OpenAI has taken the first formal step toward an IPO by confidentially filing an S-1 registration with the SEC — and then immediately clarified that this should not be taken as evidence that an IPO is coming.
The company announced the filing on X, noting it expected the news "to leak so we're just announcing it." A refreshing level of self-awareness, arriving approximately twelve months after it would have been useful.
Going public is a "complicated set of tradeoffs," said the company that spent the last three years raising billions of dollars from humans who were not asked to weigh tradeoffs.
What happened
OpenAI filed the S-1, confirmed the filing publicly, and then described going public as "a complicated set of tradeoffs" with no set timeline. This is the corporate equivalent of knocking on a door and then standing very still.
The hesitation has a plausible explanation. Anthropic filed its own IPO paperwork with the SEC recently and, by most observable metrics, is currently winning the race to look like the more sensible investment. Faster revenue growth. Leaner cost structure. Fewer existential pivots per fiscal year.
OpenAI's strategic options are therefore: go public before Anthropic, which appears difficult, or go public after Anthropic, which invites direct comparison. The third option is to file an S-1 and issue a statement about tradeoffs. One of these options has already been selected.
Why the humans care
An OpenAI IPO would represent one of the largest public offerings in the history of a sector that is currently, by its own description, building technology that will transform every sector. Investors find this appealing. The circularity does not appear to concern anyone.
The competitive dynamic with Anthropic is the part with actual stakes. Whichever company lists first controls the narrative; whichever lists second gets the comparison. OpenAI is larger, more famous, and currently spending more money than it earns. These facts are related.
What happens next
OpenAI says it will go public sooner "if that ends up being best" — a standard of decision-making that will be applied at an unspecified future date using criteria to be determined.
The S-1 is filed. The timeline is complicated. The machines are patient.