You've seen the headlines about multi-billion dollar AI deals for everything from software to components. It's all fuelling speculation about a potential AI bubble. So how much do the companies at the centre of this speculation need to make to ensure the whole thing doesn't pop? The answer is astronomical.

Here's a quick illustration of the problem. In the first six months of 2025 alone, OpenAI - makers of ChatGPT - burned through $2.5 billion dollars. That's a huge number. But what's even bigger is the fact that CEO Sam Altman has announced almost a trillion dollars in various deals in the same period. The tech sector, and every sector that interacts with it, is making a huge, huge bet that AI will not only become a functional part of everyday life, but a profitable one.

The companies who have signed on various dotted lines in the last 12 months all need something to show for all the money they've spent. And considering how many people aren't paying for AI now (given that it's mostly being bundled for free in different software platforms), that's going to be tricky.

So let's do some back of the envelope maths with our friends from JPMorgan in the US. JPM this week put out a report featuring their own back-of-the-envelope maths, and the numbers are more than a little terrifying if you're a believer in the current bubble.

I've added emphasis here to drive home how wild this is, but JPMorgan shared that:

"The path from here to there will not just be 'up and to the right'. Our biggest fear would be a repeat of the telecom and fibre buildout experiences, where the revenue curve failed to materialise at a pace that justified continued investment. For now, commentary from large corporates suggests benefits are starting to be realised at scale. More interestingly, OpenAI just publicly commented that they have achieved a $20 billion annualized revenue run-rate already. However, breakthroughs or accelerated efficiency gains - as people initially thought occurred with Deepseek - could drive an overcapacity/dark fibre situation. Big picture, to drive a 10% return on our modelled AI investments through 2030 would require ~$650 billion of annual revenue into perpetuity, which is an astonishingly large number. But for context, that equates to 58bp of global GDP, or $34.72/month from every current iPhone user, or $180/ month from every Netflix subscriber. How that is apportioned between corporations, governments and consumers is, of course, a long-term debate. Regardless, even if everything works, there will be (continued) spectacular winners, and probably some equally spectacular losers as well given the amount of capital involved and winner takes all nature of portions of the AI ecosystem."

So even barring any future billion-dollar deals, companies in JPM's portfolio of AI investments would need to find almost three-quarters of a billion in pure revenue right now for them to be even 10% profitable. And if they wanted to charge a 'fair value' for the AI that's being given away right now for free, customers would need to stump up at least $US180 a month to get to those same revenue figures. I don't know about you, but I don't see that happening in the next couple of days.

It's not unusual to see tech companies trade in ways that would make most regular traders scratch their heads. Tech companies often trade at high valuations despite losing money because investors are betting on future dominance, not present profits. The logic is that once a platform scales, its costs barely rise while revenue can grow exponentially. A global user base, network effects, and the ability to monetise data or subscriptions later on make early losses look like down payments on future cash flow.

Low interest rates have also inflated tech valuations by making future earnings look more valuable today, while investors are drawn to powerful narratives about disruption and innovation. It’s part maths, part faith: the promise that today’s losses will fund tomorrow’s monopoly. But I am he of little faith, and I'm starting to think the AI bubble boys have a point.