Anthropic IPOs: “Anthropic PBC and its seven co-founders raised a record $107.3 billion in an upsized IPO and private placement, becoming the first frontier AI lab to tap public markets for the development of thinking machines,” reports Bloomberg. Anthropic’s valuation is $1.5 trillion; demand is vastly exceeding shares available (as is pretty common for hot public offerings). “Anthropic sold 980 million shares for $83 each in the listing,” adds The Wall Street Journal.
Competitor lab OpenAI plans to go public next year. Anthropic had for several years been lagging behind OpenAI; now they’re tracking much more closely—or maybe Anthropic has overtaken them—in model sophistication, annual revenue, and other metrics.
The Reason Roundup Newsletter by Liz Wolfe Liz and Reason help you make sense of the day’s news every morning.
“In the summer of 2022, Anthropic decided not to make Claude, its A.I. chatbot, available to the public over internal concerns about safety. This allowed OpenAI—which had gotten wind of Anthropic’s internal prototype—to design a similar product, rush ChatGPT out the door first and gain a decisive lead in consumer mindshare,” writes Jasmine Sun in The New York Times. “In Anthropic company lore, this saga is put on a pedestal as an example of choosing principle over profit. Yet it may have also been a strategic wake-up call for Mr. Amodei, who called the sacrifice a ‘one-off.’ Increasingly, Mr. Amodei seemed to see how being the leader in the A.I. race conferred compounding benefits in terms of revenue, recruiting, fund-raising and developing the next-generation A.I. models. Once commercial A.I. products had been introduced, Mr. Amodei seemed unwilling to settle for second place.”
Anthropic has, of course, branded itself as the lab worried about safety, the lab that’s—at least culturally—above profit motive (though OpenAI used to dabble in nonprofitism). Dario Amodei has promoted a pause for the safety purposes, but it’s not really clear what that means or how real that is.
Here’s Matt Levine’s take on all this at Bloomberg, as of a few days ago:
The leaders of the big frontier artificial intelligence labs sincerely believe that there is a high probability that their products might kill everyone on earth, and that being slower and more deliberate about product development would meaningfully reduce that probability. However, each lab finds itself unable to unilaterally slow its product development, because if it slows down and its competitors do not, (1) the competitors will out-compete it for customers, researchers, investors, etc. and (2) the competitors will kill everyone on earth anyway. The leaders of the big frontier AI labs would be willing to get together in a room and agree to collectively slow down development, and they think that is the only practical way to reduce the probability of killing everyone. However, when they muse about doing that, their antitrust lawyers jump up and say “NO! You can’t get together in a room with your leading competitors and agree to slow down product rollouts! That is a conspiracy in restraint of trade, and you personally could go to prison for even suggesting it!” The leaders of the big frontier AI labs do not want to go to prison. They don’t want to kill everyone on earth either, but the prison stuff is so much more salient. So they don’t get together in a room to slow development, and instead post essays online saying “hey it would be great if the government made us slow down, or at least waived antitrust law so that we could get together to agree to slow down.”
Levine points out that, unless we really don’t know what the frontier labs have agreed to, there’s some strange doublespeak happening from Amodei and co.: “It just feels like if this is the most important problem facing humanity, then what you want to do is hire many very very good antitrust lawyers and pay them a lot of money to work on it full-time.”
Only time will reveal whether Amodei’s preference for slowdown or preference for trouncing OpenAI is more dominant. Now that the company is public, there are many other pressures with which he must contend.
Scenes from New York: School crackdowns on cell phones may be helping children and teenagers improve test scores. “Schools with strict bans—requiring students to lock their phones in pouches all school day—saw math and reading scores improve as if students had an additional week or two of school a year,” wrote The New York Times of a new study, published earlier this week by the Center for Education Policy Research at Harvard. The study assessed 3.3 million students total, all third through tenth graders.
“Two earlier studies reached mixed conclusions,” notes the Times. “One, of a single district in Florida in 2025, found small increases in test scores by the second year of the ban. Another, using schools’ average scores on a variety of standardized tests nationwide, found no statistically significant increase in scores.”
Though my son is not and probably will not be enrolled in New York City public schools due to, uh, ideological differences (I help run a small homeschooling co-op instead), I’m glad schools here have decently appropriate tech policies. Children enrolled in 2-K (essentially, daycare/preschool for 2-year-olds, run by the city) through second grade are not allowed to use screens; children in third through fifth grade are not supposed to have more than 30 minutes daily of screen time; and children in middle school are not to have more than 45 minutes daily.
There was no need to wait for (mostly fake) social science “studies” to sanction getting the phones out of schools. Our collective experience of what screens do to our ability to sustain attention should have been more than sufficient to arrive at this conclusion years ago.
— Tyler Austin Harper (@Tyler_A_Harper) November 18, 2026
Leave a comment