Steve Eisman, the investor immortalized in “The Big Short” for his prescient bet against subprime mortgages, has a new concern. This time it’s not collateralized debt obligations. It’s the entire foundation of Big Tech’s AI revenue engine.
In an appearance on CNBC’s “Fast Money,” Eisman laid out a blunt thesis: OpenAI and Anthropic together account for roughly 70% of AI-related revenue flowing to the major hyperscalers, including Microsoft, Amazon, Alphabet, and Oracle. That level of concentration, in Eisman’s view, is the “Achilles’ heel” of the AI trade.
Two companies, one massive dependency
That AI-driven revenue represents approximately 25-35% of these companies’ total cloud revenue. Cloud has been the growth story propping up premium valuations across the sector for years.
Microsoft, Amazon, Google, and Oracle have collectively poured hundreds of billions into AI infrastructure. The return on that investment is, to a remarkable degree, dependent on the continued spending and success of just two private AI labs.
The China variable
Eisman added another layer to his warning: cheaper Chinese open-weight AI models that could spark a price war across the industry.
Open-weight models, which make their parameters publicly available for others to build on, have been gaining traction as viable alternatives to the proprietary systems developed by OpenAI and Anthropic. Chinese labs have been particularly aggressive in releasing competitive models at significantly lower price points.
If those models reach a quality threshold where enterprise customers start switching, or even just using them as leverage in pricing negotiations, the margin compression could be severe. The hyperscalers have justified their enormous capital expenditures on the assumption that AI workloads will generate healthy returns over time. A price war would blow a hole in that assumption.
A pattern of skepticism
Eisman’s latest comments aren’t an isolated burst of pessimism. Earlier in July 2026, he flagged potential diminishing returns in the AI sector and disclosed that he had personally reduced his exposure to AI-related investments.
His skepticism sits within a broader debate that has intensified throughout 2026. Tech companies have been racing to announce ever-larger AI infrastructure budgets, with capital expenditure figures climbing quarter after quarter.