Michael Burry Warns of OpenAI, Anthropic IPO Risk

Michael Burry, the investor made famous by his successful bet against the US housing market before the 2008 financial crisis, has issued his strongest warning yet about the artificial intelligence investment boom. Burry says markets should fall sharply enough to prevent OpenAI and Anthropic from going public, arguing that the two AI companies could ultimately absorb and destroy trillions of dollars of investor capital.

Burry’s comments come as both companies move toward potential public listings and as investors continue to pour enormous amounts of money into AI infrastructure, computing capacity and model development. His warning is not a forecast that OpenAI or Anthropic will literally lose trillions of dollars themselves; rather, he is arguing that excessive valuations and capital spending could eventually transfer enormous losses to public-market investors if the expected economic returns fail to materialise.

Key takeaways

• Michael Burry says markets should “tank hard” to prevent OpenAI and Anthropic from going public.
• He argues the companies could “destroy TRILLIONS of dollars of capital.”
• OpenAI has confidentially filed for a US IPO but has not set a timetable.
• Anthropic has also confidentially filed and is reportedly considering a listing after the November US midterm elections.
• Anthropic’s reported IPO materials show enormous future commitments to cloud, computing and infrastructure.
• Burry’s broader thesis is that AI infrastructure spending could become unsustainable if revenue growth fails to justify the investment.
• Burry has also disclosed bearish positions against several AI-linked companies, meaning his views are accompanied by financial exposure to a decline.
• His warning is an investor opinion, not evidence that an AI crash or IPO failure is inevitable.

What Michael Burry is warning about

Burry’s latest comments are unusually direct.

In a post on X on September 29, he said that, “For the benefit of humanity,” markets should fall sharply enough to prevent OpenAI and Anthropic from going public. In a follow-up comment, he said the companies could absorb and then destroy trillions of dollars of capital, adding that capital destruction would be the least serious damage they could cause.

The statement combines two separate concerns.

The first is financial: Burry believes investors could allocate extraordinary amounts of money to AI companies at valuations that may ultimately prove difficult to justify.

The second is technological and social: he has repeatedly argued that the risks associated with rapidly scaling AI could extend beyond financial losses.

For investors, however, the financial argument is the most immediately measurable.

Why the OpenAI and Anthropic IPOs matter

OpenAI and Anthropic are not ordinary technology companies preparing to list on a stock exchange.

Both are among the largest private AI companies in the world and require enormous amounts of computing infrastructure to train and operate increasingly powerful models.

An IPO would change who ultimately finances that spending.

Private investors, strategic partners and venture funds currently carry much of the risk associated with the companies. Once publicly listed, retail investors, pension funds, mutual funds and public-market institutions could gain direct exposure.

That is precisely what concerns Burry.

If public investors assign extremely high valuations to the companies and then continue funding aggressive expansion, a subsequent slowdown in AI demand could create losses across a much wider investor base.

OpenAI has already moved toward the public markets

OpenAI confidentially filed for a US IPO in June 2026, according to Reuters.

The company did not disclose the size or terms of the offering and said it had not established a timetable. Reuters had previously reported that OpenAI could seek a valuation of up to $1 trillion in a future listing.

OpenAI has since indicated that an immediate listing is not its priority.

CEO Sam Altman said in September that going public in 2026 would be “ill-advised,” according to reporting cited by Yahoo Finance. The company has pushed its IPO expectations into 2027 and continues to focus on expanding its technology and business.

That means Burry’s warning is aimed at a potential future event rather than an IPO that is already scheduled.

Anthropic may reach the public market sooner

Anthropic appears to be further along.

The company confidentially filed for a US IPO in June after a major private funding round that valued it at approximately $965 billion, according to Reuters.

More recent reporting indicates that Anthropic could pursue a public listing after the November US midterm elections.

That potential timing makes Anthropic particularly important to the current AI investment cycle.

If Anthropic becomes public before OpenAI, investors would receive a market-based valuation for one of the industry’s most important frontier AI developers.

The resulting share price could influence how investors value other AI companies, including cloud providers, chipmakers, data-centre operators and software companies.

Anthropic’s capital requirements explain Burry’s concern

The economics of frontier AI help explain why Burry is focusing on capital destruction rather than simply saying that AI companies are overvalued.

