Former White House Economist Furman Warns AI Bubble Could Cut S&P 500 by 25%
Jason Furman, former chair of the U.S. President’s Council of Economic Advisers, is warning that markets may be badly underestimating the damage from a burst AI bubble. The trade has gone from Wall Street’s favorite theme to a portfolio risk that demands reassessment. A selloff in AI-linked stocks has fund managers rethinking their exposure. Furman, now a professor at Harvard University’s Kennedy School of Government, estimates in a Sunday article that a near-term AI bubble burst could send the S&P 500 (SPX) down 25% in his baseline case. A more pessimistic case calls for a 35% drop; an extreme, less likely case, 50%. Under the baseline, U.S. households would lose about $19 trillion from direct stock holdings and indirect exposure through mutual funds and retirement accounts. Even a 25% decline would deal a heavy blow to household wealth, he says.
Valuation corrections could magnify the move. Furman used several methods to size the shock. If the largest AI stocks by market value fell 50%, the broader market would drop about 11%, before any spillover to other industries. Add further valuation adjustments and cuts to earnings estimates, and the total decline could reach 18% to 41%. He says those outcomes far exceed the risk now reflected in options. Cboe Global Markets closing data last Friday implied roughly a 19% probability that the S&P 500 falls 10% by December next year, and only a 2% probability of a 50% decline.
Furman stops short of declaring an AI bubble. Smartphones and jet engines did not end as bubbles, he notes, and AI could still deliver long-term productivity gains. He also sees key differences from the period before the dot-com crash: Alphabet (GOOGL.O) and NVIDIA (NVDA.O) have strong AI-linked revenue growth, and chip demand and pricing are not comparable to the fiber-optic buildout that left vast networks underused.
AI risk has moved beyond a handful of tech stocks. Reuters reported that fund managers are auditing portfolios for direct and indirect AI exposure as they reassess the trade. According to the MSCI Global AI Index, at least $25.3 trillion is invested in the world’s largest companies most directly tied to AI. The indirect risks and opportunities from AI-driven disruption are harder to measure. Nomura Asset Management Chief Investment Strategist Andy Goldberg points to David’s Bridal, the bridal retailer that plans to spin off and list its AI wedding-planning business next year; CEO Kelly Cook says the company has long been an early adopter of AI in retail. Goldberg also cites trucking companies using AI to optimize logistics, and municipal bonds whose revenues and credit ratings could be affected by data-center construction.
Goldberg says no one fully understands how much AI risk they are taking on, calling it a massive one-way trade. A 1% to 2% cut in AI allocations could spark a sharp selloff, he warns.
The risk review has reached fixed income. PGIM, Prudential Financial’s investment arm, manages nearly $1 trillion in fixed-income assets. This year it began applying two AI-driven screens to bonds, loans and structured securities. One assesses how AI affects corporate business models and creditworthiness. The other models slower AI-related capital spending and the spread of a market crisis to non-traditional AI companies. Brian Barnhurst, PGIM’s head of global credit research, says investors readily spot direct exposure in data centers and chips. The harder question is how AI will hit health care, industrial and telecom companies.
Pension holdings are also becoming more concentrated in AI assets. The Equable Institute, a nonprofit pension researcher, says that as of June 2026, stocks of 52 publicly traded AI-related companies made up 38.5% of public equity holdings at certain large pension funds, up from 20.7% in early 2010. Those funds have about 9% to 11% of assets directly invested in AI-related companies, excluding bonds, venture capital and other private equity.
BlackRock iShares U.S. Equity ETFs head Jay Jacobs says the rising weight of AI-linked core holdings in major indices is deepening market concentration. That makes diversification harder than investors once assumed.
Investors are starting to pick AI winners, and volatility may persist. The sharp swings have pushed some investors toward more cautious strategies. BCA Research Chief Economist Peter Berezin sees the fierce selling in some AI-themed stocks as an early signal that investors need to scrutinize real-world AI adoption. AI may lift corporate productivity without delivering a matching gain in profits, he says. Powell Industries (POWL.O), which makes power-management systems for data centers, shows the valuation swings: its stock rose from about $8 four years ago to a peak of $328 this year, then retreated to $146. Berezin says data-center construction bottlenecks may eventually ease, eroding the excess profit enjoyed by suppliers such as Powell, though the timing is uncertain.
Policy and industry shifts add another layer. Janus Henderson technology analyst Shaon Baqui says market worries have built since AI valuations peaked in June. Democratic gains in the U.S. midterm elections have deepened doubts about the AI sector’s growth environment. He expects investors to focus more on separating truly competitive companies from speculative stocks with weak profit fundamentals.
Rob Arnott, founder of Syzygy Asset Management, argues that too much emphasis on overvaluation risk could blind investors to other opportunities. Capital flows are concentrated in a few AI beneficiaries, he says, while companies that can create value through AI but have not yet won market recognition may be overlooked.