AI Giants Have 'No Moats,' Says Big Short Investor Eisman

Steve Eisman made headlines this week with a stark warning about artificial intelligence spending on his podcast, The Real Eisman Playbook. "You're talking about massive companies spending trillions of dollars for something that may have no moats, and that's not a recipe for longevity," he said. Eisman's core argument is that the world's largest companies—presumably referring to Microsoft, Alphabet, Meta, and others—are pouring vast sums into AI infrastructure and development without building defensible competitive advantages.

The evidence Eisman cites is straightforward: customers switch freely between ChatGPT (OpenAI), Gemini (Alphabet), and Claude (Anthropic), meaning no single AI developer can lock in users or charge premium prices. This lack of pricing power undermines the business case for trillion-dollar capital expenditures. To illustrate the problem, Eisman drew a parallel to the airline industry, another sector that demands enormous capital investment but offers little pricing power. He concluded that investors would be better served owning the suppliers—companies like Nvidia that sell chips and infrastructure to AI developers—rather than the developers themselves, a strategy that proved successful during the dot-com era for companies like Cisco Systems.

Eisman's guest on the podcast was Torsten Slok, Apollo's chief economist, who provided context on AI's macroeconomic footprint. Slok estimated that AI spending now accounts for roughly one percentage point of U.S. GDP growth, representing nearly half of the nation's economic growth this year. Slok also raised broader financial concerns, warning of a $500 billion（約80兆円） maturity wall hitting the software sector in 2028 and 2029. He noted that the software industry carries heavy debt loads and weak coverage ratios even before AI disruption, painting a picture of a sector financially vulnerable to the very disruption it is trying to navigate.

Eisman also singled out Oracle for specific criticism, highlighting concentration risk tied to OpenAI. Roughly half of Oracle's reported $600 billion（約96兆円） backlog depends on OpenAI contracts. "It's a little scary what's going on," Eisman said. Slok added that the traditional 60/40 portfolio—60 percent stocks, 40 percent bonds—no longer provides diversification. Mega-cap stocks like Microsoft and Apple dominate the equity side, while hyperscalers including Microsoft are flooding the bond market with $700 billion（約110兆円） in investment-grade debt this year, meaning both halves of the classic portfolio ride the same AI theme. On interest rates, Slok stated there is likely zero chance of a rate cut this year given inflation near 3.5 percent, a view broadly shared by Polymarket traders, who price zero rate cuts at 81 percent. Despite Eisman's warnings, traders appear less convinced by his bear case for now.

Steve Eisman's critique cuts to a central debate in AI valuation: whether the massive capital expenditure by tech giants will generate sustainable returns. His argument rests on a simple observation supported by market behavior—users switch freely between competing AI chatbots from OpenAI, Alphabet, and Anthropic—which means these companies cannot lock in customers or command premium pricing. This is a direct parallel to airlines, where high capital requirements do not translate into pricing power or durable profits. By recommending suppliers like Nvidia over AI developers, Eisman echoes a strategy that worked during the dot-com era, when picks-and-shovels companies outperformed the businesses they served.

Eisman's concern about moats is sharpened by broader financial vulnerabilities Apollo's chief economist Torsten Slok identified. Slok estimated that AI spending now accounts for roughly one percentage point of U.S. GDP growth, nearly half of this year's total, underscoring the sector's size and systemic importance. However, the software industry—which depends heavily on AI—faces a $500 billion（約80兆円） maturity wall in 2028 and 2029 while already carrying heavy debt loads and weak coverage ratios. Meanwhile, the classic 60/40 portfolio has lost its diversification power; both equity and bond allocations are now dominated by mega-cap stocks and hyperscaler debt, all riding the same AI theme. These structural vulnerabilities suggest that Eisman's skepticism about AI's longevity without moats may resonate with credit markets and policymakers even if equity traders remain unconvinced for now.