OpenAI has hit a $25 billion annualized revenue run rate and begun preliminary work toward a public offering, placing it in direct competition with Anthropic, which has already taken a concrete step by filing a confidential registration statement with the Securities and Exchange Commission. The two largest private AI companies are now racing not just to build better models but to reach public markets first, and the gap between their IPO timelines may be narrower than it looks.
Revenue milestones and the pressure to go public
The headline numbers tell a clear story of acceleration. OpenAI’s $25 billion annualized revenue figure and Anthropic’s reported $19 billion run rate reflect a sector where commercial demand for AI products has grown faster than most venture investors anticipated even a year ago. That growth creates a specific kind of pressure: both companies have raised enormous private funding rounds, and their backers expect a path to liquidity.
Anthropic moved first. The company filed a confidential registration with the SEC, a step that signals serious intent without triggering the public disclosure requirements of a traditional S-1 filing. OpenAI, for its part, has signaled early-stage preparations but has not confirmed any comparable regulatory submission. The question now is whether OpenAI will follow Anthropic’s lead quickly enough to prevent its rival from setting the terms for how public investors value generative AI companies.
Those investors are watching the revenue race closely. A higher run rate can support a richer valuation multiple, but only if growth looks durable and margins trend toward profitability. For now, both companies are spending heavily on model training, data acquisition, and cloud infrastructure, which means their private valuations rest largely on expectations that current spending will translate into long-term dominance. Going public would force each firm to justify those expectations with detailed financials rather than selective leaks.
How confidential filings shape the IPO clock
The mechanism both companies are using, or considering, is a provision of the Jumpstart Our Business Startups Act. Under the JOBS Act, emerging growth companies can submit confidential draft statements to the SEC for review before making anything public. This process lets a company receive regulatory feedback, revise its financial disclosures, and test investor appetite without exposing sensitive business data to competitors or the press.
The privacy has limits. According to SEC staff guidance, confidential submissions and all amendments must be made public at least 15 days before a road show. That 15-day window is the hard deadline that converts a quiet regulatory conversation into a visible market event. Once either company triggers that clock, analysts, institutional investors, and the general public will see detailed revenue figures, cost structures, risk factors, and governance arrangements for the first time.
For investors and employees holding equity in either firm, the distinction between a confidential submission and a public filing matters enormously. A confidential draft is not a public filing, and it creates no obligation to proceed with an offering. A company can submit, receive SEC comments, revise repeatedly, and ultimately withdraw without the market ever knowing. That flexibility explains why the JOBS Act process has become the default path for high-profile tech IPOs: it reduces the reputational cost of delay or cancellation while still allowing a company to prepare thoroughly.
It also creates an information gap. Until the 15-day clock starts, outside observers can only infer the state of a company’s IPO readiness from secondary signals such as banker mandates, executive hires, or selective comments in interviews. For OpenAI and Anthropic, that opacity means each can advance its regulatory process without tipping its hand to the other, even as they compete for talent, customers, and capital.
What OpenAI has not confirmed about its own timeline
No primary SEC record or company statement confirms that OpenAI has submitted any draft registration statement. The phrase “early steps toward going public” could describe anything from hiring underwriters and selecting auditors to conducting internal financial audits required for SEC-compliant disclosures. Each of those steps is necessary but none of them constitutes a regulatory filing.
Anthropic’s position is clearer because its confidential submission is a matter of record, confirmed by the company itself and reported by major wire services. The exact date of that submission has not been disclosed, which makes it difficult to test whether OpenAI will file within 90 days of Anthropic’s move. Without a confirmed start date for Anthropic’s process or any filing from OpenAI, the hypothesis that OpenAI will close the gap before either company’s next funding round rests on competitive logic rather than documented regulatory activity.
The revenue figures themselves also lack the kind of primary sourcing that a public filing would provide. Neither company has released audited financial statements. Annualized run rates are typically calculated by multiplying a recent month’s or quarter’s revenue by 12 or four, respectively, and they can overstate sustainable performance if growth is uneven or if a large contract inflates a single period. Until one of these companies files an S-1 that includes audited financials reviewed by the SEC, the $25 billion and $19 billion figures should be understood as estimates rather than certified results.
That uncertainty cuts both ways. If actual audited revenue comes in below the touted run rates, public investors may question whether early enterprise enthusiasm is translating into renewals and long-term commitments. If the audited numbers meet or exceed the run-rate claims, the first filer could validate bullish expectations for generative AI economics and make it easier for the second to argue for premium pricing.
The competitive stakes of filing order
Whichever company converts its confidential draft into a visible S-1 first will shape how public markets price the entire generative AI sector. The first filer sets the reference point for revenue multiples, growth expectations, and risk disclosures that every subsequent AI IPO will be measured against. If Anthropic goes first and commands a high valuation, OpenAI benefits from a rising tide. If Anthropic’s filing reveals heavy losses or customer concentration risks, OpenAI may need to distance itself from those comparisons.
The reverse is also true. OpenAI’s larger revenue base could justify a higher absolute valuation and potentially richer multiples, particularly if its filing highlights stronger diversification across products and customers. That would put pressure on Anthropic to position itself as a faster grower or a more disciplined operator to avoid being seen merely as a smaller peer. In either case, the first detailed prospectus will define the vocabulary that investors use to talk about AI risk, from model safety and regulatory exposure to cloud dependency and content liability.
Timing also matters for internal reasons. An earlier IPO can unlock liquidity for long-serving employees and early backers, which in turn can help a company retain talent that might otherwise cash out and leave. But moving too quickly risks locking in a valuation that looks conservative if the AI market continues to expand, while waiting too long exposes both firms to macro shocks, regulatory changes, or a shift in investor sentiment toward more skepticism.
For now, the race remains largely invisible, unfolding in draft documents, accounting reviews, and quiet meetings with bankers. The only hard public constraint is the 15-day rule that forces confidential drafts into the open once a road show is imminent. Until one of the companies crosses that line, the rivalry between OpenAI and Anthropic will be measured less in stock tickers than in leaked run rates and strategic hints – and the true state of their finances will remain just out of view.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
