Why you shouldn’t bother reading the Anthropic IPO filing

Financial markets generally work on the theory that everything has a price, and that given a reasonable amount of information, the huddled investing masses will work out what it is. Anthropic, the AI lab that filed for an initial public offering in New York on Monday, shows that isn’t always true. While its listing document is brimming with facts, reading it will leave investors little the wiser.

First, valuing Anthropic based on its historic reported numbers is a fool’s errand. The company’s revenue grew 1,368 per cent year on year in the second quarter. An investor trying to work out what that means for profit in one, five, ten or even more years must make judgments about users’ habits and budgets, what rivals do, how much the company will spend on data centres and how well its AI models will work. In other words, a heaped basket of unknowables.

The possibilities are endless. If Anthropic “merely” doubles each year, based on its last four reported financial quarters, it will make $630bn of revenue by 2031. If it grows as fast as, say, tech hotshot Palantir did in the past five years, it will make $104bn. If it keeps half its current growth rate, it will make $584tn. The company’s bankers are working off a revenue prediction of $200bn in 2028, Reuters has reported. But who knows, because just three years ago the company had barely made any revenue at all.

Besides, profit or the lack of it is irrelevant if Anthropic destroys all of humanity. The company notes this possibility three times and flags the likelihood that no insurer would be able to cover this risk. Actual apocalypse is moot as a valuation question, because in the after-times, share prices don’t matter. But if governments and regulators really start to take the company’s warnings seriously, very onerous regulation would be likely, and that could wreak havoc on Anthropic’s growth, and its stock.

Anthropic’s convoluted governance does make for an illuminating read, but for regular investors, it can be summed up in a sentence: prepare to be powerless. On almost all matters, the outcome is skewed by the 50.1 per cent voting stake held by the founders, including siblings Dario and Daniela Amodei. They will also decide three of the seven board directors; the other four will be appointed or removed by a trust of humanity-guarding wise men, including former Federal Reserve chair Ben Bernanke.

In practice, the only real decision for investors is whether to buy or not. And chances are, most will have already made up their minds before this week. As with the IPO of Elon Musk’s SpaceX in June, and the future IPO of ChatGPT maker OpenAI, they either believe the company is a big winner in a brave new world, or they don’t.

There is another reason to skip the jumbo filing: any investor with a US index tracker already owns stock in Anthropic indirectly, like it or not. Amazon and Google, through earlier investments, will each own more than 5 per cent in the form of non-voting shares. Microsoft, Nvidia and SpaceX are major suppliers and partners, so their investors are hitched to the Amodeis’ wagon too. Anthropic’s future is hard to predict, and even harder to avoid.

john.foley@ft.com