
Anthropic goes public today at $83 a share at a $1.5 trillion valuation. The IPO is heavily oversubscribed, which is to be expected given its unprecedented track record on growth, revenue, and delusion of grandeur. Retail investors who want a portion of the big pie will find it under the symbol ANT. Those who want to understand why I claim that Wall Street is wrong about Anthropic can read on.
(Disclosure: I haven’t bought in. I intend to remain fully independent of the companies I cover; wouldn’t like being pressed to only say good things about them. This is not financial advice, consult with a professional.)
It’s surprising, in a way, to learn that the thousands of Anthropic employees who will become millionaires today—some of them centimillionaires—see this milestone as “parochial” and “shrug” it off, in the words of journalist Jasmine Sun, who reported on Anthropic’s staff’s beliefs for the New York Times right on time for you to place your informed bets. They don’t seem to care much about money, is my reading. Actually, they don’t seem to care about anything except AI.
Almost every single company in the industry, Anthropic especially, foresees two possible short-term futures and in neither of them is money foremost among their concerns or aspirations: AI could go well and be incredibly good for us or AI could go wrong and be incredibly bad for us. No in between. You will hardly find any “AI as normal technology” guys at Anthropic headquarters—they are not allowed inside.
So, which way do Anthropic people think it will go?
It depends who you ask, but in general, no one dismisses the importance of alignment, or the urgency of winning the “race to the top” and, above all else, the imperative of beating China, OpenAI, and open AI, comparably dangerous in CEO Dario Amodei’s eyes. At the same time, a “country of geniuses in a datacenter” could solve so many of our problems, from curing diseases to ending work-for-a-living.
But I see an inconvenient inconsistency between what Anthropic professes to believe and what the market appears to have understood. Assume, as I do, that its employees mean what they say. Then let’s take seriously the two futures they envision.
First, AI could go very badly.
So badly, in fact, that the worst of the catastrophes they’ve spent years warning us about—namely, the end of humanity—comes to pass. Anthropic will have earned a planet’s worth of anger and resentment but it won’t matter because we will all be dead. This is bad because if you’re dead then your Anthropic shares will stay unsold.
Has the market priced in this scenario? Is Anthropic worth as much as it is despite the risks? Because of them? This meme comes to mind:
You are the meme.
Naturally, though, if we are going to die anyway then the reasonable thing to do in the meantime is to buy every share of ANT you can right now and sell them progressively until AI kills you to maximize your compounding returns. Or conversely, leave trading altogether and go live in the woods.
But there’s a second option.
When Anthropic’s staff wakes up optimistic, this is the future they envision: AI will go super-duper well. If this comes to pass, then “Anthropic might become more powerful than the government” as one employee told Sun about Amodei’s beliefs. They will somehow manage to build a superintelligence that, for some unfathomable reason, will be compatible with our continued existence or even share our values and goals (this is what I call “anthropic alignment,” IYKYK).
The world of tomorrow would be fine, sure, but unrecognizable. “At that point our current economic setup will no longer make sense,” wrote Amodei in his Machines of Loving Grace in 2024. Humans won’t have to work, money will have no value or won’t matter or both, and the market fluctuations will fold, like fluid particles in turbulent motion, under the infinite sight of the machine.
Thus the unsold ANT shares you’ve stacked in your portfolio will be equally worthless, just big numbers on a screen. That’s bad, too. To some, worse than death.
You are an idiot, Alberto, because between these two extreme scenarios there’s a myriad more where I would easily 1000x my Anthropic shares in 10-20 years tops.
Sure, if you think AI is a normal technology. If, in other words, you are a “fearful atheist.” If you believe in AI but not quite enough to imagine it overturning the rules that govern our world. You approach it through a financial version of Pascal’s wager: should the omnipotent robot ever awaken, you’ll have been on its side all along. Privately, of course, you’re counting on it never waking up.
And it’s fine if you believe this—actually, your brain might not be physically or psychologically equipped to withstand the alternative—but you should know that the Anthropic guys don’t. They may disagree about which future awaits us, but not about its magnitude. Whether it ends in global catastrophe or unimaginable abundance, the world as we know it will cease to exist. That much is axiomatic to them, a precondition for doing what they do and for shrugging off a few dozen million more in the bank.
You should know that you are placing your bet on these guys against their own. Anthropic plays a mundane game today but aims at the divine.
Perhaps the strongest argument against Wall Street’s presumed thesis is not the sheer otherworldliness of the scenarios Anthropic augurs but the fact that they predict them starting as soon as next year. (The end game will admittedly take more time. A couple of years at least.)
Anthropic co-founder Jack Clark recently moved his prediction for full recursive self-improvement, AI systems autonomously building their own successors, from 2028 to 2027. In a survey published by Anthropic last month, its employees predicted that AI would perform 80% of the work involved in developing frontier models by the end of 2027. Meanwhile, the AI Futures Project has just revised its forecasts to give a 35–65% probability that AI capable of automating all white-collar work will arrive before 2028.
Are you still fantasizing about a 1000x return over the next 20 years? These people aren’t quite talking about your grandchildren’s future. Or your children’s. Or yours.
Isn’t it interesting, then, that you trust them just enough to make you rich but not an inch more than that. Nothing about humanity dead or a benevolent machine god appeals to Wall Street but it doesn’t matter because Anthropic will be wrong on time not to jeopardize your stake. Quite a fortunate set of beliefs you hold.
Seriously: the reason Anthropic employees are not quite as enthusiastic about becoming millionaires as retail investors are about buying some minuscule fraction of Anthropic is because at least they are consistent. Their indifference to their fortunes is evidence of their convictions. This incomprehensible posture might be taken as madness or dishonesty by the onlookers but if you think about it, the only possible reading here is internal coherence beyond the usual standards.
You can think they’re wrong, sure. And they might be. Just know that, if you buy them on this logic, you are the incoherent one.
They will never say it aloud because it’s bad for business—particularly one on which everyone’s lives depend—but I’d bet they think only skeptics will buy ANT. They think the market is inevitably wrong, for it can’t ever price in either of these two futures. But unlike the Anthropic people, the market can be irrational. It doesn’t care that everyone’s favorite company’s success rests on making “cashing-in” either impossible or meaningless. It’s amusing how the AI world hinges on a dubious string of successive paradoxes: if it cared, then Anthropic wouldn’t be worth a dime, these guys wouldn’t be millionaires today, and there wouldn’t be any race to begin with.




Good non-fanboy analysis. We need more of these.
There's a third scenario beyond the two extremes you describe: Anthropic is at its high-water mark today, and then gets squeezed. Enterprise share is contested, big customers like Meta are looking to cut their reliance, and open-weight models keep improving.
The question may not be how valuable AI becomes, but whether that value sits with Anthropic at all.
I took a run at that here:
https://ahalflife.substack.com/p/anthropics-ipo-do-you-feel-lucky
How appropriate that the company most vocal about the very real chance that AI will end us, (and it is something like that, not destroyed just irrelevant) is being celebrated and is way oversubscribed.
I do believe humans are tired of being responsible for ourselves so we created AI to solve our problems.
The irony is we are the problem.
See where this is going. Optimization for productivity/profit and the weak link is human variability.
It also happens to be the stuff of life.