Is the AI Boom Becoming a Bubble? Motley Fool Hosts Weigh Risks, Positioning, and Google's Nuclear Deal

On Oct. 7, 2026, three Motley Fool contributors sat down for an episode of the Hidden Gems Investing podcast and took up a question the host, Travis Hoium, admitted he usually avoids: is the AI boom turning into a bubble? His reasoning was that you can ask it every day, and one day you will be right, but most days you will be wrong.

The prompt was a recent interview with Ray Dalio, who has said a burst could come relatively soon, with higher interest rates as one possible catalyst. Hoium said his own worry is the debt. He noted that several trillion dollars are going into AI spending, that few sustainable business models have emerged so far, and that nearly every company, hyperscalers included, keeps borrowing more. He pointed to a report from the previous day that SpaceX was seeking another $40 billion in debt.

Jon Quast urged caution about reading too much into Dalio. He called him a very smart person, but one who was warning about market bubbles before AI was a factor. Quast said Dalio did not claim a burst is close. His claim, as Quast understood it, was that the market is one step before being close to a point where a bubble could burst. Quast summed it up with a line borrowed from Aragorn in The Lord of the Rings: it may come one day, but not this day.

Lou Whiteman agreed with the substance and questioned the timing. He said Dalio could have made the same statement six months or a year ago, and in fact did, and could make it again six months from now, so he does not treat it as a signal. He thinks of markets as a pendulum that will not swing one way forever. That does not mean avoiding AI or shorting it, he said, but it does mean sizing positions accordingly. He also pushed back on debt as the trigger for now, calling it manageable until it is not, and saying that moment has not arrived.

On positioning, Hoium recalled holding JDS Uniphase in the dot-com era and watching it become a disaster. He asked whether investors want to be buying the hottest names without knowing if it is 1997 or early 2000. Whiteman said he finds few individual AI stocks appealing after the run and sees better pricing elsewhere, citing banks as ridiculously cheap. He also holds many index funds. Because AI makes up so much of the S&P 500, he gets oversized exposure if the trade keeps working, so he is letting that ride.

Quast then laid out the three catalysts Dalio cites. A bubble needs a pin, and those pins are rising interest rates, stock issuance, and a sudden need to convert wealth into cash. Quast used Elon Musk as an illustration: a trillionaire on paper, but with wealth tied up in assets. Hoium added mass retirements as another way money could leave the market. Dalio himself highlights a potential wealth tax, which Quast said would not arrive until the next election cycle at the earliest. Quast said he sees little to fear if this is a bubble, observing that Micron now earns more operating profit than Apple, which he considers unlike the dot-com names at the last peak.

The conversation then turned to energy. Google announced a 20-year power purchase agreement with Constellation Energy that adds 890 megawatts of nuclear capacity, backed by $4.3 billion of investment. Quast likes the approach of upgrading existing plants for speed. But he cited BloombergNEF's estimate that new data centers may need about 100 gigawatts by 2035, a figure he said he could not verify. The Google deal would be under 1% of that, and the contracted power is not expected until the end of 2032.

Whiteman said the retrofit approach is both the bull and bear case for nuclear, since the easy existing capacity is running out and new construction will take years. Hoium noted that Microsoft's Three Mile Island deal was among the first of its kind. The discussion was cut off as Hoium raised the 15-year Vogtle expansion. The panel also planned to cover Apple's smart-home push, though the excerpt ends before that segment.