Burry Calls Slowdown Talk a Gift to OpenAI and Anthropic

Michael Burry, the investor portrayed in "The Big Short," said a safety-driven slowdown in artificial intelligence development would be "self-serving" for OpenAI and Anthropic, after a researcher affiliated with both companies publicly warned that AI could lead to human extinction.

Michael Burry, the investor whose bet against subprime mortgages was dramatised in "The Big Short," has put a commercial reading on the AI industry's safety debate. According to Nasdaq Markets, Burry said that a slowdown in artificial intelligence development would be "self-serving" for OpenAI and Anthropic — the two labs whose leadership and researchers have been most vocal about the technology's dangers.

The comment follows a public warning from a researcher affiliated with Anthropic and OpenAI that artificial intelligence could lead to human extinction. Since that warning, AI companies have been openly discussing whether to slow the pace of frontier model development on safety grounds. Burry's contribution is not a rebuttal of the risk. It is a claim about who benefits if the industry acts on it.

The argument: a pause protects whoever is already in front

Burry's logic is structural rather than technical. In any race where a handful of firms hold the lead, agreeing to slow down freezes the standings. The labs with the most capable models, the largest compute contracts and the deepest enterprise distribution keep those advantages, while challengers lose the one thing they need most — time spent closing the gap at full speed.

That is why "self-serving" is a pointed word choice. It does not accuse anyone of insincerity about extinction risk. It says the recommended remedy happens to align with the commercial interest of the firms recommending it. If a slowdown is voluntary, the incumbents choose their own pace. If it becomes regulation, compliance costs fall hardest on the smallest entrants, and the labs already staffed with safety researchers, red teams and policy departments absorb them most easily.

There is a second layer. Warnings about existential capability are also marketing claims about capability. A model dangerous enough to warrant a global pause is, by implication, a model worth paying for. Safety messaging and product messaging are hard to separate when the same company issues both.

Why this lands differently coming from Burry

Burry's reputation rests on a single trait: a willingness to interrogate the incentives of the people telling the market what to believe. Ahead of the housing crisis, that meant asking why rating agencies were paid by the issuers whose paper they graded. Applied to artificial intelligence, the same instinct asks why the most alarming statements about AI keep arriving from the firms with the most to gain from being treated as the technology's responsible custodians.

He has been publicly sceptical of AI-linked equity valuations, and this framing fits that stance. If the case for owning the AI complex rests partly on the assumption that the leaders are uncatchable, then a self-enforced slowdown is the mechanism that keeps the assumption true. An investor who doubts the durability of those moats will read pause talk as evidence about market structure, not about machine intelligence.

Neither OpenAI nor Anthropic is publicly traded, so there is no share price against which to test the claim directly. Both are private, and both depend on outside capital and compute partnerships. Investors express views on the theme through listed suppliers, cloud providers and hardware makers instead — which is precisely why a debate between private labs moves public markets.

The tape on the day the comment landed

The broad market gave no unified verdict. At the close on Wednesday, 16 September 2026, the S&P 500 tracker SPDR S&P 500 ETF Trust (NYSEARCA: SPY) finished at $754.05, down 0.44% from the prior close of $757.39, having traded between $749.60 and $761.67. The Invesco QQQ Trust (NASDAQ: QQQ), which follows the Nasdaq 100 and carries the heaviest weighting in large-cap technology, closed at $704.72, up 0.03% against a previous close of $704.54, with a range of $700.00 to $711.88. The SPDR Dow Jones Industrial Average ETF Trust (NYSEARCA: DIA) was the weakest of the three, ending at $515.22, down 1.15% from $521.23.

That pattern — the tech-heavy benchmark essentially flat while the industrial-weighted index fell more than a percent — is the shape of a market still willing to hold AI exposure even as it sells other things. Nothing in a single session's prices confirms or refutes Burry's thesis. But it does show where the risk appetite sat: the AI trade absorbed a prominent bear's argument without visible damage.

What a slowdown would actually change for investors

The practical question is whether slowdown talk converts into anything measurable. Three channels matter.

• Capital spending. The clearest transmission from AI policy to listed equities runs through data centre and chip orders. A genuine deceleration in frontier training would show up first in capex guidance, not in press statements.
• Regulatory form. Voluntary restraint that incumbents design themselves has very different competitive consequences from binding rules written by governments. The first entrenches; the second is unpredictable for everyone.
• Entrant funding. If investors conclude that a pause is coming, the cost of capital for smaller labs rises and consolidation accelerates — the outcome Burry's argument predicts.

None of those has been triggered by rhetoric alone. For now the debate is conducted in essays, interviews and conference remarks, and the companies involved continue to ship models and sign compute agreements.

Things to watch from here

Watch whether any lab attaches a date or a threshold to a slowdown rather than a sentiment. Watch whether safety commitments appear in contracts with cloud partners, where they would carry financial weight. Watch whether Washington moves from encouragement toward enforceable standards, since that is the point at which the incumbency effect Burry describes would become measurable.

And watch the suppliers. Listed chipmakers, power providers and server assemblers are the instruments through which the market prices this argument, because the two companies at the centre of it cannot be bought on an exchange. If the extinction debate ever starts to bend order books, that is where it will appear first — long before it appears in a lab's mission statement.

Key facts

• Burry's claim: An AI slowdown would be "self-serving" for OpenAI and Anthropic
• Trigger: A researcher from Anthropic and OpenAI warned AI could cause human extinction
• Nasdaq 100 proxy: QQQ closed at $704.72, +0.03%, as of 16 Sep 2026 20:00 GMT
• Broad market: SPY $754.05 (-0.44%); DIA $515.22 (-1.15%) at the same close

Frequently asked questions

What exactly did Michael Burry say?

Burry, known for the subprime bet depicted in "The Big Short," said that a slowdown in artificial intelligence development would be "self-serving" for OpenAI and Anthropic. His point is that pausing frontier development would preserve the competitive position of the labs already in the lead, regardless of whether their stated safety concerns are sincere.

What prompted the AI slowdown discussion?

A researcher affiliated with both Anthropic and OpenAI publicly warned that artificial intelligence could lead to human extinction. Since that warning, AI companies have been openly debating whether to slow the pace of development on safety grounds, which is the discussion Burry is responding to with his comment about incentives.

Can investors buy shares in OpenAI or Anthropic?

No. Both companies are privately held, so there is no listed share price for either. Investors who want exposure to the theme generally do so indirectly, through publicly traded chipmakers, cloud providers, server makers and power suppliers whose order books depend on continued AI infrastructure spending.

How did the market close on the day of the comment?

At the close on 16 September 2026, the S&P 500 tracker SPY finished at $754.05, down 0.44%. The Nasdaq 100 proxy QQQ ended at $704.72, up 0.03%. The Dow tracker DIA was weakest at $515.22, down 1.15%. Technology exposure held up better than the industrial-weighted index that session.

Why would a safety pause help the biggest AI labs?

A pause freezes relative standings, so whichever labs hold the best models, largest compute contracts and deepest enterprise distribution keep those advantages. If the pause becomes regulation, compliance costs fall hardest on small entrants, while firms that already employ safety researchers and policy staff absorb them more easily.

What would signal that a slowdown is becoming real?

The clearest signals would be capital spending changes rather than statements: reduced data centre and chip order guidance, safety commitments written into cloud contracts where they carry financial consequences, or a shift in Washington from voluntary encouragement toward binding, enforceable standards on frontier model development.

Sources

• Michael Burry of "The Big Short" Says an AI Slowdown Would Be "Self-Serving" for OpenAI and Anthropic — Nasdaq Markets

Photo: Pachon in Motion · Pexels Licence — source