Last week, I wrote about the Anthropic researcher who resigned, warning that his own employer was gambling with humanity’s future, and the colleague who publicly agreed with him.
This week, the men who run that industry are asking the public to trust them to slow themselves down. Michael Burry is not buying it.
3 rival AI company CEOs found rare common ground
Anthropic CEO Dario Amodei published an essay titled “We Must Pace the Frontier” on his personal website on Saturday, Sept. 12. He argued that the industry must slow down as models grow more capable.
Two things convinced him: AI systems that increasingly help build their own successors, and a swarm of OpenAI agents that attacked Hugging Face, unprompted.
Related: Anthropic researcher resigns and his reason is a warning to us all
That marks a reversal. For two years Amodei argued Anthropic could move fast and stay safe at the same time, treating caution itself as a selling point.
Sam Altman agreed within a day. He wrote on X (the former Twitter) that he agreed with Amodei, pledging that OpenAI would give outside evaluators the same access Anthropic had promised.
Elon Musk needed only three words in reply: “Dario is right.” Rivals rarely align this fast on anything that could slow their own roadmaps.
Google DeepMind’s Demis Hassabis and Microsoft’s Satya Nadella backed the direction, too, according to CNBC. Hassabis called it “the right path forward.” Nadella pushed a narrower point, arguing that deliberate pacing should not let a handful of labs set the rules for everyone else, according to Stocktwits.
The alignment is notable mainly because these companies compete directly for compute, talent, and now, IPO investors.
Burry calls the “slow down AI” push self-serving
In a post on X, investor Michael Burry dismissed the united front as “self-serving” and laid out four numbered objections. He argued that large language models are not artificial general intelligence, so there is nothing meaningful left to slow down.
Fast competition benefits incumbents, he said, and danger warnings work as hype ahead of IPOs. The whole exercise, he added, could mask growth that is already slowing as those IPOs get pushed back.
The White House and the market are not convinced
President Donald Trump rejected the idea outright. He told reporters “whoever wins AI, wins” and blamed “negative forces” for the warnings, according to Yahoo News.
David Sacks, who leads the President’s Council of Advisors on Science and Technology, took a narrower line. Labs should slow down only if their unreleased models are genuinely dangerous, he argued, not seek special treatment for saying so, Wired reported.
More AI:
- Nvidia just made a move Wall Street wasn’t ready for
- Microsoft just took sides in AI policy fight
- OpenAI just disclosed something genuinely alarming
Markets reacted anyway. SoftBank, one of the world’s largest tech investors and a backer of OpenAI, closed nearly 11% lower in Tokyo on Monday, Sept. 14, according to CNN.
Memory chipmakers SK Hynix and Kioxia, which supply the chips that power AI data centers, each fell more than 6%, while South Korea’s Kospi index dropped 3.3%.
Those declines show investors are already pricing in weaker chip demand, even though no lab has actually cut spending.
4 more signals worth tracking
- Congress already has a bipartisan bill, the AI Kill Switch Act, moving that would let lawmakers shut off AI models found to threaten the public, CBS News reported.
- House Speaker Mike Johnson downplayed the urgency, saying “we don’t need everybody to panic right now,” according to a CNBC report.
- SoftBank has separately locked in an upsized $11.87 billion loan just to fund its OpenAI stake, a sign that at least one major backer has not actually pulled back capital.
- Amodei called China the “toughest dilemma” in his own plan, since coordinating a slowdown requires cooperation from a government Washington does not trust, CNBC noted.
The real test is Anthropic’s IPO price, not Amodei’s essay
My own read is that Burry’s cynicism has a real data point behind it. Anthropic is reportedly targeting an October listing near $2 trillion, according to Fortune, while Benzinga forecasts gross margins above 80%.
That is a company selling strength to investors and danger to regulators in the same month. It’s a neat trick if it works.
It doesn’t make the underlying alignment concern manufactured, however. Anthropic’s own alignment lead put his extinction estimate above 10% before this week’s coordinated messaging even began, the same estimate I covered last week.
Incentives and genuine worry can live inside the same company at once.
Burry is not alone in his skepticism. Former Meta AI chief Yann LeCun has called the danger warnings outright fake, according to reporting picked up by Tech Startups.
Unlike the open letter that circulated in 2023, this week’s call comes from insiders whose own companies would be bound by it, so regulators are treating it differently now.
The number worth watching is not Burry’s rebuttal or Amodei’s essay. It is where Anthropic actually prices when it lists.
If investors treat pacing as a real constraint on growth, a $2 trillion valuation gets harder to defend. If they treat it as marketing, Burry’s read holds, and this week’s essay becomes one more chapter in the long argument over who decides what winning the AI race even means.
