OpenAI has filed confidentially for an IPO, but the more revealing fact is that the company still sounds uncertain about public life. When a business needs public-market money and its own chief executive has called the job of being public really annoying, you should pay attention to the gap.
OpenAI is no longer just the company that made ChatGPT feel unavoidable. It is now a capital-hungry AI business preparing to test whether public investors will fund the same enormous spending spree that private backers have been willing to tolerate. The Wall Street Journal reported last week that OpenAI said it had filed confidentially for an initial public offering, while adding that it had not decided on timing and that it may be a while. That caveat matters. This is an IPO story, but it is also a hesitation story.
Sam Altman has made that hesitation unusually easy to see. He has described the prospect of running a public company as 0% exciting and really annoying, even while OpenAI has moved closer to exactly that outcome. You can dismiss those comments as founder candor if you want, but investors will not. A public company is not only a funding machine. It is quarterly earnings, analyst calls, disclosure rules and a permanent audience judging whether the spending still makes sense.
The spending is the reason OpenAI is here. According to figures reported by Yahoo Finance, OpenAI's internal projections show roughly $13 billion in 2026 revenue against about $14 billion in losses, with losses continuing before an expected profitable year in 2029. HSBC analysts have estimated a $207 billion funding gap through 2030. Those are not normal startup numbers with bigger zeros. They are infrastructure numbers, and they explain why even giant private rounds eventually stop looking sufficient.
OpenAI already changed its structure to make this future easier. In October 2025, as The Guardian reported, the company converted its main business into OpenAI Group PBC, a public benefit corporation overseen by the OpenAI Foundation, after review by regulators in Delaware and California. Microsoft received roughly a 27% stake under the new arrangement, and the business was valued at about $500 billion at the time. The PBC label is not meaningless, but don't mistake it for armor. Once public investors own the stock, the pressure moves from the charter to the calendar.
That is where the story gets harder for OpenAI. A company built around long-term model research may soon have to explain itself every three months. Public investors can tolerate losses when they believe growth is still clean, dominant and obvious. They get less patient when the market starts splitting. VentureBeat has reported that Anthropic's Claude has been gaining ground with business customers, while Google's Gemini continues to take share in consumer chatbot use. ChatGPT is still the name most people know, but name recognition is not the same thing as unchallenged control.
The timing is also less comfortable than it looked a year ago. Tom's Hardware, citing The Wall Street Journal, reported that a coalition of 42 U.S. state attorneys general served OpenAI with a subpoena on June 12, seeking documents tied to advertising, data practices, minors, health information, model behavior and safety policies. A subpoena is not a finding of wrongdoing. But if you are preparing to sell investors on a trillion-dollar AI story, a multistate probe is not background noise. It belongs in the foreground.
Sarah Friar's reported caution makes the internal picture more interesting, not less. The Economic Times, citing The Information, reported in April that OpenAI's CFO had raised concerns about Altman's plan to go public in late 2026 and spend up to $600 billion over five years. A CFO is supposed to worry about those numbers. Still, when the person responsible for the financial story is reportedly concerned about timing and scale, investors are right to ask whether the IPO is a confident step or a forced one.
Frankly, the public markets may be the only place large enough for OpenAI's ambitions. Training frontier models, leasing compute, building data-center capacity and competing with Anthropic, Google and Meta costs more than even rich private investors can comfortably absorb forever. If OpenAI wants to keep setting the pace, it needs money on a scale that ordinary venture capital was not built to provide.
But you should not confuse inevitability with readiness. OpenAI can be the most important AI company in the world and still be a difficult public company. It can have a mission structure and still face Wall Street pressure. It can have Altman's vision and still need Friar's caution. The S-1, whenever it becomes public, will have to do more than show revenue growth. It will have to show whether this business can turn extraordinary attention into economics that survive outside the private-market bubble.
Also read: The US government just pulled the plug on Anthropic's most powerful AI models and the whole industry is watching, Britain's top data regulator quits over misconduct findings just as its AI enforcement credibility is on the line, and Jio Platforms files its IPO papers with SEBI and Mukesh Ambani wants you to see it as an AI company
