OpenAI confirmed on June 9 that it confidentially submitted a draft S-1 registration statement to the SEC, eight days after Anthropic filed its own S-1 on June 1. The back-to-back filings mark the opening of a public-market era for frontier AI labs that until now existed entirely as private, venture-backed entities. No pricing window has been set for either company.
The company entered its filing at a reported $852 billion valuation, with Goldman Sachs and Morgan Stanley named as underwriters. Bloomberg and Reuters both reported a September 2026 debut as the internal target, a timeline OpenAI did not confirm in its own announcement. The company’s statement was careful: the filing preserves the option to move quickly if conditions favor it, but the decision to remain private is still on the table.
OpenAI’s revenue run rate sits at roughly $2 billion per month, equal to about $24 billion annualized. The company reported approximately $13.1 billion in revenue for the prior fiscal year. It is not profitable. Compute, data center buildout, and training costs continue to outpace income by a wide margin, a fact that institutional investors will need to price against a valuation approaching $1 trillion.
Anthropic filed at a reported $965 billion valuation thirteen days earlier. That filing and this one are now the two data points institutional investors will use to establish a market-clearing price for frontier AI equity. The gap between the two valuations, roughly $113 billion, reflects different revenue profiles, safety positioning, and enterprise contract structures, none of which has been fully disclosed in either confidential filing. Confidential submissions shield financial details from public view until the company elects to move forward with a roadshow.
The timing of OpenAI’s announcement carries its own signal. According to OpenAI’s own statement, the company anticipated the filing would leak and chose to announce proactively. That is consistent with a company managing its own narrative ahead of a months-long investor process where information asymmetry works against the issuer.
Context from the week’s coverage matters here. The question of who holds equity in frontier AI labs at IPO has been active on three simultaneous tracks: institutional (these S-1 filings), retail (whether ChatGPT’s reported 900 million weekly active users become retail investors), and state (ongoing conversations about US government stakes in frontier labs). Each track feeds the others. A government-negotiated equity position, if it materializes before either IPO prices, becomes a disclosed item in the final prospectus and a factor in investor confidence.
For builders and operators who have been watching the AI infrastructure spend data this week, including enterprise software allocation surveys and subsidy analyses, these two filings convert that narrative from abstract to priced. The institutional investors who will anchor both offerings are the same parties funding AI infrastructure buildout. What they pay for OpenAI and Anthropic equity sets the implied floor for the entire category.
OpenAI’s ChatGPT reportedly reached 900 million weekly active users, a figure cited in coverage by Bloomberg. If accurate, it is the kind of top-of-funnel metric that consumer IPOs are priced on. OpenAI is not primarily a consumer company at this point, but that number will appear in every roadshow deck regardless.
If both companies list within the same calendar quarter, equity analysts and portfolio managers will have to make a direct choice between the two. That head-to-head pricing dynamic has not existed before in frontier AI. The result will produce the first market-derived answer to a question the industry has been debating for two years: which architecture of frontier AI, safety-led or capability-led, commands a premium at scale.
Investors tracking the AI sector should treat September 2026 as the planning horizon for both listings and watch for the public S-1 filings, which will disclose the financial details currently shielded by the confidential submission process.
OpenAI (openai.com), 2026-06-09. Also covered by Fortune, Bloomberg, and Reuters.
