Steve Eisman highlights vulnerabilities in the AI trade, pointing to reliance on a small number of start-ups and the rise of cheaper Chinese models that could reshape the sector’s profitability and competitive landscape.
Steve Eisman has identified what he sees as a fragile point in the artificial intelligence trade: its dependence on a very small number of fast-growing start-ups. Speaking on CNBC’s “Fast Money”, the investor argued that OpenAI and Anthropic now sit at the centre of the AI revenue stream for Microsoft, Amazon, Alphabet’s Google and Oracle, and that the largest technology groups are increasingly exposed to whether those two companies continue to expand.
The scale of that dependence is what worries Eisman most. He said the two start-ups account for a large share of AI-related revenue at the big cloud providers and a meaningful slice of their cloud income overall. That concentration matters because investors have largely treated the build-out of AI infrastructure as a broad, durable growth story, even though the direct profitability of those investments remains difficult to measure, according to Axios. The same report noted that the largest technology groups continue to spend heavily on data centres and related capacity while offering little public disclosure on the returns.
A second risk, in Eisman’s view, comes from cheaper open-weight models developed in China. He said those systems could pressure pricing across the market if they continue to win users, especially as companies search for lower-cost ways to run AI workloads. That concern is increasingly visible elsewhere in the market: reporting from Axios and other technology outlets has shown rising government scrutiny of open models, while industry data cited by CNBC indicates that Chinese models have gained substantial usage on enterprise platforms because they are often 60% to 90% cheaper than leading Western rivals.
The pricing gap is already changing behaviour. CNBC reporting cited in related coverage found that some businesses are adopting Chinese models for routine tasks while reserving more expensive frontier models for heavier work, and that this shift is beginning to show up in corporate spending decisions. The Atlantic recently reported that Chinese systems such as GLM-5.2 are drawing attention in Silicon Valley for combining strong performance with lower costs, adding to fears that a prolonged price war could squeeze margins across the sector. Eisman said he has reduced his exposure to AI, underscoring how quickly sentiment can turn when a boom depends on only a few highly valued customers and a pricing structure under pressure.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
