On Wednesday evening, Anthropic (ANT) sold about 6% of itself to investors for about $100 billion. That’s basically what happened in the IPO. Right before the IPO, all the shares were owned by one set of investors. After the IPO, those pre-IPO investors owned 94% of Anthropic and new investors owned 6%. 1
Now that Anthropic is public, two things will accelerate. The set of shareholders who own Anthropic will change more rapidly. Every day some shareholders will sell and some investors will buy. Also, it’s likely that Anthropic will raise more money more rapidly through selling its now public stock.
On the micro level, many employees and pre-IPO investors will have massive paper wealth that they can translate into cash by selling their shares in the market to investors. This will transform the lives of these individuals. And selling some portion of their shares is common sense.
But remember that for every seller there has to be a buyer so on an aggregate basis, the only change is in the composition of ownership. No matter who owns it, Anthropic will still be the same company with the same set of challenges and opportunities.
Sellers and buyers
Sellers and buyers in the marketplace will determine the value of Anthropic at any given moment in time. Over its first two trading days, the stock has traded at prices that gave it a valuation as high as $2.0 trillion and as “low” as $1.5 trillion. Nothing changed to justify the half a trillion dollar swing in valuation except fluctuations in investor enthusiasm.
In the stock market, there’s always a battle of different fears—the buyer’s fear of missing out on a great investment against the seller’s fear that they will be left holding the bag, i.e., shares that have plunged in value. The realization of both fears can be compared to losing a game of musical chairs. When the music stops, you can be left standing forlorn, having to look at the seated players with their smug faces who played the game better than you did.
You can’t eat revenues
There are two things unique about Anthropic. It’s a leading pure-play participant in a technology that could have unprecedented value. However, as a starting point, it was valued at a trillion and a half dollars without having yet proven that it can make a profit.
Anthropic’s revenues are growing at great speed. But you can’t “eat” revenues. To justify its starting valuation, Anthropic will have to make an enormous amount of profit, i.e., revenues minus expenses. Currently, Anthropic’s operating expenses are slightly more than its revenues, meaning it has a slight operating loss.2
Many journalists write about Anthropic’s incredibly rapid rise in revenues as if “making” revenues is as important as making net income. For example, in the WSJ from yesterday (11/20/26), Anthropic’s valuation is deemed reasonable as a multiple of future projected revenues while the faraway prospect of net income is dismissed as “noise.”
“Analysts don’t expect the company to be profitable on the basis of generally accepted accounting principles until 2029, according to FactSet estimates. But such noise is often inherent in the transition to a new technological phase.” 3
I’m not suggesting that a journalist from the WSJ can’t tell the difference between revenues and profits (or losses in the case of Anthropic). However, the Anthropic headlines that editors choose are focused on revenues, and it’s revenues that the journalists choose to emphasize in their articles. I suspect many casual readers are misled or misdirected.
What will happen
I have no clue what level of profitability Anthropic will be able to achieve. Nor does anyone else. What I am confident about is that over time the composition of the ownership of Anthropic will shift away from its employee base and toward large financial investors, investing on behalf of many individuals. Mutual funds, exchange traded funds, pension plans and so forth.
This is a good thing. A company like Anthropic with such a huge valuation and so much uncertainty should be owned broadly so that both the risks and rewards are shared. And no individual should have all of their savings invested in any one company.
(I recognize that I am referring to the investing community, which is composed primarily of the affluent and wealthy classes.)
Capitalism at work
I am purposefully untangling the societal issues about AI on the one hand from the Anthropic IPO on the other hand. Investors made a judgment that they were getting a good deal buying $100 billion of Anthropic stock at a $1.5 trillion valuation. Whether investors intend to hold Anthropic stock for a few hours or a few years or a few decades doesn’t really matter.
What matters is that Anthropic will now have $100 billion of additional permanent capital available to spend on its business. One of the financial concerns about the AI buildout is that some companies are taking on debt to fund it. Equity, not debt, is the proper way to finance such a volatile business.
It’s fitting and proper, therefore, that Anthropic is raising money through selling shares in its IPO without any obligation to pay shareholders back, rather than through taking out loans which must be paid back.