Training and operating advanced AI systems requires enormous amounts of computing infrastructure.

Anthropic’s IPO prospectus, reviewed by Reuters, reportedly identifies roughly $518 billion in future cloud, computing and infrastructure commitments. The company also reported a $42 billion net loss in 2025, according to Reuters’ reporting on the filing.

Those figures do not mean Anthropic is necessarily financially unsustainable.

A rapidly growing technology company can deliberately spend ahead of revenue if the investment creates a sufficiently large future market.

The issue is whether the eventual revenue and cash generation will be large enough to justify the infrastructure spending and valuations investors are currently assigning to frontier AI companies.

That is the fundamental question behind Burry’s bearish thesis.

AI spending is becoming the bigger investment story

The debate is increasingly moving beyond individual AI companies.

The entire technology ecosystem is spending heavily on AI.

Nvidia is supplying GPUs. Cloud providers are building data centres. Semiconductor companies are expanding advanced packaging and memory capacity. Utilities and infrastructure companies are preparing for higher electricity demand.

AI labs such as OpenAI and Anthropic sit at the centre of this spending chain.

If demand for AI computing continues to grow rapidly, the investment could generate substantial economic returns.

But if AI revenues fail to keep pace with infrastructure investment, the consequences could spread throughout the ecosystem.

That is why Burry sees the potential problem as a capital-allocation cycle rather than simply an issue with one company’s valuation.

Burry’s bearish thesis has moved closer to the present

Burry has been warning about an AI bubble for months.

In a recent Substack post, he said he was increasingly confident that his bearish thesis would play out within the next year and brought forward his previous base-case timing for an AI reversal.

He has argued that companies are spending heavily on chips and data centres while accounting assumptions and depreciation schedules could make profits look stronger than the underlying economics suggest.

He has also criticised the enormous capital expenditures being made by technology companies.

His central question is simple: What happens if AI investment grows faster than the profits generated by AI?

If companies continue buying computing infrastructure because they expect future AI demand, but that demand eventually disappoints, excess capacity could lead to falling returns on invested capital.

That could pressure both private AI valuations and public companies that supply the infrastructure.

Burry is also positioned for an AI downturn

An important qualification is that Burry is not a neutral observer.

Reporting on his latest views says he holds put options or other bearish positions against several AI-linked companies.

The list has included Nvidia, Palantir, Micron, Oracle and the Nasdaq 100.

A put option generally becomes more valuable when the underlying asset declines, although the exact economics depend on the strike price, expiry and other terms.

That means Burry has a financial incentive to be correct about an AI-driven market decline.

It does not make his argument wrong.

But investors should distinguish between an analytical warning and an independently established prediction. His position is a thesis, not proof that AI valuations will collapse.

Why the “trillions” claim needs context

Burry’s use of “trillions” can sound like a prediction that OpenAI and Anthropic themselves will lose several trillion dollars.

That is not necessarily what he means.

His warning is about capital flowing into the AI ecosystem and subsequently being destroyed if investments fail to produce adequate returns.

Consider the structure.

Suppose investors value an AI company at $1 trillion and the company subsequently raises additional capital at high valuations. If future growth disappoints and the valuation falls dramatically, shareholders can lose hundreds of billions of dollars in market value without the company itself having lost an equivalent amount of cash.

The same process could occur across multiple AI companies and infrastructure providers.

If enough companies are affected simultaneously, aggregate shareholder losses could reach trillions.

That is the mechanism behind Burry’s warning.

The IPO could become a test of AI investor appetite

An OpenAI or Anthropic listing would provide one of the clearest tests yet of how public investors value frontier AI.

Private-market valuations can remain elevated because funding rounds involve a limited number of investors and negotiated terms.

Public markets are different.

Once shares trade every day, millions of investors can buy or sell them based on earnings, cash flow, growth expectations and sentiment.

The valuation can therefore change much faster.

A successful IPO at a very high valuation would reinforce the bullish AI narrative.

A weak listing or subsequent sell-off could have the opposite effect.

It could make investors question valuations across the AI supply chain.

The bullish case Burry is challenging

There is also a credible argument on the other side.

