AI Slowdown Could Trigger $1.2T Capex Crunch, Warns IMF and Fitch
What happened: Calls to slow the development of advanced AI models are colliding with an $800 billion investment boom among U.
What happened: Calls to slow the development of advanced AI models are colliding with an $800 billion investment boom among U.S. hyperscalers in 2026, with global AI capex projected to reach $1.2 trillion in 2027. Goldman Sachs estimates these firms need $300 billion in annual AI revenue to break even, leaving a $230 billion gap at current run rates. The IMF warns that an AI investment reversal could shave 1.5% off U.S. GDP and 1.2% off global output, while Fitch sees a 35% equity shock and possible U.S. recession if the AI buildout stalls.
Why it matters: The debate over AI’s pace pits economic risk against safety concerns. Proponents of a slowdown (including Dario Amodei, Sam Altman, and several policymakers) argue for “pacing the frontier,” while opponents (Trump, Mark Zuckerberg, Jensen Huang) warn against stifling a key growth engine. The risk of overbuild—AI infrastructure at 1.5x socially efficient levels per BIS—suggests that a sudden capex retrenchment could have outsized effects on both tech and broader equity markets.
Source: Cointelegraph