How would an AI investment bust affect recession and recovery?
Core argument: Paul Krugman cites Bernstein and Cummings’ estimate that an AI bust would deliver a negative shock three times larger than the dot-com bust, noting AI-related tech spending and stock market valuations as shares of GDP exceed 1990s boom levels.
The Nixon, Ford, and Volcker recessions resulted from deliberate Fed tightening to curb high inflation, pushing interest rates to very high levels. Since the 1980s, however, inflation has been much less of a problem. Why, then, have recessions continued? Because of private-sector overreach: overenthusiastic companies and lenders taking excessive risks. The 1990–91 recession followed overinvestment in commercial real estate, fueled by runaway lending by newly (and disastrously) deregulated S&Ls. The 2001 recession followed the collapse of the 1990s tech bubble. The 2007–09 recession, the worst since the Depression, followed the bursting of a housing bubble fueled by effectively unregulated “shadow banks.” If another recession comes soon, AI-industry overreach may be responsible. Tight-money recessions are easy to end: the Fed can simply lower rates and loosen credit. But after a burst bubble, lower rates may not revive borrowing and spending, even before rates hit zero [though in two of the three busts rates did indeed hit zero]. Fiscal stimulus is a heavier lift.

