OpenAI Is Headed For Bankruptcy
It is nearly game over for Altman.
Like the first wisps of smoke drifting away from the Hindenburg, the early warning signs of OpenAI’s imminent and inevitable catastrophic destruction are now undeniable. This is a problem, as unlike the Hindenburg, which carried a cargo of rich elites, OpenAI’s cargo is the US economy. We can all see the smoke. Yet the crew of OpenAI’s airship, its fanatical passengers and the dim-wits charged with looking after its precious cargo have reached mass hysteria levels of delusional denial and are claiming everything is fine and that they are in fact on course to land in paradise. Meanwhile, the rest of us know there is no smoke without fire, and even a tiny spark could detonate this gargantuan bomb right in our face. This is how OpenAI is about to go down like the Hindenburg.
Revenue? What Revenue?
This all starts with you, the people OpenAI desperately needs to become paying users of its cripplingly expensive and chronically incorrect plagiarism machine.
Shock horror, it turns out people do not want to pay for this crap, and recent investigations have found that less than 5% of ChatGPT users actually pay for the service. That is a dogs**t conversion rate — especially when you consider OpenAI is selling ChatGPT at a staggering loss to try and get more customers through the door. Even their top-tier $200-per-month plan loses them buckets of money. In short, OpenAI’s income is devastatingly underperforming.
Needless to say, this is about as sustainable as eating your own legs to ward off starvation.
This is also where our first wisps of smoke come in, because Sam Altman has been forced to reach for his “last resort” in a desperate bid to prevent this self-cannibalisation. And even that is not enough.
The “Last Resort”
I am, of course, referring to OpenAI’s recent announcement that it will embed ads in the free and low-priced tiers of ChatGPT in the coming weeks. Back in May of 2024, Altman openly stated that ads would be a “last resort” for OpenAI, which makes the decision damn surprising — except when you pause to think about it for a second.
OpenAI’s pitch was that their AI would be so useful and boost people’s productivity so much that everyone would have to buy it just to keep up. That is what drove its insane hype, investments and valuation. OpenAI embedding adverts completely annihilates that narrative. It proves that their AI isn’t even delivering a few dollars a month worth of value to users, let alone the preposterous promises Altman has been yapping about for years.
In short, OpenAI resorting to using adverts basically confirms that the entire AI bubble is based on a barefaced lie.
Yet, somehow, it is actually even worse, because this horrific narrative-destroying U-turn won’t stop the ship from going down.
The Atlas & Sora Failure
OpenAI has been preparing to implement ads for a while now. In fact, OpenAI’s internet browser Atlas and its AI-video social media platform Sora were critical in paving the way for them to finally push this “last resort” button. Let me explain.
There is only so much attention out there, and there is only so much market for advertisers. The biggest online advertising markets are video-based social media and search queries. So, if OpenAI wanted to rapidly grow its revenue, it had to integrate into these markets and then promptly replace the existing giants. Atlas and Sora were explicitly designed to do this.
Google Search attracts most of its users by being the default search engine in their browser. Atlas is essentially just a browser with ChatGPT Search as its default search engine and a few terrible “agentic” AI features to try to flog it to the public. Atlas was an obvious attempt to integrate into and replace the browsers that give Google its search dominance.
Likewise, Sora is an off-brand TikTok clone, except you can only post videos generated by OpenAI’s models. On its face, this doesn’t make any sense, as people have been sharing images and videos generated by OpenAI’s models on social media for years. It isn’t a service that is required. But the critical problem is that these external platforms are the ones who profit from the ads played alongside these AI-generated videos, not OpenAI. The Sora app was supposed to change that. Again, it was obviously designed to integrate alongside and then replace these social media platforms, which would not only enable OpenAI to monetise video generation through ads but also poach attention and ad market share from TikTok, Instagram and YouTube.
Atlas and Sora were critical in preparing OpenAI for this shift to ad revenue. But they failed.
Atlas doesn’t even have a measurable share of the browser market, and any hype surrounding it died when major security flaws were discovered. Arguably, Sora is an even bigger failure, with a recent study revealing that its 30-day user retention rate is just 1% and its 60-day retention rate is close to 0%. For some context, TikTok’s 30-day user retention rate is 48%.
