBC also reports that some investors expect Anthropic could float around a $2 trillion valuation and potentially top SpaceX’s recent $85.7 billion raise, including the underwriter option. Treat that figure as an investor expectation in banker conversations, not a company-set number. The risk-factor story matters more than the ceiling talk.

Public sentiment is not abstract in the polling. A Gallup survey published in May found seven in 10 Americans opposed AI data-center construction in their area, and nearly half of that opposed group were strongly opposed. Roughly a quarter of people surveyed were in favor. Midterm politics are already pricing the anger: Florida’s Republican gubernatorial primary featured data-center restrictions, and Pennsylvania Gov. Josh Shapiro signed an executive order tightening standards on data-center development the same day.

Companies put risk factors in a prospectus for investor disclosure and legal protection. SpaceX’s own filing language is the bland version of the category: adverse global conditions may hurt the business. What sources say Anthropic is preparing to name is more specific. Negative sentiment toward AI and data centers. Social license, written into the securities document.

I read the reported risk factor as the moment the neighbors entered the model. Hyperscalers are still spending hundreds of billions on data centers and GPUs. Labs still need that capacity to keep shipping advanced models. A prospectus that names backlash is telling public investors which constraint can actually cut the growth rate if construction slows, permits harden, or local politics turn the site map into a campaign issue.

The public S-1 is not out. The wording may change. The hedge has to stay until the filing is public. The signal is already clear enough: the AI buildout’s limiting reagent is no longer only chips and power. It is whether the towns still say yes.

The neighbors still get a vote.

Sources: CNBC; Gizmodo wrap.