How should we view the current AI bubble? Bridgewater founder and billionaire Ray Dalio has warned that the AI bubble is about to burst.

On October 7 local time, Dalio said at a forum that a large amount of debt is being used to fund AI investment. "We are now before the breaking point, but we are getting close. I think it is already very close," he said. On asset allocation, Dalio revealed that he is shorting bonds, recommends allocating 5% to 15% of assets to gold, and is looking for investment opportunities outside the United States. However, some Wall Street investors believe that although AI is in a bubble, investors should still enter immediately.

High interest rates raise AI bubble risks

Dalio said he does not doubt the development of AI technology itself. What concerns him is whether the capital investment and asset prices surrounding AI can sustain their current pace.

Tech giants are investing hundreds of billions of dollars in data centers, chips and power facilities. In the early stage, these expenses were mainly supported by companies' own cash flow. As the scale of investment expands, the share of debt instruments such as corporate bonds and project financing has risen, and the AI investment cycle has become more sensitive to interest rates.

This week, the S&P 500 Index and the Nasdaq 100 Index hit record highs again, as tech giants continued to invest hundreds of billions of dollars in building artificial intelligence infrastructure. At the same time, global bond yields have risen to multi-decade highs, with the 10-year U.S. Treasury yield still above 5% and once approaching 5.3%. Fiscal financing needs, economic resilience and global competition for capital have pushed up the price of long-term funds, while AI infrastructure has created a new wave of massive funding demand.

Market concentration is also increasing. The weight of chip, cloud computing, data center and large internet companies in the indices continues to rise, and U.S. stock performance is increasingly dependent on the AI investment cycle.

Dalio believes that the combination of rapidly rising asset prices, capital concentrated in the same investment theme and continuously expanding debt financing is a typical feature of a bubble.

Dalio said the expansion of the AI industry is increasingly reliant on debt financing, and some large companies are already beginning to face constraints in obtaining credit. He also worries that excessive investment by companies competing for an uncertain future market may amplify bubble risks in the technology sector.

Beyond interest rates, Dalio also pointed to another mechanism that could pierce the bubble: wealth monetization.

Dalio said that if investors want to actually use their paper wealth, they must sell assets for cash, and large-scale selling would break the market cycle that depends on rising prices.

"Everyone says 'I'm worth a billion dollars,' but you have to try to spend it," Dalio said. If you want to spend money, you first have to sell assets for cash. "Bubbles are usually pricked at this point."

Investors should enter immediately

At the same forum, another billionaire, Galaxy Digital founder Mike Novogratz, said AI is in "the biggest bubble of our lifetime," but he also advised investors to enter immediately.

Novogratz said: "Bubbles don't end like this. I know how bubbles end - they end spectacularly, and this is not spectacular enough yet." In other words, in his view, the bubble is real, but it has not yet inflated to the extreme.

Novogratz's bullish logic is based on valuation. He believes AI-related stocks are still cheap on a price-to-earnings basis. "If you haven't invested in AI, you might as well go home and bury your head in a bucket of ice water."

On the policy level, Novogratz believes AI does not face a political brake. He pointed out that Washington "will find it very difficult to slow AI down, even to the extent that those building AI themselves would like."

Novogratz mentioned that from labor and copper to power generation equipment, every part of data center construction is rising in price. "There is a lot of inflation in the data center construction industry."

In his view, these investments are also one of the reasons the U.S. economy remains resilient in a high interest rate environment. Data centers do not exist only in tech companies' financial statements; the money ultimately flows into construction, power, equipment and supply chains.

A debt crisis may emerge in the next two years

Dalio also warned about debt risks. Dalio said governments and large tech companies continue to issue debt, while demand from important buyers is weakening. Increased supply coexists with insufficient demand, and this imbalance may turn into a crisis "sometime in the next two years."

"We are clearly in a relatively risky period," Dalio said. He gave a more specific judgment on U.S. debt: a crisis may occur within the next three years, and the United States is approaching the limit of its debt capacity.

Dalio also said rising interest costs are consuming more of the government budget and squeezing room for other spending. The faster debt accumulates, the greater the future repayment pressure. About one-third of U.S. debt financing relies on foreign capital. If foreign demand weakens, it will further push up long-term yields, and the financing environment for AI companies will continue to tighten.

On specific asset allocation, Dalio revealed that he is currently shorting bonds, while believing that an ideal diversified portfolio should allocate 5% to 15% of assets to gold. Among bond assets, he favors inflation-linked bonds, believing this type of asset can help hedge persistent price pressure.