.png)
4 hours ago
1
It is almost funny how the world has only just cottoned on to the AI bubble. Every time I read the news, it feels like an arsonist complaining that the building is on fire. Every major banker, venture capitalist, financial pundit and media outlet that made AI hype their entire personality a year ago is now warning that when the AI bubble pops, it will crash our economy. Great… And whose fault is that, exactly? But what I find truly fascinating about this sudden snap back to reality is that it is totally surface-level. For example, none of these people have even tried to look under the hood of OpenAI, which is overwhelmingly responsible for this bubble. That is almost offensive, because they could probably build a mass grave with all the skeletons in their closet. So, let’s dig into OpenAI and put a name to these ghouls so terrifying no one can talk about them.
Earlier this year, The Information analysed OpenAI’s predictions and found that they were on track to post a $14 billion loss in 2026 (read more here). This was on the heels of OpenAI nearly going bankrupt and experiencing a $6 billion investor bailout at the end of 2024. So, needless to say, if losses grew to that level, OpenAI would be seen as a doomed venture.
Well, recently, The Information posted another report on OpenAI and found that in the first half of this year, they made $4.3 billion in revenue and posted a net loss of $13.5 billion. So, in the first half of this year, OpenAI lost as much money as they were predicted to lose for the entirety of next year! This means that OpenAI is losing about three times more money than it’s earning and is on track to post a $27 billion net loss by the end of the year!
Or, to put it another way, OpenAI’s 2025 revenue is on track to only be $3.1 billion more than last year, while its annual operational costs are set to be $24.1 billion more than last year. So, for every dollar of revenue growth OpenAI has, it is costing them $7.77!
I cannot stress how unprecedentedly dreadful that is. It shows that the promised future investors were piling their money into is a fairy tale. This is a money black hole.
Now, The Information did point out that half of these losses came from convertible equity issued to investors. Convertible equity is an odd blend of raising capital with both debt (loan) and equity (selling shares in the company). Essentially, it starts off as a loan, but one that can be “paid off” with a set amount of equity down the road rather than cash. In other words, OpenAI is having to pay off a huge amount of debt that it used to grow but didn’t grow enough, and now that debt is registered as a loss. This kind of issue won’t stop, though. As I covered in a previous article, the entire AI industry, including OpenAI, is turning to debt more and more to keep the lights on, as equity only gets you so far.
The Financial Times, being a ‘responsible’ fiscal news outlet, didn’t take this convertible equity into account and found that OpenAI had posted a loss of $8 billion for the first half of the year. Sounds better, right? Even if you take this moronically rosy analysis to heart, that still means OpenAI is on track to post a net loss of $16 billion in 2025! Again, that is still more than OpenAI itself predicted they would post next year.
No matter how you try to spin it, this shows that OpenAI’s expenditure is out of control and they are not getting anywhere near the returns they need to even be remotely financially sustainable.
But are we so sure the future is this bleak? The entire AI bubble is predicated on the notion that these tools will get radically better thanks to the truly gargantuan investment in AI and will eventually displace jobs and hoover up exponentially more revenue.
Well, earlier this year, OpenAI said it planned to reach profitability by 2029 and was aiming for $100 billion in annual revenue by then. That $100 billion figure relies on some wildly optimistic AI adoption rates, which have already started to slow down significantly this year (read more here).
But even OpenAI’s own numbers didn’t add up, and if you actually look into them, they suggest that by 2029, they will post annual losses in the low hundreds of billions of dollars, as the realistic overall operational costs of their $500 billion Project Stargate AI data centre are going to be several hundred billion dollars a year (read more here).
However, since then, OpenAI has announced another $500 billion in AI data centre deals over the next five years with AMD, Broadcom, Nvidia, and Oracle. So OpenAI is doubling its planned expenditure over the next five years!
I need to put this into context, because this is going to make OpenAI’s revenue problem so much worse!
