It’s hard to avoid the hype surrounding AI these days: Promises of new developments like personal robot assistants and miracle cancer cures are everywhere, and company executives take every opportunity to tout AI’s capabilities to eager investors — and not-so-enthusiastic consumers.
But not everyone is surprised by the AI craze. James Ferguson, founding partner at the U.K.-based macroeconomic research firm MacroStrategy Partnership, worries that investor enthusiasm for AI is creating a concentrated market bubble reminiscent of the dot-com era.
“Historically, these things end badly,” Ferguson told Bloomberg’s Merrin Somerset Webb. Merrin talks about money Podcast. “So if you’re a little bit older and you’ve seen this kind of thing, you’re more likely to believe that it’s going to end up badly.”
Veteran analysts have argued that hallucinations – the tendency of large language models (LLMs) to fabricate facts, sources, etc. – may prove to be a more intractable problem than initially anticipated, making AI far less viable for use.
“I would argue that AI is still completely unproven, and while fake it till you make it may work in Silicon Valley, I think it suits us as regular people to stay away from AI once you’ve been burned,” he said. “If you can’t trust AI, then in my opinion it’s effectively useless.”
Ferguson also noted that AI may end up being too “energy intensive” for many companies to be a cost-effective tool. As he noted, a recent study by the Amsterdam School of Business and Economics found that by 2027, AI applications alone could consume as much electricity as the Netherlands.
“Forget that Nvidia keeps raising the price of their chips. It’s also getting more and more expensive to run those chips on a server. The result is something that’s very expensive and has yet to prove worth the price in any but the most marginal use cases,” he said.
For investors, especially those enthusiastic about AI, Ferguson warned that the tech hype based on dubious promises is very similar to the period before the dot-com bubble burst, when market returns were concentrated in tech stocks that traded on Wall Street’s overly lofty earnings growth forecasts, he said.
But despite those lofty predictions, dot-com-era hardware giants Cisco and Intel have bitterly disappointed investors ever since, and Ferguson argued that today’s AI hardware hero, Nvidia, could suffer a similar fate, especially given its lofty valuation.
“What multiple of revenue is Nvidia a bargain at if you consider that no matter how phenomenal its growth rate is today, it probably won’t exist in 10 years?” he asked, suggesting that Nvidia may not be worth it at the roughly 40 times revenue investors are paying today.
Ferguson argues that AI tech stocks like Nvidia are overvalued, but acknowledges that no one can predict when the bubble will end. This dynamic has led many bearish investors to feel “forced to trade” in the market even when stocks appear expensive, which can hurt them greatly, he said.
“That’s exactly what happened in the dot-com market. [bubble]”For example, most people, other than retail bettors, looked at this and said, ‘Well, it’s not going to last, but having said that, if it lasts one more quarter and I don’t play, I’m going to lose my job,'” he explained.
The good news, Ferguson said, is that the current stock market bubble is centered around AI-related stocks, so there’s still value in them.
Of course, investors will feel widespread pain if the AI bubble bursts, but after that, Ferguson recommended turning to currently out-of-favor small U.S. stocks that aren’t as expensively valued and could benefit from lower interest rates.
“There’s a lot of value in the U.S.,” he said. “The problem is you have to find it the old-fashioned way – digging through small caps and looking for companies that are growing the old-fashioned, steady way.”
