Approaching That: What Ray Dalio's AI Bubble Warning Gets Right About the Debt

Ray Dalio says markets are approaching the point where the AI bubble bursts, not because sentiment on AI will sour, but because debt-funded spending and forced liquidation will do the pricking. The numbers behind hyperscaler capex and bond issuance already back up the mechanism he's describing.

Ray Dalio told the Forbes Global CEO Conference in Singapore on October 7 that markets are nearing the point where the AI bubble bursts. "We're in the part of the cycle that is before that but approaching that. I think we're close to that," he said. What makes the comment worth more than a headline is the mechanism he described, which is not the one most bubble warnings reach for. Dalio is not predicting that investors will suddenly decide AI is worth less. He is describing a plumbing problem: too much of the money behind the buildout is borrowed, rates are climbing, and paper wealth eventually has to be converted into cash to be spent, a conversion that forces selling regardless of anyone's opinion about the technology itself.

The Bursting Mechanism Isn't a Change of Heart

Dalio's second point is the more specific one, and the one allocators tend to skip past. "Everybody says 'I'm worth a billion dollars' but OK, try to spend that. In order to spend that you have to sell wealth in order to get money, and so the bubble usually pricks at that," he said. Wealth taxes, margin calls, and the ordinary need to realize gains to fund spending all force the same action: converting an unrealized, paper valuation into an actual sale. A bubble built on borrowed money is especially exposed to this, because rising rates increase the carrying cost of the debt at the same time they make the future cash flows the debt was supposed to be repaid from look less certain. Dalio's framing does not require AI to stop being useful or even to stop growing. It only requires the financing underneath it to come due faster than the cash it generates.

The Capex and Debt Numbers Behind the Warning

The scale of the borrowing Dalio is pointing to is no longer a marginal detail. JPMorgan's Jamie Dimon put hyperscaler AI infrastructure spending at roughly $700 billion in 2026, more than double the roughly $300 billion spent the year before, with outlays on track to cross $1 trillion in 2027. Morgan Stanley forecasts global AI-related debt issuance will more than double to nearly $570 billion in 2026. Hyperscalers alone issued $219 billion of investment-grade bonds in the first half of the year, including $62 billion in non-dollar currencies, which on its own already exceeds the $121 billion those same companies issued across all of 2025. Spending that used to come out of free cash flow is now increasingly coming out of the bond market, at a pace that keeps revising upward every time these companies report.

The Same Concentration Problem Dalio Is Pointing At, in Equities Too

The equity side of the same story reached its own milestone this month. Nvidia, Apple and Microsoft together now account for more than 21% of the S&P 500, according to Creative Planning data, the highest concentration ever held by three stocks in the index's history. For scale, IBM, AT&T and Exxon Mobil together peaked at 13.4% of the index in the mid-1980s, less than two thirds of today's reading from just three names. A forced-selling event of the kind Dalio describes does not need every AI-adjacent asset to be fairly valued or overvalued uniformly. It only needs enough of the market's weight concentrated in one trade that a liquidity event in that trade moves everything else with it, which is precisely what this level of index concentration guarantees.

Why This Matters Beyond a Single Billionaire's Opinion

Dalio's specific timeline is a guess, and his track record on calling bubble peaks is mixed like anyone else's. What is not a guess is that the mechanism he is describing, debt that has to be serviced regardless of sentiment, is already visible and already large, covered in more detail in an earlier piece on this page about the hyperscaler debt load itself. A portfolio that has priced AI exposure as a single story about whether the technology delivers has not necessarily priced the separate, more mechanical question Dalio is raising: whether the financing structure underneath that story can be serviced smoothly, or whether it forces a round of selling that has nothing to do with whether AI works.

The Question Worth Sitting With

Dalio is not arguing that AI is fake. He built part of his career on exactly this kind of distinction, that a technology can be real and a financing structure around it can still break. The debt figures behind hyperscaler capex and the concentration figures behind the S&P 500 both describe the same underlying condition from different angles: a very large, very fast-growing trade with a narrow base of names and lenders underneath it. Whether that structure gives way on Dalio's timeline or a different one, the honest question for a portfolio is not whether AI is a bubble. It is whether the portfolio is exposed to the specific mechanism, forced selling to meet debt service or cash needs, that actually pricks one.