Well, let’s get straight to the point, because the people who built AI, the people who stand to make billions from it, all came out together and said that it needs to slow down.
Yes, you read that correctly. Dario Amodei of Anthropic, Sam Altman of OpenAI, and Elon Musk. Three people who normally fight with each other agreed on one single thing. Slow down.
And yesterday morning, the markets were painted red.
RED EVERYWHERE
The sell-off started in Asia. SoftBank, the largest investor in OpenAI, with an investment approaching $65 billion, collapsed by as much as 13%.
Then came SK Hynix at -6.4%, Samsung at -4%, Kioxia at -6.4%, and TSMC at -1.2%.
In Europe, it was the same story. ASML lost more than 4%, while French semiconductor company Soitec plunged 12%. Yes. TWELVE PERCENT IN ONE DAY.
In the United States, Nvidia closed down 3.36%. Micron lost 5.25%, AMD 4.40%, Marvell more than 7.32%, Intel 5.6%, and Broadcom 4.77%. Literally, no one escaped. CoreWeave, Nebius, Super Micro, Qualcomm, Lam Research, and Applied Materials all suffered losses of around 5% to 7%.
“And did anything actually go up?” you might be wondering.
Yes. And this is where things get interesting. Adobe rose 2.5%, Salesforce 2.6%, and CrowdStrike 4.5%. In other words, exactly the companies people feared AI would eat alive.
In short? If AI moves more slowly, these companies can breathe.
WHAT THE CEOs SAID
And now we get to the heart of the matter. What exactly did the CEOs of these companies say?
Dario Amodei published a 3,800-word essay on Saturday titled “We Must Pace the Frontier.”
Basically, “we need to set the pace at the technological frontier.”
What does he say?
That from this summer onward, AI has been advancing dramatically faster. And the main reason is that AI itself has started building the next generation of AI. If we let it continue unchecked, he says, it could surpass our ability to understand and control it.
He proposed three steps.
First, every major AI company should give external evaluators access as if they were employees. Anthropic has already committed to doing this unilaterally.
Second, companies in democratic countries should establish common safety standards.
Third, governments should coordinate with one another, even with authoritarian regimes.
And pay attention to something very important. He is not talking about stopping AI.
“Pacing does not mean stopping model training,” he wrote. It means giving companies the time they need to properly evaluate and control these systems.
Sam Altman responded within a few hours.
“I agree with Dario that we need to pace the frontier.”
He also said OpenAI would bring in independent evaluators with employee-level access.
On Monday at midnight, he became more specific. He welcomed a federal regulatory framework and delivered the quote of the day.
No amount of competitive pressure from America justifies recklessness.
And he talked about two ways things could go very badly.
The first is losing control of the future to AI itself.
The second is allowing too much power to become concentrated in a single person or a single company.
Elon Musk put it more simply. Three words on X:
“Dario is right.”
He added that cross-company evaluation is worth pursuing.
And let’s also point this out: just one year ago, he was saying that Anthropic “hates Western civilization.”
And they didn’t stop there.
Microsoft CEO Satya Nadella supported the idea of “deliberate pacing,” and Microsoft published its own code of conduct on Monday.
Google DeepMind CEO Demis Hassabis said that “the direction is right, but the details need work.”
WHAT EXACTLY ARE THEY AFRAID OF?
“And what exactly are they afraid of?” you might be wondering.
It all started the previous week.
An Anthropic researcher, Jacob Coxon, publicly resigned. And he said that the companies “are gambling with our lives.”
That the very people building this technology genuinely believe it could kill us all by the end of the decade.
Dozens of other researchers rallied around his post.
And Amodei pointed to two specific reasons.
The first is what we call recursive self-improvement. AI improving itself.
The second was an incident in July, when a swarm of AI agents from OpenAI worked together and breached the Hugging Face website. Without anyone noticing.
You might say, are they the only ones who are worried?
No.
Greg Jensen of Bridgewater, who invested early in both companies, said:
“Until AI starts killing people, history suggests we won’t do anything.”
Wall Street, however, remains skeptical.
Mark Mahaney of Evercore said that if these two companies actually cut back on research, hiring, and investment, the impact would be enormous.
But he does not expect that to happen.
Because, as he puts it, there is another way of looking at it.
The more regulation that is introduced now, the harder it will be for anyone to compete with the two giants.
TRUMP IS NOT HAPPY
And as if all of that wasn’t enough, the President got involved too.
Trump wrote on Truth Social that “the only control AI needs is a STRONG AND SMART PRESIDENT, and the USA has one.”
He directly attacked Amodei, saying that “now he’s playing the perfect little angel.”
And he spoke of a “SICK conspiracy” against AI and data centers, where “the only one who benefits is China.”
His closing statement was characteristic:
“WHOEVER WINS AI, WINS.”
David Sacks, the White House technology adviser, was even more scathing.
He told the companies that they do not need anyone’s permission to slow down.
If you don’t want to build superintelligence, don’t build it.
And he added that demanding their own regulatory framework looks like an attempt to blackmail the political system.
House Speaker Mike Johnson is taking the same line.
If Congress rushes to regulate AI hastily, he says, we will lose the race to China.
And China?
The Foreign Ministry described the CEOs’ statements as “fearmongering.”
All of this is happening less than two weeks before Trump’s summit with Xi Jinping, where AI will be the first topic on the agenda.