A year is a long time in AI. Just 12 months ago, Sam Altman was predicting his company OpenAI would build a super intelligence and fundamentally remake society. Now the boss of the ChatGPT developer is walking back those ideas after failing to make money from ads and erotic chatbots.

Meanwhile, rivals are storming ahead with plans to expand and go public on the stock market, in what is widely expected to be a season of record-setting initial public offerings (IPOs).

Elon Musk’s SpaceX, which owns xAI, is going to float this month, while Anthropic confidentially filed for an IPO on Monday in what the New York Times said could be a “once-in-a-generation” moment for Wall Street. The AI chip designer and Google’s parent company, Alphabet, is meanwhile raising $80bn to fund a larger AI infrastructure buildout – in what analysts say is the largest equity fundraising ever.

These flotations – and OpenAI’s, if it happens – will reveal much about the ambitions, and state of, the AI economy. The numbers involved are eye-popping; already there is talk of whether the IPOs themselves might strain limited pools of capital. Questions hang over the progress of the AI infrastructure build-out, and the degree to which AI tools can replace human labour. Some of the biggest questions hang over OpenAI, the “poster child” of the boom.

One might have been forgiven, back in the halcyon days of early 2025, for thinking that money – or at least, the question of a profit – wasn’t everything to Altman.

The OpenAI chief executive was blogging about creating “science fiction” tools that could “massively accelerate scientific discovery,” and “massively increase abundance”. US vice-president JD Vance was reportedly among those reading with concern the doomsday scenario AI 2027, which envisions a company loosely modelled on OpenAI building a super intelligence that destroys humanity.

Vast sums chased these visions of ultimate power; OpenAI announced a $500bn (£370bn) investment into US AI infrastructure, called Stargate, “to secure American leadership in AI”. (For good measure, it announced a less-expensive Stargate in the UK, which it shelved in April this year.)

But in 2026, it turns out that money is important – and OpenAI needs more. It may not be the best time to float, but there might not be a better one.

The bare facts of the business have not changed in the past year, even if belief in the machine god has tempered. This winter, OpenAI successively announced – then failed to execute – several strategies to monetise ChatGPT, including advertisements, which Altman had earlier said was a “uniquely unsettling” proposition and a “last resort”. Meanwhile, after teasing the idea of erotic chatbots in late 2025, OpenAI backed out of the idea in March, saying its other work was a “higher priority”.

Meanwhile, the Information reports that OpenAI made revenue of $5.7bn in the first quarter of this year – but with adjusted negative margins of -122%, meaning it lost $1.22 for every dollar it spent. These numbers are impossible to confirm, as OpenAI does not release financial details, but underscore a fundamental truth of the AI economy: the computing power which makes ChatGPT possible is expensive, and isn’t getting cheaper with scale.

Can OpenAI – valued at $852bn at its last funding round – mount a successful IPO? What happens if it all goes sideways?

“Until we see the numbers, it’s very difficult for us to assess the valuation,” says Russ Mould, investment director at AJ Bell. “We need to know what the projected market cap might be, what the revenues are … what cashflow looks like.”

In some ways, OpenAI is acting like a mature, IPO-ready business, says Adrian Cox, a Thematic Strategist at the Deutsche Bank Research Institute. At an event earlier this month in Australia, Altman struck a sharply different note around AI-related job loss than he did a year ago, saying he didn’t expect a “jobs apocalypse.” (Last year, he said that people will “find new things to do, new ways to be useful to each other” following AI-related social change.)

“The demands of a public company are very different from a private company and the transparency that is required in going public has caused not only a change in strategy, but also a change in tone around the company,” says Cox.

“The company’s clearly engaging in an important shift as it positions itself as the kind of mature company that you would expect to go public.”

There are caveats, though, including the reported clashes between OpenAI’s chief financial officer, Sarah Friar, and Altman over the company’s timeline to go public. Friar has reportedly expressed doubts that OpenAI is ready to go public this year, and reservations about whether it can cover its computing costs.

The question is, can it afford not to? “This feels like the year for mega AI IPOs,” says Cox. “The demand seems to be there from retail investors. The demand is also there from users for AI.”

There could be consequences if it does not. The market is at record highs, having weathered Iran-related wobbles earlier this year. There is only so much capital available to share with others planning to go public such as Anthropic – which last week overtook OpenAI with a valuation of $965bn – and SpaceX.

“Some might say that there is only a finite amount of financing to go around, there may be a downside to waiting under those circumstances,” says Cox.

Mould says: “You’ve got a market with enormous potential with a lot of very, very big people jockeying; very big players jockeying for position. We don’t know who’s going to win. They can’t all be valued as winners in the long run.”

What happens if OpenAI’s stock is a flop? For some – including Gary Marcus, noted AI critic and professor at New York University – it could signal the start of something bigger: cracks in the “fantastical thinking” that undergirds it – and its rivals’ business models, and, if it were to slide, potential knock-on effects for the broader economy.

A wide range of people might be exposed to this. Significantly, US indexes including the S&P 500 and the Nasdaq are considering changing their rules – or have changed them already – around including newly floated companies on their indexes. This could lead to ordinary investors being more exposed to OpenAI’s fortunes.

Then again, Facebook’s IPO was also a flop and it is now worth more than $1.5tn. The market has changed, said Cox – OpenAI is no longer the symbol of the AI boom, the way it was even a year ago. Its fortunes are not necessarily a sign of how AI is going generally.

“There has been a diversifying of the AI narrative over the past couple of years,” Cox adds. “So any one IPO is probably less likely to be taken as a bellwether for the entire industry than it would have been before.”

OpenAI declined to comment.