Burry Says AI IPOs Could Destroy Trillions in Capital
Michael Burry says public listings by OpenAI and Anthropic could wipe out trillions, citing a prospectus disclosing at least $518 billion of future compute and infrastructure obligations.
Investor Michael Burry warned that initial public offerings by OpenAI and Anthropic could destroy trillions of dollars in capital, pointing to an IPO prospectus that discloses at least $518 billion in future cloud, computing and infrastructure commitments.
Michael Burry, the investor best known for betting against subprime mortgages before the 2008 crisis, has aimed his skepticism at the next big thing in public markets: the eventual listings of the largest private artificial-intelligence labs. Burry warned that initial public offerings by OpenAI and Anthropic could destroy trillions of dollars in capital, according to GuruFocus.
The figure anchoring his argument is a disclosure buried in an IPO prospectus: at least $518 billion in future cloud, computing and infrastructure commitments. That is not a forecast or an analyst estimate. It is a contractual obligation schedule — money an AI company has already promised to pay data-center operators and cloud providers in years to come, whether or not the revenue arrives to cover it.
What a $518 billion commitment line actually means
Commitment disclosures are among the least glamorous pages in any prospectus and among the most revealing. They sit near the lease obligations and purchase commitments, and they tell a reader what the company must spend regardless of how the business performs. For a software firm, that line is usually small relative to revenue. For a frontier AI lab, it is the whole business model written down in advance: compute is the cost of goods sold, and it must be booked years ahead because chips, power and buildings cannot be conjured on demand.
The reason a figure of this scale unsettles a bear like Burry is straightforward. A commitment is a fixed claim on future cash. Revenue is not fixed. If model pricing falls — and inference costs have been falling across the industry as competition intensifies — the gap between what has been promised and what can be paid is absorbed by equity holders. In a private round, that gap is papered over with another financing. In a public market, it is repriced daily.
Why the risk travels beyond the two labs
The uncomfortable arithmetic for public investors is that the $518 billion does not vanish into a private black box. It is revenue on somebody else's income statement. Cloud providers, chip designers, data-center REITs, power developers and networking vendors have all built capacity plans and, in several cases, investor narratives around exactly these kinds of multi-year commitments.
That is the circularity critics keep pointing to: the same handful of balance sheets appear on both sides of many of these contracts, sometimes as supplier, sometimes as investor in the customer. When a buyer is partially funded by its own vendor, a commitment number looks less like independent demand and more like a closed loop. Burry's warning is essentially that the loop unwinds violently the moment public shareholders, rather than venture funds, are holding the equity.
Listed AI infrastructure names have been priced for the commitments being honored in full. Any repricing of that assumption would not be confined to the newly listed shares.
The market has not blinked yet
If there is anxiety about AI capital destruction, it was not visible in Wednesday's tape. As of the last trade at 15:16 GMT on 30 September 2026, the S&P 500 tracker (NYSEARCA: SPY) changed hands at $768.64, up 0.58% from the prior close of $764.20 and near the top of a $766.00–$769.41 day range. The Nasdaq 100 proxy (NASDAQ: QQQ) — the index most exposed to the AI complex — was the day's stronger performer at $743.60, up 0.77% against a $737.93 close, with an intraday range of $739.76 to $744.73. The Dow 30 fund (NYSEARCA: DIA) lagged at $513.43, up 0.11%.
The pattern is familiar and worth reading carefully: tech-heavy exposure outperforming the broad market, and the industrial-weighted index barely moving. Investors are still paying up for the AI build-out narrative on the same day a prominent bear is telling them the bill has already been signed. Burry's track record is built on being early and unpopular, which is precisely why the market's indifference proves nothing either way.
What an IPO filing would force into the open
The most consequential thing about a public listing by either lab has little to do with the first day of trading. It is disclosure. Going public means audited financials, quarterly revenue recognition, a segment breakdown of where compute spending goes, and a commitments table that analysts can compare with the figures the vendors report on the other side.