Data centres are expensive to use. They cost roughly 3–5 times their build cost in operational costs over their 15-year lifespan, averaging out to an annual operational cost of 26% of their build cost. But to utilise a data centre, you need AI developers, people collecting data, people sorting data, people beta testing new models and such. This is why data centre operational costs are only around 40% of an AI company’s operational costs.
In other words, this $1 trillion planned data centre investment could cost OpenAI $650 billion in annual operational costs by 2029!
OpenAI’s predicted $100 billion revenue by 2029 was wildly optimistic and had very little backing it up. Surely, they have a much higher and more robust revenue prediction that would justify this enormous investment and colossal inflation of their operational costs?
Well, they increased that prediction to over $125 billion, but again, this number is largely plucked out of thin air and has no real justification. It is also far too small to justify this expenditure! OpenAI have said that they are “looking into” new revenue streams, though all they have suggested is more specific tools for businesses and erotic chatbots. However, as I will explain in a bit, these types of business tools are either useless or don’t need this colossal amount of infrastructure. The entire global porn industry is also only worth $100 billion, so even if OpenAI dominated that market, it wouldn’t even begin to bring in enough cash!
So, even if OpenAI hits its $125 billion 2029 revenue target, it will still be making an annual loss of half a trillion dollars.
But here is the thing: OpenAI’s revenue growth is slowing down dramatically. In 2023, they increased their revenue by 169% over 2022, and in 2024, they increased their revenue by 250% over 2023. In 2025, they are set to increase revenue by only 56% over 2024. With AI adoption rates falling and AI pilots failing across the board (more on that in a minute), OpenAI’s revenue growth is set to continue to slow down at the very least. By contrast, for OpenAI to even think about breaking even by 2030, they would need to at least triple their annual revenue every year between now and then. In short, OpenAI’s revenue growth is catastrophically off the mark, and I don’t say that hyperbolically.
But I can hear those who have Big Tech’s balls tickling their chin saying that such huge revenue increases are on the horizon, as this enormous investment in expanding AI training data and computational power will solve its current problems, allowing AI to rapidly displace workers and soak up huge amounts of revenue.
There is just one problem with this narrative: even OpenAI themselves don’t believe it.
We need to backtrack a little. What is preventing AI from taking everyone’s job right now? Well, a recent MIT study found that 95% of AI pilots failed to yield profit or an increase in productivity, and most of them actually significantly reduced productivity. Likewise, a METR report found that AI coding tools, which are meant to be the most promising application for generative AI, actually slow developers down. Both studies cited the same issue, “hallucinations”.
AI hallucinations are one of the best bits of PR ever. The term reframes critical errors to anthropomorphise the machine, as that is essentially what an AI hallucination is: the machine getting it significantly and repeatedly wrong. Both MIT and METR found that the effort and cost required to look for, identify, and rectify these errors was almost always significantly larger than the effort the AI reduced.
In other words, for AI (specifically generative AI) to be even remotely useful in the real world and have a hope in hell of generating revenue by augmenting workers at scale, let alone replacing them like it has promised to, it needs to cut “hallucinations” down to basically zero.
Enter OpenAI’s latest research paper. It found that hallucinations are a core part of generative AI technology and can’t be fixed or reduced from their current level by simply adding more data and computing power to these models. That is horrific news, because OpenAI’s colossal investment in data centres is trying to do just that, i.e., trying to make AI more accurate by buying a bigger sausage machine and feeding it more slop. They also found that “reasoning models”, like all of OpenAI’s latest models, which use hidden chain prompts to “appear” to think, make them worse. Loosely, these models turn a single query into many, and that just creates more opportunities for a hallucination to mess things up.
They were able to find a botched workaround called “active learning”, where AI systems have an algorithm slapped on top of them, which forces them to repeatedly ask clarifying questions in an attempt to reduce uncertainty. However, even this doesn’t totally eliminate hallucinations, and it is questionable exactly how much it can reduce them. What’s more, operating such models is so inherently expensive that OpenAI found that it is almost always significantly cheaper to have a human do the task instead.
