OpenAI’s CFO Just Told Employees the IPO Is Coming in 2027 or Sooner

OpenAI CFO Sarah Friar told employees at an all-hands meeting in San Francisco that the company plans to go public in 2027 or sooner, framing the IPO as "another fundraise" rather than a finish line. Friar cited internal figures showing revenue run rate up 35% quarter to date and enterprise revenue up 50%, and noted that Codex has reached 20 million users, up 3.5x. The company raised $122 billion in March 2026, and the listing is seen as providing liquidity for employees and early investors.

OpenAI’s finance chief told the company’s staff this week that a public listing is coming, and the message was unusually direct for a firm that has spent years signaling ambivalence about the public markets. At an all-hands meeting in San Francisco, Sarah Friar said the company plans to go public in 2027 or sooner and framed the eventual listing as part of a longer funding arc rather than an endpoint. The most useful line she delivered reframes the entire event. “The IPO is not a finish line. It is a milestone; call it another fundraise,” she told employees, according to CNBC’s Kate Rooney. That is striking from a company that raised $122 billion in March 2026. A business with that kind of private capital does not need public markets to fund the next training run, which means the listing is really about liquidity for employees and early investors, and about creating stock that can be used as acquisition currency. Public investors should understand which of those two things they are being invited to underwrite. What Friar Actually Told the Room Friar’s headline commitment was a timeline. “We will be a public company in 2027. We may go sooner if the business continues to inflect,” she said. The framing of “or sooner” matters because it is conditional on growth continuing at its current pace. Friar cited internal figures showing acceleration, with revenue run rate up 35% quarter to date and enterprise revenue up 50% according to CNBC’s reporting. Those numbers are self-reported at an internal meeting and were not audited or filed. You should treat them as directional signals about the pace of the business, not as disclosures a prospectus would carry. Friar has credibility on the mechanics. She has previously worked as a banker and as a public company CFO, so her use of the word “fundraise” to describe the IPO is deliberate rather than casual, and it should shape how investors read the eventual S-1 when it lands. Why the Enterprise Number Matters More Than the Headline The 35% overall run rate figure will get the coverage, but the enterprise line determines what kind of company OpenAI actually is at the point of listing. Enterprise contracts renew, expand seat-by-seat, and produce the sort of recurring revenue that public markets pay a durable multiple for. Consumer subscription momentum is real, and Codex reaching 20 million users, up 3.5x, is evidence of it. Consumer AI subscriptions are also easier to churn out of than a signed enterprise agreement with committed usage. A company going public on a consumer growth story trades on user counts and cohort behavior. A company going public on enterprise contracts trades on net revenue retention and gross margin, and the two get very different valuations for the same top line. The mix at the time of filing will therefore matter more than either growth rate in isolation. If enterprise keeps outpacing the blended figure into the filing window, that is the tell that OpenAI wants to be priced like an infrastructure business rather than a consumer product. Anthropic Timing Question Friar also addressed the elephant in the room: Anthropic is expected to go public as soon as September 2026 https://247wallst.com/investing/2026/07/07/what-betting-markets-really-think-about-the-openai-anthropic-and-databricks-ipos/ . Her response to staff was that OpenAI is “running our own race.” That is the right posture, although the sequencing has real consequences. Whichever frontier lab lists first sets the comparable multiple that every private AI company will be measured against, from foundation model rivals down to the application layer. Being second is workable if the first listing prices well and trades well, because a healthy comparable lifts everyone. Being second is harder if the first listing breaks, because the market’s willingness to accept AI-scale losses against AI-scale revenue gets tested in public in a way it has not been before. Confidential filings are a common step ahead of a listing, and the SEC’s confidential submission process gives issuers room to iterate with staff before going public. What you should watch between now and then is the enterprise disclosure OpenAI chooses to make, the pace at which Anthropic’s process moves, and any change in Friar’s language from “2027 or sooner” to something more specific. Until an S-1 is filed, a CFO’s timeline is a statement of intent. Contact email protected for any questions or corrections.