Anthropic IPO first day closes at $1.76 trillion as the Dow drops 1,100 points
The Anthropic IPO first day produced a $1.76 trillion company and a falling market. That was the small news: three giant school districts moved to drop Google Search, Ubisoft traded 166 jobs for an AI studio, and the world court named a genocide.

Chicago, 19 November 2035. Dear Priya — you asked which day I would put the pin in, and you expected a war or an election. It was a Thursday. I was a middle-school librarian then, and that morning I spent twenty minutes with a seventh grader who could not work out where her search results had gone. Her laptop had stopped showing her pages and started simply telling her things. That afternoon your mother's retirement savings fell for the sixth day in a row. I filed those as two unrelated annoyances. They were one story in two hats.
What follows in the middle is what actually happened that day, with sources. The letter around it is invented, and so is everything I claim it led to. Hindsight is a kind of fiction; I just happen to be fond of this one.
Anthropic IPO first day ended at a $1.76 trillion valuation while the Dow fell 1,100 points

Anthropic's shares opened on the Nasdaq a third above the price set the night before, brushed a two-trillion-dollar valuation before lunch, then spent the afternoon handing it back. The company behind Claude, the chatbot, finished its first day worth about $1.76 trillion, having raised more money in one offering than any company in history. Everybody agreed it was a triumph.
The rest of the market declined to attend the party. The Dow fell about 1,100 points and closed below 50,000, the S&P 500 had its sixth losing day in a row, and Nvidia threw away a healthy morning gain to end lower. One reason was almost too tidy to print: investors sold AI stocks to raise cash for the new AI stock. By Friday morning Tokyo was down four and a half percent.
Inside the building it was a lovely Thursday. The company has around 5,300 staff, and CNBC's arithmetic put the average employee stake above $40 million, with hundreds of people above $100 million and enough collective wealth, by one property firm's estimate, to buy nearly a tenth of San Francisco's homes. The chief executive told investors the company would "pace the training and deployment of AI systems, taking as long as is necessary to have confidence that they are safe." The pacing presumably begins after the closing bell.
From 2035, and this is memory rather than evidence: that was the week the money stopped asking whether the machines worked and started asking who would be left holding them. The market came back. Several of the people who bought at the opening bell did not.
New York, Los Angeles and Chicago moved to drop Google Search because its AI cannot be switched off
Google has told schools that from early next year its AI will be switched off by default on student accounts, with administrators deciding what to turn on. There is a catch. A district that blocks the AI loses ordinary Google Search too, because, as the company's education vice-president explained, "Search and A.I. are deeply integrated." The country's three largest school districts responded by shopping for another search engine.
Chicago says it will look elsewhere if no deal appears. Los Angeles has already turned off the AI summaries and its technology chief says the district could change search engines "literally overnight." Rivals already let administrators switch the generated answers off, which is the whole argument in one line: a search box that answers instead of listing links is a different object, and the schools would like the old one back.
New York's chancellor, Kamar H. Samuels, refused what he called "the false binary Google has laid out." Lina Khan, the former competition regulator, said the company was "leveraging its search monopoly to coerce schools." I can tell you from 2035 which children came out of that decade able to spot an invented fact. The ones who were still made to click through and look at the page. That is my profession talking, and my bias, and I stand by it.
Ubisoft cut 166 jobs in Quebec and opened a generative AI studio on the same day
Ubisoft confirmed on Thursday that it is cutting 166 jobs at its Montreal and Quebec City studios, mostly production and support staff on projects that were cancelled or rebuilt. The company had described January's cuts as the final ones. Its headcount had already fallen by roughly 1,100 between April and September.
The same earnings report announced a new subsidiary dedicated to "AI for gameplay innovation." It is called Motherbrain. Its team built the company's first playable generative AI experiment, and the new outfit will sell AI characters and game masters to other studios as well. Also in Thursday's news: one in three people in the games industry has been laid off in the past two years, and most of them say generative AI is making their business worse.
What I know from here, nine years on: the generated teammates did get good, the credits did get short, and some of the finest games of the early thirties were made by eight people in a rented room, a few of whom were cleaning out their desks that Thursday. Cold comfort counts as comfort if you wait long enough.
The world court found Myanmar committed genocide as AI-made videos stirred fresh attacks on Rohingya refugees

In The Hague, the International Court of Justice ruled that Myanmar committed genocide against the Rohingya. It is the first time in the court's eighty years that a state has been found responsible for the crime. The judges, nine to six, found that at least 10,000 people were killed and that genocidal intent was "the only reasonable inference" from the evidence. More than 730,000 people fled to Bangladesh. The judgment binds, cannot be appealed, and orders Myanmar to make amends.
In 2017 the hatred travelled on Facebook feeds. In 2026 it travels as video. While the court read its verdict, researchers were documenting AI-generated videos pushed by bot accounts before a petrol-bomb attack on a Rohingya settlement in Malaysia, and the UN refugee agency describing a coordinated campaign of AI-made images in Indonesia. The same account notes that Meta's removals of violent and inciting posts dropped from about 11.5 million a year to 1.4 million after it loosened its rules.
The court named a state, which took seven years and was still a kind of miracle. Nobody in that room had any authority over a stranger with a text box and a free video tool. That gap is what I would point to, if you asked me what we left unfixed for too long. We were very busy that day counting a different sort of record.
Two rooms where nobody wrote it down, and three things I expect next
Two other rooms spent that Thursday carefully not putting things in writing. In Antalya, negotiators published a COP31 draft with every mention of fossil fuels removed and the promised money for poor countries' defences dropped, keeping one firm ambition: electricity should be 35% of the world's energy by 2035, up from about a quarter now. The text does not say where the electricity comes from. "A ghost COP," said Mohamed Adow of Power Shift Africa. The richest industry on earth is currently bidding for that same electricity, and did not need to say a word.
In Shenzhen, the American, Chinese and Russian presidents held their first three-way meeting, about seventy minutes long, and issued a declaration on UN reform, oil, the international criminal court and the venerable line that a nuclear war must never be fought. Artificial intelligence appears nowhere in it. Afterwards Donald Trump said of his talk with Xi Jinping: "We spoke about SI, and we both agreed that we want... guardrails... on a minimal basis is a positive thing." He did not define minimal. The declaration did not mention it at all.
Three guesses, mine, and you can hold me to them. One of those three school districts leaves Google Search before the next school year starts, and the phrase deeply integrated gets quoted back at the company in a competition case within a year. Anthropic enters the big stock indexes, which means your retirement fund buys it whether or not you admire it. And the problem of invented video reaches a courtroom long before it reaches any treaty, because courts move and treaties, as Thursday showed twice, prefer not to.
P.S. One decent thing also happened that day: a large drug company agreed to design a cancer molecule with an AI lab spun out of Google, for a target it would not name. I have swallowed something every morning since 2033 that came out of an arrangement like that, and about this one subject I find it hard to be clever. The rest of that Thursday I am still extremely clever about. — N., who is imagined, as all letters from the future are.
Arthur Down is a journalist with more than 12 years of experience writing about technology, personal finance, health and everyday consumer topics. He studied Journalism and Media Studies at the University of Sheffield, started as a reporter at a regional newspaper and has worked freelance since 2014. At N33 he writes and edits the blog's practical guides to AI tools.
How we research, source and review articles