This is why OpenAI’s plan to increase revenue by building specialised tools for businesses doesn’t make sense. Any form of generative AI will have these hallucinations, and yes, making a more specific tool will reduce them, but not by enough. Moreover, building these specific tools is insanely expensive, even more so than “active learning”, and yet it still won’t reduce hallucinations to the same extent. It is a non-starter. But OpenAI could be talking about AI data analysis tools, rather than generative AI. Such tools are hyper-specific, don’t hallucinate, and are extremely useful. However, these tools have been widely used for literally decades; it is an established industry already in which OpenAI has no advantage, as these are a totally different kind of AI. The entire industry is also “only” projected to be worth $100 billion by 2030, so even if OpenAI could dominate it (which they can’t), it wouldn’t increase revenue enough to meet their self-inflicted costs. Furthermore, these tools are a lot more efficient than generative AI and don’t require the giant computational power OpenAI has built. Most of them can work fine on a bog-standard computer. So, this revenue stream is a non-starter too.
To summarise, OpenAI’s losses are growing twice as fast as predicted. Yet it is doubling its planned AI infrastructure expenditure between now and 2030. Just to cover the operational costs of that infrastructure investment alone, OpenAI needs to at least triple its annual revenue each year for the next four years, which is at odds with the fact that its revenue growth has fallen off a cliff and could even reverse. For revenue to even begin to grow enough, AI needs to actually boost productivity in real-world businesses, and to do that, it needs to almost entirely eradicate “hallucinations”. But OpenAI itself knows that their solution to AI development of chucking more data and more expensive computation power at the problem won’t solve this issue at all, and the only possible solution that can be found might not reduce hallucinations enough and is so expensive it renders the profitability of AI totally off the table. Meanwhile, the only viable additional revenue stream they have identified, erotic chatbots, hasn’t a hope in hell of meaningfully increasing their revenue enough and is more a sign of how desperate they are getting.
With all of this in mind, do you think OpenAI will survive to see the 2030s? Or, do you think the money will have run out by then?
Let’s also not forget that OpenAI controls around 61% of the US generative AI market. It also soaks up a truly colossal amount of AI investment. To give you an idea of how much OpenAI dominates this space, Bloomberg estimates that venture capitalists have poured $192.7 billion into AI so far this year, which is less than 20% of the investment deals OpenAI has announced this year.
OpenAI itself is basically the AI bubble. So when it inevitably kicks the bucket, do you think it will take down the entire industry and all those who heavily invested in AI with it?
At this point, you might be asking, “Why is this happening?!” And I get it. It’s as if they are fully aware they are driving headfirst into a wall at 100 mph, and instead of stepping on the brakes, they are mashing the throttle.
But there is a painfully simple answer. Those who control AI companies, like your Sam Altmans of the world, don’t make money from the company being profitable. In fact, many don’t even take a salary. Instead, they make money from their shares in the company shooting up in value. And here is the kicker: AI companies aren’t valued on their current models’ performance, their revenue, or even their planned business fundamentals. No one cares about that. Instead, they are valued based on their spending on data centres, as the market falsely believes this is the only key to unlocking human-replacing AI. So all these AI CEOs, along with the venture capitalists and banks jumping on the bandwagon, are pumping dramatic amounts of money into AI infrastructure, knowingly pushing the industry into catastrophic losses and putting the entire financial system at risk, just to add yet more billions to their already overflowing bank balances, in the dashed hope they can exit before it all comes crumbling down.
It is greed at the cost of everything. That is why this is happening. And it is why OpenAI is so much more screwed than we realise. They aren’t developing AI. They are trying to make the bottom line go up at any cost.
Thanks for reading! 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: The Register, The Information, Will Lockett, Will Lockett, Will Lockett, Investopedia, Science Direct, OpenAI, SQ Magazine, Ed Zitron, The Information, Will Lockett, Getlatka, Bloomsberg, Reuters, SOS Int, VM, GVR, AU Investing
