Can IPO Disclosures Test an AI Growth Narrative?

The accusation is testable only in pieces

Michael Burry argued on September 14 that calls by major AI companies to slow frontier development were self-serving. Among his explanations, he suggested that describing AI as unusually powerful could strengthen an IPO narrative and that talk of slowing development could provide cover for weaker growth. Those are Burry's interpretations, not established facts about either company. (Substack)

There is, however, a factual capital-markets backdrop to the argument. Sam Altman said OpenAI would not go public in 2026 and cited the need to focus on AI safety and alignment. Anthropic had already confidentially filed for a U.S. IPO in June, according to Reuters, and its listing preparations were still progressing when Altman made those comments. (Reuters)

For an options researcher, the useful question is therefore narrower than deciding whose explanation is right: what evidence could eventually distinguish a weakening business from a company deliberately changing its pace for another reason?

An IPO filing can answer parts of that question. It cannot reveal an executive's private motive.

A public filing creates a financial baseline

The SEC says an IPO registration statement, typically filed on Form S-1, contains a prospectus describing the company, the offering and information about its business and financial condition. Among the sections the SEC highlights are risk factors, use of proceeds, financial data, management's discussion and analysis, the business description and audited financial statements. (SEC)

That makes the eventual public filing more useful for investigating the growth part of Burry's claim than the rhetoric surrounding an IPO.

A researcher could compare reported revenue trends across periods, examine management's explanation for changes in operating performance, identify major customer or supplier dependencies, review spending and financing requirements, and see how the company describes the risks to future growth. Those disclosures provide evidence that can be compared with earlier public statements.

They still would not establish why executives made a particular safety argument. Strong financial growth would not disprove genuine safety concerns, and weaker growth would not establish that those concerns were manufactured for financial reasons. The filing can test measurable parts of a narrative without resolving an allegation about intent.

There is another limitation in Anthropic's case: Reuters reported that its June submission was confidential. Confidential IPO preparation does not give the public the same financial record that a publicly available registration statement eventually can provide. (Reuters)

Options on other companies are not a substitute

Before a private company has publicly traded shares and exchange-listed options tied to those shares, there is no direct equity-options market for researchers to inspect.

OCC describes a standard equity option as a contract covering shares of an underlying security. That relationship matters when a private-company story spills into options activity in chipmakers, cloud providers, software companies or broad technology funds: those contracts have different underlying securities. (OCC)

Activity in such instruments may show that market participants are repricing risks around the wider AI theme. It cannot isolate OpenAI's or Anthropic's revenue trajectory, determine what either company would disclose in an IPO filing, or establish why an executive advocated a slower development pace.

This distinction becomes especially important during a large thematic event. A semiconductor company can be affected by expectations for AI infrastructure spending while simultaneously responding to interest rates, its own earnings outlook, product news and other factors. Treating its options activity as evidence about the motives of a private AI company collapses several separate questions into one.

The filing comes before the options interpretation

If one of these companies eventually becomes public, the research sequence becomes clearer.

Start with what the registration statement establishes about the underlying business: historical financial performance, operating explanations, financing needs, stated risks and the planned use of new capital. Compare those disclosures with earlier claims only where the documents provide comparable evidence.

If exchange-listed options later become available, their prices and activity add another layer of market information around that now-public underlying security. They do not replace the underlying disclosures or retroactively prove why management made an earlier statement.

That is the useful boundary exposed by Burry's accusation. The claim about slowing growth may eventually become more testable as financial disclosures emerge. The claim about motive remains a different proposition. Options research is more informative when those two questions stay separate.