Today, most of what the market believes about frontier AI economics comes from selective disclosure and from what the suppliers choose to say about their order books. A prospectus compresses that into a single, legally consequential document. If the commitments genuinely outrun the revenue, the filing is where it becomes undeniable — and where the equity market, not a private syndicate, sets the price.
Things worth tracking from here
- Whether other AI companies disclose commitment schedules of comparable scale as they approach public markets.
- Whether vendors' reported backlog and remaining performance obligations reconcile with the customer-side commitment figures.
- The degree to which supplier financing, equity stakes and prepayments appear inside those contracts.
- Any sign of contract renegotiation, deferral or cancellation, which is usually the first crack in a commitment-driven capex cycle.
- Whether AI-exposed indexes continue to outperform the broad market once filings make the numbers comparable.
How to weigh a bear call of this size
"Trillions in capital" is a deliberately large number and should be treated as a directional claim rather than a modeled output. Burry has not published a valuation for either company, and no listing date has been set for either. What is documented is the $518 billion commitment line — a hard figure in a legal filing, not a projection.
For an ordinary investor, the practical exposure is rarely direct. Very few will get IPO allocations. Far more already own the infrastructure side through index funds: a broad Nasdaq 100 position is, by construction, a bet that AI capital spending converts into durable earnings. That is the position Burry is arguing against, and it is the one most portfolios hold without having chosen it deliberately. Reviewing how much of a portfolio sits in AI-linked capital expenditure is a more useful response to this warning than deciding whether to chase a listing that has not yet been announced.
Key facts
- Disclosed commitments: At least $518 billion in future cloud, computing and infrastructure obligations
- QQQ (Nasdaq 100): $743.60, +0.77%, as of 15:16 GMT, 30 Sep 2026
- SPY (S&P 500): $768.64, +0.58%, prev close $764.20
- DIA (Dow 30): $513.43, +0.11%, day range $512.27–$514.42
Frequently asked questions
What exactly did Michael Burry warn about?
Burry warned that initial public offerings by OpenAI and Anthropic could destroy trillions of dollars in capital. His argument centers on an IPO prospectus disclosing at least $518 billion in future cloud, computing and infrastructure commitments — contractual obligations that must be paid regardless of whether revenue materializes to cover them.
What is a commitment disclosure in an IPO prospectus?
It is a section listing money a company has legally agreed to spend in future years, such as multi-year cloud contracts, equipment purchases and leases. Unlike forecasts, these are binding obligations. For AI labs, compute commitments are effectively the cost of doing business, locked in years ahead of the revenue meant to pay for them.
How did markets trade on the day of the warning?
Markets rose. As of the last trade at 15:16 GMT on 30 September 2026, SPY was at $768.64, up 0.58%. QQQ, the Nasdaq 100 proxy most exposed to AI names, led at $743.60, up 0.77%. DIA trailed at $513.43, up 0.11%. There was no visible risk-off reaction.
Why would AI commitments affect listed companies?
Because those commitments are revenue for someone else. Cloud providers, chip makers, data-center operators and power suppliers have built capacity and investor expectations around multi-year AI contracts. If a customer cannot honor its obligations, the shortfall shows up in supplier backlogs and order books, not just in the AI company's own shares.
Have OpenAI or Anthropic announced IPO dates?
No listing date has been announced for either company based on the available information. Burry's warning addresses the prospect of such offerings rather than a scheduled transaction. The $518 billion commitment figure comes from an IPO prospectus disclosure, which is the concrete data point underpinning the broader argument.
What does 'circularity' mean in AI financing?
It describes arrangements where a supplier also invests in, or finances, its own customer. When a chip maker or cloud provider takes an equity stake in an AI lab that then commits to buying its products, the resulting revenue is partly self-funded. Critics argue this inflates apparent demand and obscures how much independent buying exists.
Sources
Photo: valaymtw · Pixabay Content License — source
