AI Infrastructure Bubble More Severe Than Dot-Com Boom, Says Legendary Short-Seller Jim Chanos

Legendary short-seller Jim Chanos, founder of Chanos & Co., recently warned that the current artificial intelligence infrastructure investment boom poses systemic risks comparable to—and potentially more severe than—the late 1990s dot-com bubble. According to Chanos, while capital inflows have reached historic scale, the economic fundamentals supporting this wave remain unproven.

Chanos highlighted that mega-cap cloud companies' incremental capital returns have declined from approximately 40% about 18 months ago to roughly 20% today, with further deterioration possible if spending growth persists. He projects this fundamental mismatch will force executives to make critical strategic decisions within 12 to 18 months, with a reckoning expected by late 2026 to 2027. The core issue, Chanos argues, is a duration mismatch: companies are committing billions in near-term capital expenditures to finance 20-year asset strategies, while large unrealized GPU and data center investments remain undeployed and off depreciation schedules, obscuring true economic performance.