OpenAI is leaning toward delaying its initial public offering until next year, three people involved in the company’s deliberations said, a turn that underscores the uncertain future of fast-growing artificial intelligence giants.
The maker of ChatGPT hired bankers and lawyers with an eye toward a public offering as early as the third or fourth quarter of this year, the people said. Sam Altman, the company’s chief executive, pushed those advisers to find a way to value the startup at $1 trillion, up from the company’s last private valuation of $730 million, according to the people involved, who did not want to be named because they were not authorized to speak publicly about internal considerations.
But a cascade of recent developments has caused OpenAI’s leaders to move away from their most aggressive ambitions. Top of mind is what has happened to Elon Musk’s SpaceX following its IPO this month. It was the biggest ever, raising more than $85 billion and reaching a valuation of $1.77 trillion upon its debut. But since then, SpaceX’s stock has been on a downward spiral, with shares falling to $153 at the close of trading on Thursday after hitting a high of $202 last week.
Global markets have also been volatile in recent weeks, with technology stocks dragging down indexes as investors question whether AI companies will live up to their lofty promises.
That has prompted OpenAI’s advisers, in talks with the company over the past week, to warn that it might not find much enthusiasm from retail investors for its own shares, two of the people involved said.
OpenAI’s brakes on its IPO plans could disappoint Wall Street and Silicon Valley. Public offerings by OpenAI and its rival Anthropic, which also said it planned a Wall Street debut, could trigger a wave of generational wealth. OpenAI said this month that it had filed confidential paperwork with securities regulators to jumpstart the process of going public, but it did not publicly commit to any timeframe.
A $1 trillion valuation on the public market, which would exceed Walmart’s market cap, would be staggering for OpenAI, a start-up that is not believed to have turned a profit and is aggressively spending on new data centers.
A spokesperson for OpenAI declined to comment further beyond the company’s previous statement.
OpenAI’s advisers presented company executives with the option of waiting until 2027 to go public at a $1 trillion valuation, or lower the target valuation for a faster IPO. Mr. Altman, a person briefed on the matter said, responded that any change to the trillion-dollar valuation was a nonstarter.
OpenAI also struggles with other issues. Late last year, Sarah Friar, the company’s chief financial officer, said it was not pursuing an IPO at the time and was focusing on strengthening its finances. Since then, it has continued to pour money into data centers and computing power, with no signs of slowing down.
The company also spends heavily on marketing and recruiting high-profile engineering talent from companies such as Meta and Google. It is looking for other revenue lines, including dabbling in placing ads in ChatGPT and striking e-commerce deals with companies like Shopify and Stripe that would allow people to buy things from online stores directly inside ChatGPT.
These initiatives are still in early experimental stages, two OpenAI employees said. OpenAI reported about $13 billion in revenue by 2025, one of the people said, a figure the company hopes to triple this year. OpenAI said this year it generated $2 billion in revenue each month.
But some OpenAI executives appeared to have changed their minds about an IPO just a few months after Ms. Friar said the company was not looking to go public. The Wall Street Journal reported that the company planned to go public by the end of 2026.
That surprised some employees because they believed the company was not on strong enough financial footing, said two people familiar with the company’s plans.
OpenAI is facing acute pressure. Anthropic, which offers a Claude Code tool for creating sophisticated software code, has had success selling its service to enterprises. At the same time, Google’s Gemini, the tech giant’s flagship product for consumer AI, has become popular among users.
After several years of increasing downloads of ChatGPT’s consumer app, those numbers have slowed and continue to hover around 900 million users, surprising investors who thought the company would easily hit a billion.
Over the past six months, OpenAI has undergone an almost complete overhaul. Under Fidji Simo, the managing director of artificial general intelligence deployment, OpenAI has begun to drop “side tasks” — a term describing non-essential tasks in a role-playing game — including money-losing divisions like its video generator app, Sora. And to match Anthropic, OpenAI is building a sales team to push Codex, its coding product, to larger enterprise customers.
Despite the hesitation about an IPO, OpenAI executives believe the company is moving in the right direction, according to two employees. More than five million people use Codex on a weekly basis, the company said in a blog post this month.
The company also recently announced that it had more than two million business customers. And last week, the company recruited Noam Shazeer, away from Google, a hire widely seen as a coup across the isolated AI research community. Mr. Shazeer was one of the authors of a 2017 paper that introduced the “transformer architecture” in AI (or the T in ChatGPT).
(The New York Times has sued OpenAI and Microsoft, alleging copyright infringement of news content related to AI systems. The two companies have denied those claims.)
