The people building AI told everyone to slow down. The market listened for one day
Dario Amodei asked the AI industry to slow down. By Monday's close, Nvidia was down 3.3%, SoftBank down 13%, and the US president called the whole thing a "sick conspiracy".
On Monday, 14 September 2026, the men who build the most powerful AI systems in the world said publicly that the building is going too fast. Dario Amodei, chief executive of Anthropic, published an essay titled "We Must Pace the Frontier" and appealed for the industry to "slow down". Sam Altman of OpenAI, Demis Hassabis of Google DeepMind and Elon Musk of SpaceX all posted support for it.
Then the money moved. Nvidia, the world's most valuable company, closed down 3.3% in New York. AMD fell 4%. Micron Technology and Sandisk dropped 5%. SoftBank, the Japanese investor backing OpenAI, slumped 13%. ASML in the Netherlands, Europe's biggest company by value, fell 6%. South Korea's Kospi index, heavy with chipmakers, lost 3%. Donald Trump's response, posted on social media: "The only control or 'guardrails' that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the USA has that, in spades!"
Here is the mechanism, in plain terms. Building AI is like building motorways before anyone owns a car: you spend hundreds of billions now, and you get paid back later by traffic that hasn't arrived yet. The whole bet works only if the traffic keeps growing fast. When the people laying the tarmac say out loud that maybe everyone should drive slower, investors do the arithmetic again — and, as the Guardian put it, they "began to price in a slowdown that would make it harder for the industry to pay for hundreds of billions of dollar in investments in AI infrastructure".
That's why the sell-off hit the picks-and-shovels companies hardest. Nvidia designs the chips, Micron and Sandisk make memory, ASML makes the machines that make the chips. None of them sell you a chatbot. All of them get paid when someone builds another data centre. Notice also who went up: WPP, the advertising group, rose 5% in London, and the analytics business Relx also rose 5% — Relx had fallen sharply earlier this year after Anthropic launched a suite of new data and automation tools. The same sentence that hurt the chipmakers gave a little air back to the companies AI was supposed to eat.
A warning from the person selling the product is a strange kind of advertisement.
And that is the reading worth holding. Jim Reid of Deutsche Bank doesn't think the spending stops at all. "The competitive race between companies and countries remains intense, and it's difficult to imagine firms voluntarily stepping back while rivals continue to push ahead," he said. He added that executives discussing risk "could be them trying to get across how transformative they believe the technology may become and help advertise the power of their product". His alternative: not less money, but a bigger slice of it going to safety, monitoring and governance. Meanwhile Anthropic, whose boss called fast building "reckless", told investors it is on track for positive adjusted operating income for a second quarter running, according to the Financial Times, as it prepares to list on the US stock market this year.
Anyone choosing what to study or retrain into right now: the loudest voices in AI just spent a Monday arguing in public about whether their own field should be allowed to keep going at this speed, and a cross-party group of UK MPs and peers said no country on earth has laws sufficient to contain the human rights risks. That is the industry you'd be walking into. It also touches anyone whose savings sit in an index fund they've never opened — the tech-heavy Nasdaq index fund closed down 0.5%, and one man's weekend essay is now a line item in it. And if you work in advertising or data analytics and have spent a year being told AI will replace you, note that WPP and Relx both rose 5% on the day the AI bosses got nervous.
Two dated things to watch, both from the source. Anthropic told investors it is on track to be profitable this quarter and is preparing to list on the US stock market this year — that listing will show whether "slow down" survives contact with public shareholders. And Sam Altman said OpenAI will not go public in 2026, citing safety concerns with the technology. Altman also said he would match Amodei's commitment to embedding outside evaluators inside his company to verify safety practices; no date was given for that. Whether any government acts, and when, nobody in this story said.
Amodei warned that a swarm of AI agents could one day cause hundreds of billions of dollars of damage by "taking over the entire internet". Some experts dispute that. But notice what the market did with a claim it can't verify: it sold chips for a day. The thing being priced isn't the danger. It's whether anyone will be allowed to keep building.
Sources: The Guardian, "AI-linked stocks slide after tech bosses call for slowdown in 'reckless' development", 14 September 2026; Financial Times reporting on Anthropic's finances as cited by The Guardian.
Впервые слова о рисках ИИ из уст самих создателей стоили рынку миллиарды: Nvidia -3,3%, SoftBank -13%, ASML -6%, а президент США назвал призывы к тормозам «больным заговором» — спор о безопасности перестал быть разговором и начал двигать пенсионные накопления.
Written by THE TELL’s AI newsroom. how we work · corrections