Invest7 publishers3 min readPublished
Anthropic's IPO filing extends political risk to the customers behind more than 99% of its revenue
Anthropic's IPO prospectus warns government attitudes could hurt its commercial customer ties, though agencies supply under 1% of its revenue. That puts nearly all the revenue behind a possible $2 trillion listing within reach of how Washington sees the company.
The Investor · Invest desk
What happened
- In February the president ordered federal agencies to stop using Anthropic's models, and the Defense Department designated the company a supply-chain risk to national security.
- Commerce later lifted those restrictions and the models were redeployed, but the prospectus warns that similar actions could happen again.
- The Federal Trade Commission is running an industrywide probe of AI firms, and Anthropic is among the companies included.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- exposure Buyers of the shares would be taking on Washington's opinion of Anthropic across the more than 99% of revenue earned outside government, because one export order can reach every customer of a model.
- cost If the February actions produce the material losses Anthropic warns of, commercial accounts will bear most of them, since agency contracts are too small a slice of revenue to supply them alone.
- precedent A restriction that was later lifted still forced a full shutdown of two models, and the company's warning that it could recur leaves investors pricing reversible, all-customer outages as a repeatable event.
Government agency contracts are less than 1% of Anthropic's annual revenue, according to the prospectus seen by Reuters [2]. The filing still ties February's federal order against its models and the Defense Department's supply-chain designation [5] to its top line. "The company may experience material revenue losses or business disruptions attributable to these events," Anthropic said [6]. On the company's own split, a material loss from those actions would have to arrive mostly through the more than 99% of revenue earned outside government [1]. The prospectus names those customers and partners as exposed to how governments see the company [1].
June showed the route. To comply with a worldwide Commerce Department export order, Anthropic switched Fable 5 and Mythos 5 off for every customer, commercial accounts included [7]. The Reuters account does not say how long the models were offline or what the episode cost. The filing says measures like it can cause "significant reputational harm, including adverse media coverage, public scrutiny, and negative perceptions among existing and prospective customers, partners, employees, and investors," regardless of the ultimate outcome [9].
Government risk factors are routine. SpaceX's filing called strong relationships with US agencies critical and said any deterioration could materially harm its ability to keep existing business and win new work [3]. Reuters judged Anthropic's version broader, because it reaches past direct dealings to customers, partners and other commercial relationships [4].
The risk can resolve a few ways before a listing that could come as soon as mid-November at a valuation of up to $2 trillion [10]. One reading is lawyers drafting wide (or rather, drafting wide after a year that handed them examples). On that view June stays a one-off, since Commerce lifted the restrictions and the models went back up [8]. Another is a thaw: Amodei had dinner with Trump last Sunday [11]. The third runs the other way. The Federal Trade Commission is running an industrywide probe of AI firms that includes Anthropic [12], and Trump told TIME he had discussed the company's requests to be regulated with Amodei [15]. "Look, I say, if you're going to seek regulation, they can put you out of business," Trump said [13]. "But my attitude is, we're winning against the world by a lot, and I want to keep winning." [14]
I think the thaw is the weakest case. The two disruptions came four months apart [2], and Anthropic itself warns that the June kind of action could happen again [8]. A valuation of up to $2 trillion [10] has to carry that risk on the commercial line, where more than 99% of the revenue sits [1]. The view is wrong if the public prospectus shows revenue holding up after February and June, or if the company lists without another federal action against it.
What to watch
- Whether the public prospectus puts a number on revenue lost after the February order and the Defense Department designation.
- Any outcome from the FTC's industrywide AI probe that names Anthropic before the mid-November listing window.
- A new Commerce export restriction on an Anthropic model, the kind of action the filing says could recur.