Billionaire investor Ray Dalio warned on October 7 that the artificial intelligence boom has become a “classic bubble” nearing a bursting point, driven by rising interest rates and a heavy reliance on debt. Speaking at the Forbes Global CEO Conference in Singapore, the Bridgewater Associates founder said tech companies are increasingly borrowing to fund AI infrastructure while bond yields have climbed to levels not seen in decades. “We’re in that part of the cycle that is before that, but approaching that,” Dalio said. “I think we’re close to that.”
Fellow panelist and Franklin Templeton CEO Jenny Johnson added that complex financing structures and off-balance-sheet deals make it harder to track the true web of corporate liabilities tied to AI spending. Temasek Holdings CEO Dilhan Pillay noted that physical AI remains important but warned too much capital flowing into AI infrastructure could raise the overall cost of capital.
Separately, BitMEX co-founder Arthur Hayes outlined a scenario in which the multitrillion-dollar AI data-center buildout could create credit stress, a crash, and eventually a liquidity-driven rescue. Estimates cited in the report put US AI infrastructure costs between $2.8 trillion by 2030 and $10.3 trillion by 2032, with at least $1.3 trillion in debt already raised. Hayes argues that providers may be saddled with commitments to pay for reserved compute capacity, while Columbia economist Stijn van Nieuwerburgh estimated that a 10% return would require $3.7 trillion in annual AI revenue by 2032.
Bitcoin slid toward $83,800, with $403.58 million in leveraged long liquidations recorded in one hour, underlining the near-term risk-off pressure. Hayes said a credit shock could initially trigger falling prices and further liquidation pressure before any bailout materializes. He sees Bitcoin as a potential long-term beneficiary if excess liquidity flows into crypto, though that depends on both financial stress and a policy response that expands liquidity.
Adding to the debate, a BlackRock report noted that AI agents may choose to save in Bitcoin for long-term value preservation, while stablecoins could serve as transactional money. For now, the macro framework remains contingent: an AI-credit unwind could hurt risk assets first and only later boost Bitcoin if it forces a liquidity-heavy rescue.