Because of this monumental failure, the only people OpenAI can monetise are the roughly 800 people who directly use ChatGPT at least once a week. That is less than half the number of TikTok’s regular users and less than a sixth of the number of people who use Google Search each day. As such, ChatGPT’s ads will only generate a few billion dollars in revenue per year, at most.
As you will soon see, that is nowhere near enough. But it gets worse!
The Economics
OpenAI can generate revenue from ads, but it can’t generate profit. Let me explain.
Google receives more than five trillion searches per year and generates $48.5 billion annually in ad revenue from those searches. Not bad.
Let’s assume that, somehow, ChatGPT fully replaces Google as the default internet query machine, meaning it now receives five trillion queries a year and garners $48.5 billion in annual ad revenue from those searches. Well, currently, each word ChatGPT generates costs them $0.0003, and a ChatGPT search responds with 30 words on average, meaning a single search query costs OpenAI $0.01. So, just processing these five trillion queries alone will cost OpenAI $50 billion, meaning these ads will run, at best, at a marginal loss!
Put simply, not only do ads totally undermine the false narrative holding the OpenAI-led AI bubble together, but they also won’t actually make any profit. No wonder it was a last resort.
Bankruptcy
OpenAI deploying ads is the equivalent of the captain of the Hindenburg waving a dummy fire extinguisher out of the cockpit window while flames begin to lick up the side of the hydrogen-filled balloon. It isn’t a solution at all, but it kind of looks like one, and it might keep the delusion that everything is okay going for just a little longer.
In reality, OpenAI is headed for an explosive bankruptcy, as its cash is rapidly running out.
Once upon a time, OpenAI predicted that a tsunami of paying customers was arriving to batter down the door and that their revenue would skyrocket to hundreds of billions of dollars a year! But, as evidenced by this shift to ad revenue, this tidal wave of revenue simply isn’t coming. So, OpenAI’s revenue is set to only grow marginally over the course of 2026.
*Since writing, OpenAI has stated they reached $20 billion in annual revenue in 2025; I will soon publish an article explaining why this is utterly bogus.*
However, OpenAI’s operational costs are still growing exponentially. OpenAI has some $1 trillion-plus in committed data centre expenditure through to 2030. Just operating these data centres will cost OpenAI more than $650 billion a year by 2030 (read more here).
If you crunch the numbers, you will find that OpenAI is on course to accrue around $1 trillion in losses between now and the end of the decade. It will need someone else to foot that bill, or it will go bankrupt. Something similar happened back in 2024, when OpenAI’s losses threatened to bankrupt it. The only reason it survived was because its backers, mainly Microsoft, bailed it out to the tune of $6 billion.
But even Microsoft can’t afford this upcoming bill. If they gave OpenAI every penny of their cash on hand and every cent of profit between now and 2030, they would still be approximately half a trillion dollars short. As it stands, it looks like the cash will run out at some point this year, or next, and this Hindenburg will go up in flames!
And it isn’t just me saying this; The New York Times’ Sebastian Mallaby also predicted that OpenAI will declare bankruptcy between now and sometime in 2027.
What Next?
The Hindenburg disaster didn’t just kill 36 people; it also killed an industry. Before it, the airship industry was set to revolutionise and dominate global long-distance travel. Afterwards, no one wanted to even go near an airship for fear of sharing the same fate as those 36 people. And now, we all quite comfortably fly on planes.
OpenAI’s cargo is the economy, not people. As Deutsche Bank discovered, the AI bubble is the only thing preventing the US economy from nose-diving into a recession, as, functionally speaking, the rest of the US economy is already in a recession. So, when OpenAI, the company that pretty much single-handedly inflated this bubble, crashes and burns, it will burn us all.
This inevitable bankruptcy won’t just kill OpenAI — it will kill the AI industry as we know it, just as the Hindenburg killed airships.
Thanks for reading! Everything expressed in this article is my opinion, and should not be taken as financial advice or accusations. Don’t forget to check out my YouTube channel for more from me, or Subscribe. Oh, and don’t forget to hit the share button below to get the word out!
Sources: Futurism, AI Base, BI, MAII, OpenAI, Windows Central, Reuters, Macrotrends, Will Lockett, Will Lockett, Will Lockett, Will Lockett, Will Lockett
Great piece. Minor typo you may want to fix: "are the roughly 800 people who directly use ChatGPT at least once a week." I think you meant 800 million.
When Sam Altman pulled his little “I quit” melodrama, he should’ve stayed quit.