TS Lombard Warns AI Bubble Could Burst in Early 2028, Sending Tech Stocks Down 40%

TS Lombard strategist Dario Perkins outlined a scenario in an October 22 report in which the AI investment boom ends in early 2028, cutting U.S. tech stocks by 40%, the S&P 500 by 30% and erasing roughly $20 trillion in market value. The firm expects a mild recession rather than a 2008-style credit crisis. Tech companies spend about $1 trillion a year on AI infrastructure, with hyperscaler capex near a record share of U.S. GDP. Revenue growth still depends largely on the capex cycle; OpenAI and Anthropic draw about 80% of revenue from 1% of customers, mostly loss-making VC-backed startups. Reaching adequate returns would require $2.5 trillion to $3 trillion in annual AI revenue by 2030, about 14 times current levels. The report cites six bubble risks: revenue shortfalls, credit strain (an estimated $700 billion in new AI debt in 2027), depreciation and lease costs, IPO disclosures, semiconductor cycle whiplash and regulation. AI capex need not fall—only slow—to trigger stress. The scenario assumes a 35% cumulative default rate in 2028-2030 and $125 billion to $190 billion in credit losses, with equity holders absorbing the bulk of the damage. TS Lombard advises against shorting U.S. tech now, favoring diversified global equity exposure.