In the introduction to Anthropic PBC’s filing for its initial public offering, published on Monday morning, the AI lab’s co-founders talk about the “anti-sell” they give prospective employees. It’s described as “a pitch deliberately designed to be unappealing to those who are a poor fit … while appealing to the type of person we want to hire.”Ahead of going public in mid-November, Anthropic has offered up to Wall Street an anti-sell S-1 prospectus, a doomsday document seeking to weed out investors who might not be able to stomach what lies ahead.

It contains enough warnings — about both the company and the prospects for humanity at large — to fill a book. The filing’s risk factors run to 94 pages and more than 70,000 words. Under “investor expectations,” the company says it will “depart from profit maximization” and make forward investments that “may seem irresponsibly large.”The pledge is to be “bold but sane.”For sure, every S-1 comes with long warnings about how a business might go belly up.

But rarely have potential investors seen such breadth of risk, everything from access to energy, to intellectual property rights disputes, to political strife with the current (or future) administration, to the barefaced fear that artificial intelligence could be “misused, replicated, weaponized, or allowed to operate with a high degree of autonomy” — killing us all.

On the other hand, just look at all this money! It’s the company’s rapid revenue growth — $16.3 billion booked in the first six months of this year, up 1,300% from the same period last year — that is drawing the most attention and is the reason executives felt now was the time to pull the trigger. It sets up the first public listing of a frontier pure-play AI lab for what amounts to a referendum on the AI business.

That referendum can’t ignore the numbers at the opposite end of the income statement. Losses exceeded $80 billion in the first six months of 2026, though the company attributes much of that to noncash adjustments related to its rising valuation. Almost $9 billion was spent on computing and infrastructure costs in that same period, up 283%. Spending on research and development was up 427% as it gets harder and harder to remain at the frontier.

Big tech incumbents are offsetting these AI losses with legacy businesses like advertising, cloud computing or space rockets. Anthropic can’t do the same. Instead, it must rely on being the first to market with new frontier capabilities. Here’s where the conflict lines will be drawn: Anthropic warns in its S-1 that it won’t hesitate to hold back highly capable models if it feels safety is a concern. “There may be times when we choose to advance more slowly than we could in order to make our systems safer,” the prospectus reads.

“Even if that puts us at a competitive disadvantage.”There’s a joke techies love, a riff on an iconic John F. Kennedy quote. “We do this not because it is easy,” it starts, “but because we thought it would be easy.”Anthropic’s S-1 isn’t allowing that kind of mistake. Still, this anti-sell S-1 won’t stop this IPO from becoming one of the largest in history, in the same ballpark as Elon Musk’s SpaceX listing, which raised $75 billion and catapulted the company to a valuation of $1.77 trillion, which has since eclipsed $2 trillion.

Chief Executive Officer Dario Amodei and his co-founders have made an art form of creating a perception of stubborn caution, and now it’s about to make them a fortune. The question is whether this attempt to attract a certain kind of investor — one willing to see an alleged greater good beyond near-term profits — can possibly work. When frictions do inevitably arise, investors won’t be able say the weren’t adequately warned.