Bubble, bubble, toil and trouble

It’s Halloween and it seems like an appropriate time to confront something genuinely scary: the prospect of an AI bubble. A bubble happens when there is overinvestment, overvaluation and a general feeling of overenthusiasm. The dot-com bubble is one of the most famous ones where investors poured money into anything ending in .com. If you’re too young to remember, ask someone like me who was around at the time. When the bubble burst, it wasn’t pretty.

And it feels similarly precarious right now. Just this week, we saw Nvidia, the GPU powerhouse driving AI data center growth, reaching a staggering $5 trillion in value. How about Reuters reporting that OpenAI could go public next year with a valuation as much as $1 trillion? These are not normal numbers.

Investors have sunk billions into any company related to AI. The category dominates VC investment dollars, according to Pitchbook. In a July study, the firm reported that $74 billion was invested in AI startups over the past year across 1,603 deals, both records. Suffice to say there was a lot of money being thrown around.

We also know that trillions of dollars are being dedicated to building data centers in the coming years to support AI expansion, an output of capital the likes of which we likely have never seen before (and which could be put to work to solve a lot of the world’s problems if there was the same willingness). Synergy Research reports that 137 new hyperscale data centers came online worldwide in 2024 and there are sure to be a lot more this year, judging from all the announcements.

We have seen a ton of hoopla and gobs of investment, yet the reality is still not matching the hype. That sounds like we’ve crossed over into bubble territory to me.

Walks like a bubble

When you have the likes of Jeff Bezos, Bill Gates, Jamie Dimon and even Sam Altman calling this a bubble, maybe it’s a bubble. It’s not that AI doesn’t have a lot of potential value, it’s just that when you have investors circling around one category like AI is now, it's probably not going to end well.

In an interview during Italian Tech Week earlier this month, Bezos said that he definitely sees a bubble, but he doesn’t think it will necessarily be catastrophic. He distinguishes between what he calls a financial bubble and an industrial one, and he sees this as the latter.

“The ones that are industrial are not nearly as bad. It could even be good because when the dust settles and you see who are the winners, society benefits from those inventions,” he said. “This is real. The benefits to society from AI are going to be gigantic.”

He could be right, and fallout from such a situation wouldn't be disastrous when all is said and done, but in the short term, as we’ve written here, companies are struggling to find ROI on this technology. They are abandoning projects, and it’s fair to wonder what happens when enough companies fail to see the value.

AI over-selling

The disillusionment could be setting in because vendors are consistently overpromising when it comes to AI utility. We could already be seeing it. Consider that Forrester is predicting that executives will begin looking much more closely at their AI spend in 2026. 

“With fewer than one-third of decision-makers able to tie the value of AI to their organization’s financial growth, CEOs will lean more on their CFOs to approve AI investments based on their ROI in 2026,” the firm wrote. “As financial rigor slows production deployments and wipes out proofs of concept, enterprises will defer a quarter of their planned AI spend into 2027.”

From my experience — see my Adventures in AI every week in this newsletter — the current generation of AI has some serious limitations. I certainly don’t think it’s ready to replace people in any meaningful way, but there is a ton of potential here.

The ability to interact with the model and ask questions, have a sort of conversation, is powerful, but whether it’s enough to convince customers to keep investing in the face of continuous lack of ROI is not clear.

If businesses continue to fail to see the value and begin to scale back their AI investments, it probably won’t mark a hard stop as we’ve seen with past AI boom-bust cycles. But we could see a slow-down coming as companies try to figure out how this technology fits their overall strategy, and investors who went all in, could end up getting spooked along the way.

~Ron

Featured image by Wilhelm Gunkel on Unsplash