Investment expert David Belle has dismissed predictions of an AI bubble, arguing that today's boom differs fundamentally from the dot-com mania of the late 1990s. Belle, founder and trader at Fink Money, said doom-mongers overlook that companies driving the current rally are generating substantial profits.
Profits keep turning up
Belle said: "The 'this is 1999 all over again' crowd have been predicting disaster for years, which must get exhausting eventually. The problem for the doomers is that the profits keep turning up. In 1999, share prices ran away while earnings didn't. This time, in several of the biggest companies, profits have grown as fast as the share price or faster. You can absolutely argue that AI stocks face risks. But if you're going to call something a bubble, it helps if you occasionally look at what the companies actually earn."
Nvidia, often the focus of AI sceptics, provides one of the strongest counterarguments, according to Belle. He noted that investors are paying roughly $34 for every $1 the chipmaker earned over the past year and about $22 for every $1 it is expected to earn next year. During the AI boom, that figure has been above $200.
Valuations compared with dot-com peak
Belle said: "Nvidia's share price went through the roof, but its profits went up even faster. That's rather awkward if your entire argument is that investors have stopped caring about profits. The stocks got more expensive in dollars. They got cheaper relative to the profits they produce."
Unlike many dot-com era companies, today's biggest AI beneficiaries are established businesses earning billions in cash. Belle said: "In 1999 you were paying a fortune for profits that might turn up one day. In 2026 you're paying a high-but-normal price for profits that have already arrived. The S&P 500 is around $20 to $21 for every $1 of expected earnings next year and technology is around $22. Around the dot-com peak, the Nasdaq was often north of $70."
Japan comparison misleading
Belle also addressed comparisons with Japan in the late 1980s, when keiretsu corporate groups held significant stakes in one another. Banks, insurers and companies held roughly two-thirds to 70% of the Japanese stock market between them. Belle said today's AI market is different, with a much smaller overlap concentrated among a handful of companies alongside genuine sales to customers.
He acknowledged legitimate reasons for caution: a few tech giants represent a large market share, huge sums are being spent on data centres and AI infrastructure, and some headline profits at companies like Google and Amazon include paper gains on investments. Belle added: "There is a serious bearish argument here. The funny thing is that it isn't the one the doomers keep making. The risk isn't that these companies have no profits. They plainly do. The risk is that they spend so much money building AI infrastructure that future demand doesn't justify it."
Belle concluded: "AI shares can fall. Nvidia can fall. Markets can have corrections and investors should never assume prices only go upwards. But 'stocks can fall' is not exactly a groundbreaking prediction. The dot-com bubble was a market that left earnings behind. This market has been chasing them, and so far the earnings have been winning. I'm afraid anyone waiting desperately for 1999 to happen again may need to remain miserable for a little longer."
