(Yicai) Nov. 19 -- The “All in AI” strategy adopted by American tech giants is essentially an unsustainable capital game, and if the funding chain tightens, it could trigger systemic risks.

For the past three to four weeks, the stock prices of two major tech giants, Oracle Corporation and Meta Platforms, have been steadily declining, indicating that the market has begun to reject the model of internal order circulation among these firms -- in which Company A buys products from Company B, Company B buys from Company C, and Company C then invests in Company A’s equity, forming a seemingly perpetual capital loop.

This boom in AI infrastructure, fueled by rapidly expanding debt, is fundamentally an unsustainable capital game. Once the funding chain tightens, the pressure will first hit the downstream segments of the AI industry, as data center construction becomes even more cash-intensive. In contrast, upstream players such as chip developer Nvidia and tech giants like Google, which hold competitive advantages in both software and hardware, can remain in their comfort zones thanks to their monopolistic positions.

But this “storm in a teacup” will inevitably spill over. The modern economy essentially functions as a financial accelerator: when AI development is in a positive feedback phase, capital forms an upward spiral of “investment -- growth -- expansion.” However, once it enters a negative feedback phase, that spiral will deteriorate rapidly and become extremely difficult to curb.

At present, nearly all the pressure to sustain this spiral rests on US authorities. If the Federal Reserve maintains high interest rates and refrains from cutting them, and if the Donald Trump administration is unwilling to let fiscal policy absorb the cost of AI infrastructure investments, then a tightening funding chain will likely cause credit default swap premiums to surge (higher premiums indicate rising expectations of debt default). Companies would then face difficulties securing funds or sharply rising financing costs, ultimately stalling growth.

The direct result of this cascade is a steep drop in stock market capitalization. Of the current USD68 trillion total market cap of the US stock market, as much as USD30 trillion comes from companies in the AI industry chain, meaning that even minor disturbances could trigger systemic risk.

Short sellers in the capital markets have already begun acting. Their preferred targets are companies with tight cash flows, high debt levels, and an unrelenting need to invest. Recently, we have observed sharp increases in credit default swap premiums for Oracle, Meta, and Nvidia-backed cloud services provider CoreWeave.

In summary, the US’s “All in AI” strategy is facing three major risks: first, the fragile ecosystem of internal order circulation among major players; second, the contradiction between soaring computing power demand and aging power infrastructure; and third, unemployment and declining incomes caused by intelligent automation replacing human labor.

If these risks keep compounding, Murphy’s Law -- the idea that potential failures tend to occur -- will likely apply.

(The article’s author, Liu Yuhui, is the deputy director of the Shanghai Chief Economist Financial Development Center and a council member of the China Chief Economist Forum.)

Editors: Dou Shicong, Emmi Laine