In the San Francisco offices of OpenAI, staff are preparing to celebrate a historic milestone.

At some point in the coming months, ChatGPT, the company’s AI bot, will join a select group of apps to boast more than one billion users.

It will have done so faster than any other consumer product in history. It took Facebook eight years to get there and TikTok – the fastest-growing to date – slightly more than five.

ChatGPT will take around three and a half years.

Since its release in late 2022, OpenAI has become one of the world’s most valuable start-ups, raising tens of billions of dollars and making Sam Altman, its chief executive, one of Silicon Valley’s most prominent figures.

But even as it breaks records, OpenAI is facing questions about whether the vast sums investors have ploughed into the company will ever be repaid.

Some have even speculated that the poster child of the AI boom could run out of cash and potentially bring down much of the US tech sector with it.

OpenAI, founded 11 years ago, has simultaneously become one of the most valuable and loss-making companies in history.

The business has been valued at $850bn (£640bn), bigger than all but 10 companies on the US stock market.

Separately, it has told investors that it plans to burn through $115bn by 2029 – almost four times the record-breaking $31.5bn Uber lost before making its first profit.

Sky-high salaries for researchers and the extraordinary cost of running its energy-hungry AI systems have led to questions over whether the company can ever justify that investment.

In the past, Altman has outlined plans to spend $1.4tn on data centres to power its technology.

This led Sebastian Mallaby, the economist and writer, to suggest earlier this year that the company could run out of cash.

“My bet is that over the next 18 months, OpenAI runs out of money,” he wrote in The New York Times.

OpenAI – which is backed by Nvidia, Amazon and Microsoft – does not make its finances publicly available, but a series of defensive actions in recent weeks has suggested that the company’s funding is far from unlimited.

The company shut down Sora, its system for creating AI-generated videos, in part because of the exorbitant cost of running it. It also shut down a TikTok-like app for viewing the clips.

Using AI to make videos is much more expensive than a chatbot conversation, so it was a clear candidate for cuts.

“The economics are currently completely unsustainable,” said Bill Peebles, the head of Sora, last year.

Enders Analysis, a research group, said the system had a “clearly limited commercial runway”.

OpenAI has reportedly suspended work on an “erotic chatbot” for over-18s as part of a push to drop “side quests” that were not central to its strategy.

It cancelled a multibillion-dollar deal this month with the data centre provider Oracle to expand the flagship Stargate data centre in Texas.

A separate project in the UK, with the British data centre company Nscale, was promised by the end of this month but neither company has provided an update since an announcement in September.

Gil Luria, the head of technology research at US investment firm DA Davidson, says the company has had to pull investment plans under pressure from investors and as rival AI providers close in.

“Up until six to nine months ago, they were running away with AI and they were the dominant player. Everything they did was great,” he says. “In the subsequent six months, it became clear they were making commitments they couldn’t live up to.”

A high water mark for OpenAI came last September, when the company first unveiled plans to spend $300bn with Oracle. The announcement briefly made Larry Ellison, Oracle’s founder, the world’s richest man as his company’s shares soared.

“There was this sense that everything they touched turned to gold,” says Luria. “Then people started pointing out that it didn’t have $300bn to spend.”

The Midas touch seemed to fade in November when a defensive-seeming Altman bristled during a podcast hosted by Brad Gerstner, an OpenAI investor.

When Gerstner asked how OpenAI’s numbers would add up, Altman snapped: “Enough.”

Sarah Friar, the company’s chief financial officer, also suggested the US government could provide a financial “backstop” in comments that were later walked back by the company.

Later that month, AI rival Anthropic released a new version of its Claude bot that many experts said surpassed OpenAI’s.

Use of Anthropic’s computer programming tools exploded in December, making it appear for the first time that OpenAI was playing catch-up.

Nathan Benaich, an AI investor, says ChatGPT now “lags Claude significantly” and that the company’s video and image generation systems are also behind the curve.

Altman declared a “code red” in December (ironically, Google had used the same phrase shortly after the launch of ChatGPT threatened to decimate its search business).

Altman ordered staff to direct resources to its core technology and put a series of ambitious other projects on hold.

Plans for a physical device – developed with Sir Jony Ive, the iPhone designer – and an OpenAI project to produce its own AI chips have both gone quiet in recent weeks.

Fidji Simo, an OpenAI executive, said earlier in March that the company was “doubling down” on areas it was succeeding in, such as programming app Codex, and wanted to “avoid distractions”.

A spokesman said the company was not scaling back its infrastructure spending and planned to triple capacity this year.

Luria says that rowing back on the company’s frantic spending was probably a condition of the company’s latest funding round, in which it secured up to $120bn from investors including Amazon and SoftBank.

These fundraising numbers do not necessarily mean money in the bank, however.

The funding round included a $50bn commitment from Amazon but the majority of this depends on the company going public or hitting undisclosed milestones.

A $1bn investment from Disney announced last year was cancelled last week without the money changing hands and a $100bn commitment from Nvidia announced last year was whittled down to $30bn.

Running the company remains tremendously expensive.

Richard Windsor, an independent tech analyst, estimates that the company’s spending on computing has risen at the same pace as its revenues for the last three years.

“You can see that this is not exactly a particularly good financial proposition,” he says.

To break even, this means the company either needs to wring more money out of its users or AI models need to become much cheaper to run.

While OpenAI sells monthly subscriptions that provide access to more powerful systems and extra features, an estimated 95pc of its users use its free service.

The company has started to introduce advertising, which Altman once called a “last resort”.

Soon after, Anthropic released a Super Bowl ad mocking the move, prompting a spiky response from Altman.

But the company has little choice. It is targeting a public listing as soon as this year that it hopes will value it at more than $1tn.

“The consumer AI ecosystem is a must-win if it is ever to justify [that] valuation,” Windsor says. “If people get fed up with pouring money down a black hole, you can very quickly see how the company gets into trouble.”

Larry Fink, the boss of investment giant BlackRock, said this month that the tech industry’s exuberance was likely to lead to “one or two bankruptcies” among large AI companies.

The largest of them all has investors’ faith for now. But it would only take a couple of wrong moves for that to change.

An OpenAI spokesman said: “With user demand outstripping supply, compute is the critical resource when it comes to AI. Along with locking in our long-term compute needs via our infrastructure strategy, we are also ruthlessly prioritizing the allocation of that compute across where it drives the most long term economic value.”

Try full access to The Telegraph free today. Unlock their award-winning website and essential news app, plus useful tools and expert guides for your money, health and holidays.