How much wealth would be destroyed by an AI stockmarket crash? (www.economist.com)

A new visual analysis warns that an AI-driven equity bust on the scale of the dotcom crash could erase roughly 8% of American household wealth. Key indicators underpinning that risk: S&P 500 market capitalization reached about 177% of US GDP in Q2 (levels comparable to past bubbles), US households now have roughly 21% of their assets in stocks, and the market is unusually top-heavy — the 20 largest AI-related companies account for roughly 52% of the S&P 500’s value versus 39% at the dotcom peak. The technical danger is concentration and correlation. Largest names (Nvidia ~8.2%, Apple ~6.6%, Microsoft ~6.5% of the index) dominate passive and active portfolios, so a shock to AI expectations would propagate quickly through index funds, pensions and foreign investor positions. High market-cap-to-GDP, elevated retail and institutional stock exposure, and clustered valuation in a small set of AI winners raise systemic tail risk: rapid de-risking could trigger steep price declines, liquidity stress, and real wealth effects for consumption and retirement funding. For investors and policymakers the takeaway is clearer stress-testing, diversification away from mega-cap AI exposure, and monitoring leverage and liquidity in ETF and margin markets.