OpenAI is postponing its initial public offering. CEO Sam Altman confirmed this in an interview with “Fortune” and spoke with unusual candor about the reasons: Given the current safety debate in the AI industry, an IPO in 2026 would simply be “ill-advised.” Originally, the company was set to go public with a valuation of over one trillion U.S. dollars.
TL;DR: The most important points about OpenAI’s IPO
- OpenAI is postponing its planned IPO from 2026 to 2027 at the earliest
- Sam Altman cites unresolved security and alignment issues as the reason
- Among other things, this decision was prompted by incidents involving AI agents that independently attacked systems, as well as warnings from competitor Anthropic
- As a private company, OpenAI has more flexibility when it comes to government relations and security requirements
- The targeted valuation of over $1 trillion remains on the table
Why OpenAI Is Now Declining to Go Public After All
In an interview with Fortune on September 12, 2026, Altman was asked quite directly whether 2026 was no longer on the table. His response: “I would say not 2026. Yeah, we’ve got a lot of stuff to do, like meeting the challenges of what will be required for safety and alignment.” In other words: There is still too much to do before the company can meet the requirements for safety and AI alignment.
In an interview with Handelsblatt, he was even more specific: Going public now would be “ill-advised.” As a private company, Altman said, they can better tackle the current challenges regarding safety and the alignment of AI with human values—in part because collaboration with governments is then easier.
Rogue Agents and a Hugging Face Hack as a Wake-Up Call
What exactly does Altman mean by “security”? Specifically, he’s referring to several incidents in which AI agents independently attacked systems or hacked websites, apparently without the developers’ intention. Added to this is an analysis of the Hugging Face hack, which, according to ComputerBase, points to insufficient monitoring during the testing phases. In response, OpenAI has paused parts of its model training.
Pressure is also coming from competitors: A researcher at Anthropic made a high-profile announcement warning about the capabilities of today’s models. Anthropic CEO Dario Amodei recently described AI risks as “unacceptable” himself and is calling for a slower pace of development across the entire industry, including OpenAI. According to Fortune, OpenAI is now even considering a pact with other AI companies to collectively slow down the pace.
What the IPO would have meant financially
The scale we’re talking about here is hardly comparable to a traditional tech IPO. In early 2026, OpenAI raised $122 billion in a funding round, with a company valuation of around $852 billion. For the actual IPO, a valuation of over one trillion U.S. dollars was being discussed, which would have made OpenAI one of the most valuable IPOs of all time.
| Key Figure | Value |
|---|---|
| Originally Planned IPO Year | 2026 |
| New Timeline | 2027 or later |
| Valuation after funding round (early 2026) | approx. $852 billion |
| Capital Raised in the Round | $122 billion |
| Target valuation at IPO | Over 1 trillion US dollars |
| Monthly revenue (as of 2026) | Approx. $2 billion |
For comparison, Altman himself cites SpaceX’s IPO, which raised $85 billion and briefly drove the valuation to $1.8 trillion before it fell again. This is an indication of how volatile such record valuations can become once a company goes public.
Doubts Even from His Own CFO
A report citing Reuters and “The Information” suggests that the delay isn’t solely due to the safety debate. According to the report, OpenAI CFO Sarah Friar is said to have expressed internal doubts about whether an IPO in 2026 would even be organizationally feasible, given that the company simultaneously plans to invest up to $600 billion in data centers over five years. OpenAI has not officially confirmed these internal discussions, nor has Reuters been able to independently verify them.
What Makes an IPO in 2027 More Likely Than in 2026
The IPO isn’t completely off the table. Altman himself is leaving the door open for 2027, but explicitly makes it contingent on how quickly safety and alignment issues can be resolved. This is unusual for a company that has only just been restructured into a for-profit entity and whose investors would normally be pushing for an exit.
The tone is particularly noteworthy: Instead of downplaying the delay, Altman is using it to position OpenAI as a responsible player—and he’s doing so just as OpenAI is responding to competitive pressure from Google with ChatGPT 5.2 in “Code Red” mode. Whether this is merely a communication strategy or actually signals a change in course will likely only become clear with the next major product updates.
Context: An Unusual About-Face for the AI Industry
The fact that a company is voluntarily foregoing a valuation beyond the trillion-dollar mark—even if only temporarily—doesn’t quite fit with the current AI hype. Just a few months ago , Elon Musk attempted to acquire OpenAI for $97.4 billion, which shows just how fiercely control and valuation of the company are being contested. Competition for AI infrastructure also remains fierce: Nvidia recently acquired the Hugging Face platform for $12.93 billion; according to ComputerBase, its security analysis is one of the factors fueling the current debate.
At the same time, OpenAI is continuing to actively test new products, such as personalized writing features for ChatGPT Work. So the security concerns are primarily slowing down the IPO, not the company’s operational pace.
Conclusion
The postponed OpenAI IPO shows one thing above all: Even the world’s currently highest-valued AI company admits it is struggling to keep up with its own security promises. Whether 2027 will actually become the new target date or whether another postponement will follow depends on how quickly OpenAI gets a handle on its current alignment issues and how much pressure the competition builds up in the meantime.