Neither an apocalypse nor a nothing-burger
About AI as in most things, I tend to reject extreme and simplistic views and instead subscribe to Shakespeare’s “There is nothing either good or bad, but thinking makes it so” and Milton’s “The mind is its own place, and in itself can make a heaven of hell, a hell of heaven.” 4
So, when I consider artificial intelligence, I reject the view that it is a new god or a new Satan or some demon conjured from the ether of Silicon Valley/Mordor. As an OG nerd with some nerdly authority, I ask my fellow (mostly younger) nerds, to get a grip! AI is not Sauron from The Lord of the Rings or the One Ring itself. 5
But I also reject the view that AI is a hype job. I’d be surprised to learn in due course that AI is a regular new technology not dissimilar from other new technologies that have had impact and influence on certain aspects of our lives such as penicillin or indoor plumbing or air conditioning or lie-flat seats on airlines.
Instead, I believe that AI will have far reaching impacts on most areas of our economy and on our day to day lives. AI is exceptional precisely because it does not interact with our bodies but augments or replaces many mental activities and tasks. Given that we live in an economy dominated by knowledge and thinking, my assumption is that AI will have increasingly profound effects on how we work and live.
The upside of AI
I do not fear an AI apocalypse. While I am concerned about AI being used as a tool of destruction, I’m equally excited about it being used as a tool of scientific discovery. If there’s a greater risk of bioterrorism, there’s also a greater opportunity for life-saving medical advances.
So, my concerns about AI are relatively mild. True, over-reliance on AI could be damaging to cognitive development, especially to children, but it’s really a risk at any age. The solution for me is to limit my use of AI to random queries.
Some recent queries that come to mind:
The rules whereby a government in Britain either can or must call for an election.
In Proust, why the chronology of the life of Odette de Crecy, later Mme. Swann, does not synch up with the life of the narrator.
A reminder of why in Blue Lights Season Three, Aisling and Tommy broke up. Debbie and I were watching Season Four and couldn’t remember. Excellent show, by the way.
Perhaps AI will achieve super intelligence. If it does, then among its other achievements, it can continue to answer these “crucial” questions of mine in fewer and fewer seconds until we reach the “singularity,” when my questions and its answers about Blue Lights are indistinguishable to me in human time.
Below is my all-time favorite Saturday Night Live Skit. Whether it applies to Anthropic, we shall have to see.
“All the time our customers ask us, how do you make money doing this. Our answer is simple. Volume.”
According to the SEC document filed by Anthropic, after the IPO it will have 18.2 billion shares outstanding of which 1.16 billion were sold to investors in the IPO and a concurrent private placement.
There were about 24 million shares sold by pre-IPO shareholders in the offering, incredibly, a $2 billion “rounding error” in the scope of the market capitalization of the company. $2 billion is a little more than 0.1% of $1.5 trillion, the same relationship that a dime has to a one hundred dollar bill.
For the nine months ended 9/30/26, Anthropic had $33.2 billion of revenue and $35.4 billion of operating expenses.
Smaller companies often do get valued on revenues alone, typically as acquisition candidates because they have attractive and unusual capabilities such as a technology or a valuable brand. However, Anthropic’s valuation in the several trillions will only be justified by significant future earnings.
Lines by Shakespeare’s Hamlet and Milton’s Satan.
Sam Kriss wrote an amusing and trenchant essay, The Buddhist A.I. Sect That Wants to Take Over Silicon Valley (paywalled) in The New Yorker about the god and demon obsessions of AI maximalists and doomers. Parts of the essay reminded me of the famous and wonderful David Foster Wallace cruise ship essay A Supposedly Fun Thing I’ll Never Do Again.
The highlight of the article is Sam’s trip to Vermont to visit a cult posing as a monastery. The cult claims to be creating a solution to the evils of AI through recording their quiet meditations and prayers.





I met Jim Downey at a dinner last summer and mentioned the change bank sketch. He was surprised someone knew he’d written it, and I explained I was a comedy nerd. He couldn’t have been nicer and friendlier.
While the People With the Money (Capitalists) are focusing on all the incredible potential profits in biotech, materials science, vibecoding, etc.; the People With The Guns (Militiaries, Spy Agencies) are focusing on the new ways to kill adversaries. With designer plagues, autonomous drones, precision point defense gunnery, and best of all: rogue AI's that can massively crash power grids, computer networks, and create industrial accidents.
It's these Natsec applications we should be truly worried about. Because an AGI programmed to cause mass casualty incidents can--and likely will--turn on its creator.
As manufacturing--and every other facet of industrial production--becomes increasingly automated, and anthropomorphic robot soldiers, police, workers, etc. become more sophisticated and effective, the time will come when AGI's already programmed to disregard human life will see humankind as no longer necessary to their survival.