OpenAI has positioned itself as one of the most important tech companies throughout 2025. It has placed itself at the heart of the $500 billion Stargate infrastructure investment scheme in the US, part of the Stargate UK build-out of new data centers worth tens of billions, and it's alleged to have pledged $10 billion to Broadcom for a custom chip deal too. Not to mention the $300 billion it's promised Oracle over the next few years. The only problem is: OpenAI doesn't have anywhere near the cash to pay for any of it.
OpenAI lost $5 billion last year alone, and is on track to generate just $13 billion in revenue this year. Amidst all of this, OpenAI is undergoing a restructuring towards a for-profit model; it's still currently a non-profit, and its revenue-generating plans are still unclear.
That's not necessarily uncommon for big tech firms. Amazon famously lost billions in the late 90s and early 2000s, as it re-invested to grow the business (and some would say, avoid taxes). As Sherwood News notes, Uber, Tesla, Snap, and Netflix all burned billions of dollars in building their businesses before they reached profitability. The difference this time is the scale.
Where they collectively burned through just under $40 billion, OpenAI believes it will burn through $115 billion over the next four years, which is a staggering amount of money. Especially considering that OpenAI has only managed to raise just shy of $64 billion in its entire history.
Sure, $40 billion of that came this year, but that is contingent on it completing a for-profit restructuring before the end of the year, and even that huge amount of money wouldn't even cover a single year of OpenAI's promised investment to Oracle, let alone everything else.
So, where is the rest of the money going to come from? Immediate investments from the likes of Nvidia will help, but that's going to start with an insufficient $10 billion and trickle in, as OpenAI data centers come online in the following months and years. There's also been talk of global government investment from the likes of Saudi Arabia and India, but that's far from finalized.
It may all rest on an IPO
(Image credit: OpenAI)
A larger proportion of the investment funds needed for all of OpenAI's projects is likely to come from a future IPO. OpenAI is already valued at over $500 billion thanks to a mix of massive investments, a strong-hold on industry mindshare, and OpenAI's admittedly impressive 700-800 million weekly users (though crucially, only a fraction pay for the service). It's speculated by Nasdaq that OpenAI could end up as the first trillion-dollar IPO.
Though Microsoft will be an immediate major shareholder as part of any public listing, with the software giant expected to own at least 30% of OpenAI. That's part of the ongoing negotiations with Microsoft, which are said to be progressing, but are far from finalized.
Before OpenAI can launch an IPO, though, it needs to become a for-profit company. That alone could push a potential IPO further into the future.
That may be something OpenAI is okay with for now. CEO Sam Altman said to Bloomberg in August that an IPO for OpenAI was "not a priority," though that hardly rules it out.
Considering OpenAI needs to generate hundreds of billions of dollars over the next few years, which it can't get from revenue, and appears unlikely to generate it from outside investment, an IPO seems like the most viable solution to its financial balance sheet in the near term.
Pull the trigger before the bubble pops
OpenAI is under pressure to get its restructuring complete before the end of the year to unlock the full Softbank investment, but that's not the only timeline it needs to meet. The sheer scale of investment in the AI industry and the incredible speed at which it's come about has prompted many to draw comparisons between AI and the dotcom bubble of the late-90s and early-2000s. When that happened, many huge companies either folded, or lost enormous sums of money. A potential AI bubble could be even worse.
Even OpenAI CEO Sam Altman believes the AI industry is overinflated, and the company's chairman agreed with him in a recent interview with The Verge.
He didn't mince words either, stating: "I think we’re also in a bubble, and a lot of people will lose a lot of money."
Just not OpenAI. That's the plan at least, and perhaps why we're seeing such efforts from all the companies involved in it to keep the hype and gravy train rolling. Nvidia has invested $100 billion in OpenAI, and OpenAI is buying (or leasing) Nvidia GPUs, and buying Oracle compute power. Oracle is also backing the OpenAI Stargate project, and Nvidia is also backing CoreWeave, who just announced billions of dollars in investment in OpenAI.
It's hard not to speculate that these companies are all announcing enormous deals with each other to pump share prices and encourage further investment. In the near term, if OpenAI is going to launch an IPO at a sky-high valuation to generate the funds it needs for all future investments, it will need to do it before that bubble pops.
Everyone's trying to sell the shovels
(Image credit: Tom's Hardware)
OpenAI still hasn't quite figured out where the profit lies in AI. Indeed, neither has just about anyone in the space. There is no proven profitable business model in AI yet.
The only companies making real money in the AI gold rush for now are the ones selling the digital shovels. Nvidia effectively holds a monopoly on the fastest AI hardware with its GPUs. That may change in the years to come as other firms catch up, or AI developers pivot towards application-specific integrated circuits (ASICs) in a fashion similar to how cryptocurrency mining evolved beyond GPUs
For now, Nvidia is king, and it's clear that AI is a part of the company's core strategy moving forward, leveraging the idea that a cyclical investment is keeping AI companies going.
Nvidia wants to keep building the hardware to run the AI, and OpenAI wants to provide the software shovel to go alongside it. It also pushes the idea that compute and electricity can equate to money.
For now, OpenAI needs a lot of money, and though there are some potential sources for it, even those could fall far short of what OpenAI has already promised.
In a moment of candour, Altman even seems to acknowledge that the numbers don't quite add up. In a blog post, Altman discussed the compute in, money out equation:
"Over the next couple of months, we’ll be talking about some of our plans and the partners we are working with to make this a reality. Later this year, we’ll talk about how we are financing it; given how increasing compute is the literal key to increasing revenue, we have some interesting new ideas."
Those ideas will need to be very interesting if the numbers are ever going to add up. For now, we'll have to wait and see how OpenAI attempts to fulfill its promises.
Jon Martindale is a contributing writer for Tom's Hardware. For the past 20 years, he's been writing about PC components, emerging technologies, and the latest software advances. His deep and broad journalistic experience gives him unique insights into the most exciting technology trends of today and tomorrow.
