CHAPTER 1THE FILE THAT IS NOT A PROSPECTUS YOU CAN READ
On June 8, 2026, OpenAI said it had recently submitted a confidential draft S-1 to the US Securities and Exchange Commission. Its whole public explanation was unusually candid about optionality: "We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company." The company added that submitting now gave it the option to go public sooner if that proved best. The announcement expressly says it is not an offer to sell securities. [1]
That first sentence is a fact about a procedural step. It does not reveal revenue, cash flow, losses, debt, cost of serving users, capital commitments, related-party deals or a proposed share price. It does not mean the SEC approved the business. It does not mean ordinary investors can buy shares. And it does not mean an offering must occur at all. The distinction may feel pedantic until someone tries to price a business on a confidential document they have never seen.
A private draft is a useful option for a company. SEC staff can comment while management weighs market conditions. Under the SEC's expanded nonpublic-review procedures, an issuer using this route confirms that it will publicly file its registration statement and nonpublic drafts at least 15 days before a roadshow, or before a requested effective date when there is no roadshow. The operational details can change; the core point is that a public prospectus must eventually arrive before an offering can proceed. Investors should read that actual filing and amendments, not a rumor about its contents. [2]
In August, CFO Sarah Friar reportedly told employees that OpenAI "will be a public company in 2027" and could be sooner if the business continued to inflect. CNBC, citing people familiar with the remarks, reported both that statement and her description of an IPO as "another fundraise." [3] In September, Reuters reported Sam Altman saying OpenAI would not go public in 2026, against a discussion of AI safety. [4] These are reports of remarks made at different times and under different conditions, not a binding offering calendar. At this report's date, the sensible working assumption is no 2026 listing, with 2027 a stated management direction rather than an announced offering date. The confidential filing should not be backfilled with a fantasy prospectus simply because the future year sounds specific.
Why submit confidentially? A company may want the SEC process to begin without publishing the financial and strategic details its rivals would like to study. OpenAI's own June statement points to actions easier as a private company. The exact motives are unknowable from that statement. Our inference is narrower: optionality has particular value when a company is making capital-intensive commitments, reworking products, courting enterprise buyers and negotiating partnerships at once. That inference is not a claim about management's hidden intention.
The contrast with Anthropic helps frame the uncertainty but cannot resolve it. The Commerce Stories short companion story reported that journalists had reviewed Anthropic's confidential draft and described its finances. A leak or press review of one rival's filing does not grant access to OpenAI's. We will not treat a revenue leak, a management slide described secondhand, or a private valuation as a line item in OpenAI's prospectus. [5]
For the investor, the first discipline is to sort documents into three piles: filed financial statements; management and partner announcements; and reported leaks. They are not interchangeable. For the individual, the IPO is a reminder that a service used for homework, writing or work can be both a helpful tool and a company seeking enormous continuing capital. For the consumer, the confidential process is not itself evidence the product is unsafe or unsound. It is evidence that you cannot yet audit the entire enterprise behind it.