Bridgewater Founder Ray Dalio: AI Bubble Nearing Burst
Ray Dalio, co-founder of Bridgewater, describes the current state of artificial intelligence as a typical bubble, stating it is nearing a breaking point. His assessment is not based on technology being overvalued, but rather on rising interest rates and holders needing to convert wealth into cash.
On October 7, Bloomberg reported that at the Forbes Global CEO Conference in Singapore, he mentioned that a significant amount of debt is being borrowed to finance AI; the continued rise in interest rates marks the point where the bubble begins to burst. On October 6, he told Bloomberg TV that large-scale cloud providers, which previously relied on equity financing, are now turning to the debt market and are starting to encounter credit thresholds.
He characterized the conditions for a burst as a demand for cash, not a change in perspective. Examples include loan repayments and wealth taxes. In the same interview, he compared the AI craze to the 1929 stock market crash and the 2000 internet bubble, pointing out two pathways: forced asset sales for cash and an influx of new stock supply that dilutes existing holders.
The timeline for the debt crisis varies in different contexts. In the Bloomberg TV interview, he stated that the debt crisis would occur within the next three years, by fall 2029; in a video interview at the Greenwich Economic Forum, he said the supply-demand imbalance would turn into a crisis sometime within the next two years. Both interviews took place on October 6. He noted that AI has become heavily reliant on debt, which could tighten riskier credit like auto loans, with lower-income individuals being squeezed the hardest during this period.
On the buyer side, he is also withdrawing. About one-third of U.S. Treasury bonds are held by foreign investors, with Japan and China holding the majority; he stated that China does not want to continue increasing its holdings for both geopolitical and economic reasons, while Japan, which has lent a lot to the U.S., now wants to reclaim some of it. Some Chinese holdings may be held through other countries, with Belgium's latest report showing U.S. Treasury holdings at $470.7 billion.
His positions include shorting debt and going long on gold. In an ideal portfolio, gold would account for 5% to 15%; Bitcoin would be about 1%, which he does not support, citing that AI models can already breach digital vaults. His family office has moved to Abu Dhabi, and he is also looking at the UAE and Singapore. This is an event-driven allocation, not a new fundraise.
If funds follow his script, they will flow from AI debt and U.S. Treasury demand into gold and cash. Beneficiaries will be cloud providers that can issue debt before interest rates rise and gold holders, while those under pressure will be AI spenders relying on rolling loans to maintain capital expenditures and U.S. Treasuries dependent on continued purchases from China and Japan. For him, the bubble bursting equates to a repayment date, not a sudden failure of models.
Source: Public Information
ABAB AI Insight
Dalio's framework predates this conference. For years, he has referred to rising debt as a blight eroding the system, calling the critical point a heart attack; Bridgewater sells research based on economic machines and debt cycles, not on single tech bets. On October 6, 2026, in Singapore, he applied the same framework to AI: large-scale firms shifting from equity to debt, China and Japan reducing demand for U.S. Treasuries, with about one-third of U.S. debt relying on foreign buyers. The family office's move to Abu Dhabi indicates a shift in residence and allocation out of the original buyer circle.
The resource mobilization is characterized by empty debt and more gold. At the Greenwich Forum, he stated that the ideal portfolio includes 5% to 15% gold, with Bitcoin only around 1%, reasoning that models can breach digital vaults. The motivation is not to be bearish on chips but to be bearish on the order of financing. Debt supply comes from governments and large tech, while demand comes from institutions in China and Japan; when both sides reverse, he seeks assets that do not rely on rollover. Wealth taxes and loan repayments are categorized as the same type of event: holders are forced to convert paper wealth into cash.
The analogy is his own mention of 1929 and 2000, with mechanisms closer to wholesale financing before 2008. When the 2000 bubble burst, many internet companies had not yet stabilized cash flows and relied on equity financing; this time, he says cloud providers already have market positions but are starting to incur debt for capital expenditures. The industry is in a control struggle at the end of an expansion phase: whoever locks in spending with long-term debt first secures computing power and faces refinancing thresholds first.
This represents a transfer of pricing power. The model's capability has not failed in his statements; what has failed is the low price of equity financing. Rising interest rates return debt pricing power to lenders, while wealth realization returns asset pricing power to those needing cash. New stock supply further returns pricing power to issuers, diluting the demand that previously inflated prices. Thus, he focuses not on press releases but on repayment dates, tax rates, and buying from China and Japan. In his framework, the bubble bursts because cash is more expensive than the story.
ABAB News · Law of Cognition
Bubbles do not end because the story ends, but because cash is demanded back.
Equity speaks of growth, while debt speaks of maturity.
When buyers exit, issuers discover the price.