Famed economist Mohamed El-Erian delivered a stark warning about the evolving global economy, stating that while the underlying system remains intact, investors should brace for significant individual losses within the Artificial Intelligence (AI) sector and expect numerous “credit accidents”.
Speaking at Yahoo! Finance Invest, El-Erian framed the current environment as one in which “cockroaches” abound but “termites” do not. This differentiation is fundamental: cockroaches are unpleasant accidents that “come in groups” but do not “eat away at the integrity of the system.” Termites, conversely, erode the foundation.
While systemic shock is unlikely, the President of Queens’ College, Cambridge University, and Chief Economic Advisor at Allianz said he expects economic and credit accidents because market participants have “stretched really far for additional returns.” This has been encouraged by loose financial conditions and a strong economy, he added, and some investors seem to have gone “beyond their comfort zone and beyond their ability to do due diligence.”
The rational AI bubble
El-Erian told Yahoo that he had, in collaboration with Nobel Laureate Mike Spence, assessed the AI boom and concluded that the market is experiencing a “rational bubble.” While the aggregate value being created is significant, making it rational for investors to take a venture capital approach and “overinvest” due to the large payoff, there is a darker side: “there will be tears” and losses.
He said elements of this bubble mirror past speculative periods, such as the dot-com era, where companies applied a label—now “AI”—to their operations to attract capital. Further contributing to the bubble elements is the fact that foundational model companies are attracting significant investment, yet “not all of them are going to succeed.”
A key concern for El-Erian is the inadequate focus on diffusion—the process of getting AI into the workplace in a comprehensive and orderly manner. The U.S. currently lacks a comprehensive diffusion policy, unlike countries such as China and the UAE. If diffusion is not handled correctly, he added, the full promise of AI won’t be realized.
Regarding corporate adoption, El-Erian noted his concern about the prevailing corporate mindset, which currently views AI primarily as a “cost minimizer.” The true potential of AI, he argued, lies in labor enhancement and serving as a “productivity enabler.” If the U.S. gets diffusion right, the resulting significant productivity increase could allow monetary policy to be looser than it would otherwise have been.
Pressure on the K-Shaped Economy
Beyond the financial accidents, El-Erian cited two major issues that could pose pressure: the need to refinance a large amount of debt at higher interest rates and the significant pressure on the lower end of the income distribution.
This focus highlights concerns about the bottom of the K-shaped economy. He said lower-income consumers are “near recession,” grappling with affordability concerns—an issue that is social and political, not just economic—and high debt, including maxed-out credit cards. Furthermore, insecurity about future income, driven in part by surging layoffs reported Challenger, Gray & Christmas and the impending workplace changes brought by AI, compounds their distress.
El-Erian cautioned that this pressure is not isolated: lower household incomes may be forced to stop spending because they are unable to, and this “will contaminate upwards for the economy as a whole.” While the upper class is generally doing well on both income and wealth measures, they are not immune to the difficulties faced by low-income households.
El-Erian urged policymakers to recognize that the future will be determined by the “tails of distribution, not in the belly.” In today’s structurally changing, fragmented world, leaders must realize they are operating in a multimodal world and should not be deceived by the assumption of a normal, bell-shaped distribution.
For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.
