Jim Rickards says AI bubble is 17 times larger than dotcom era

Rickards says debt financing behind AI data-center expansion resembles the structured-credit playbook used before the 2008 housing crisis and may already sit in retirement portfolios.

Summary

Jim Rickards has expanded his warning on the AI boom, arguing not only that it resembles the dotcom bubble but that its financing structure also echoes the debt engineering behind the 2008 housing crash. In a new presentation, he says the buildout of AI data centers is being funded with large amounts of debt that is structured like real-estate deals, then bundled into bonds in a way he compares to the CDO model associated with subprime mortgages. Rickards says AI requires enormous physical infrastructure, citing an estimated $5 trillion in U.S. data-center spending, while the companies involved are already burning cash. In his telling, outside investors fund the projects, AI companies pay rent, and the resulting obligations are packaged and sold on to investors. He argues that makes AI not just a technology story but also a credit and real-estate risk. He says the danger may extend well beyond investors in AI stocks because the debt is, in his view, being distributed into pension funds, retirement accounts and 401Ks. That adds to his earlier comparison with the late-1990s internet boom, when heavy debt-backed infrastructure spending and weak underlying economics preceded a deep Nasdaq decline. The presentation says investors should understand how the debt behind AI is being packaged, assess their exposure and consider steps to protect themselves. It is available online for free.

Terms & Concepts
  • CDO: A collateralized debt obligation, a structured finance product that pools debt and repackages it into securities sold to investors.
  • tranches: Slices of a bundled debt product that carry different levels of risk, return and repayment priority.
  • subprime mortgages: Home loans made to weaker borrowers that became central to the 2008 financial crisis when many soured.