AI adoption is expanding rapidly across coding, customer service, enterprise software, research, advertising and other industries.

OpenAI says its revenue has been growing rapidly, while Anthropic has also reported sharp increases in demand for Claude.

Reuters reported that Anthropic’s revenue had grown twelvefold to nearly $4.6 billion in 2025. OpenAI’s annual recurring revenue was reported to be approaching $70 billion in September, according to Axios data cited by Yahoo Finance.

If those growth rates continue, today’s enormous infrastructure investments could eventually look reasonable.

Technology markets have repeatedly demonstrated that investors can underestimate the size of new markets.

The internet is an obvious example. Many dot-com valuations collapsed, but the underlying technology ultimately produced some of the world’s largest companies.

The same distinction matters for AI.

A bubble can exist around a genuine technological revolution.

Why the dot-com comparison matters

Burry’s argument is not necessarily that AI is useless.

It is closer to the argument that a transformative technology can still produce terrible investments when expectations become detached from economics.

During the late-1990s internet boom, companies with little revenue and weak business models attracted extraordinary valuations.

When expectations changed, investors suffered huge losses.

Yet the internet itself became one of the most economically important technologies in history.

The AI cycle could follow a similar pattern in which the technology succeeds while many individual investments fail.

That is the risk public investors need to understand.

What could prove Burry wrong?

Several developments could undermine his thesis.

The most obvious would be continued rapid growth in AI revenue.

If enterprises keep increasing spending on AI applications and AI agents, companies such as OpenAI and Anthropic could generate enough revenue to support their infrastructure investments.

Another possibility is falling computing costs.

If chips become more powerful and efficient while inference costs decline, AI companies could achieve significantly better economics without proportionally increasing capital expenditure.

A third possibility is that AI creates entirely new markets that are difficult to measure today.

In that scenario, current valuations could look less extreme several years from now.

What could make Burry’s warning look prescient?

The opposite scenario would involve a slowdown in AI spending.

If companies discover that AI projects do not generate sufficient returns, enterprise demand could weaken.

Cloud providers could slow data-centre construction. AI labs could face higher financing costs. GPU demand could normalise.

At the same time, increasingly sophisticated models could become cheaper to operate, making it harder for AI companies to maintain premium pricing.

That combination—slower revenue growth, high infrastructure costs and falling prices—would create a particularly difficult environment for highly valued AI companies.

The bigger risk may be the entire capital cycle

The most important part of Burry’s warning is therefore not whether OpenAI or Anthropic deserve a specific valuation.

It is whether the entire AI ecosystem is collectively investing more capital than it can economically recover.

The current AI boom connects model developers, chipmakers, cloud companies, data-centre operators and energy infrastructure.

Each group depends partly on continued spending by the others.

That creates a powerful growth cycle when demand is rising.

But it can also create a feedback loop in the opposite direction if expectations weaken.

A slowdown in AI demand could reduce infrastructure spending, which would affect semiconductor demand, cloud revenue and data-centre utilisation.

That is the systemic risk Burry is highlighting.

The Bigger Picture

Michael Burry’s warning should not be read as proof that OpenAI and Anthropic are doomed or that an AI crash is inevitable. His comments are an intentionally extreme expression of a broader concern: the amount of capital flowing into AI must eventually be supported by sustainable revenue and returns.

The upcoming IPOs could provide the market with its clearest test yet. If investors accept trillion-dollar valuations while companies continue to generate rapid revenue growth, the AI boom could enter another phase of expansion. If public investors demand evidence of profits and cash flow, valuations could become much more disciplined.

The technology can succeed even if some AI investments fail. That distinction will be crucial if the sector eventually moves from an infrastructure-building phase to a profitability-driven phase.

Looking Ahead

The next major milestones will be Anthropic’s potential IPO process and OpenAI’s progress toward a public listing. Their disclosures will give investors more information about revenue growth, losses, infrastructure commitments and the amount of capital required to remain competitive at the frontier.

For now, Burry’s “trillions” warning is best viewed as a stress test for the AI investment thesis rather than a precise forecast. The key question for public markets will be whether the extraordinary spending required to build frontier AI can ultimately produce equally extraordinary economic returns.