AI Stocks Trading at “Ridiculous Valuations” May Pose Bigger Threat Than Iran, Economist Warns

Tech and AI-related stocks now account for roughly 45 percent of the S&P 500’s total market value, up from about 25 percent three years ago. At the same time, a key valuation measure called the Cape-Shiller price-earnings ratio has climbed above 40 for the index, well above its long-term average of 17 and higher than the 28 it registered before the 2008 financial crisis.

NVIDIA became the first company in history to reach a $5 trillion market capitalization in October, a figure that individually exceeds the annual economic output of both Germany and Japan. That milestone has come to symbolize just how concentrated and richly priced the AI sector has become in a relatively short period of time.

It is against that backdrop that economist Liam Halligan, writing in The Telegraph, is sounding the alarm, arguing that AI stocks trading at what he calls “ridiculous valuations” may represent a greater threat to global financial stability than the geopolitical risks currently dominating headlines, including the ongoing conflict in the Middle East.

The Debt Financing the AI Boom May Be Built on Shaky Ground

Halligan argues that a significant portion of AI infrastructure investment is financed through debt from unregulated private lenders, often secured against the AI chips themselves. There is, he warns, a meaningful gap between how those chips are valued on paper and how long they actually remain commercially useful.

Funding models typically depreciate AI chips over five to six years, according to the economist. Their practical lifespan, however, is often closer to two to three years, as faster and more efficient hardware enters the market in rapid succession. That gap means the collateral backing a large portion of AI-related debt may lose its value well before the loans against it are repaid.

That structural mismatch, the economist argues, is what makes the current cycle unusual. When previous technology booms collapsed, the physical infrastructure left behind, fiber-optic cables and rail networks, retained real-world value. The assets underpinning today’s AI debt cycle are far more vulnerable to obsolescence.

History Offers a Complicated Precedent for AI Investors

The railway boom of the 19th century and the internet bubble of the late 1990s are two of the clearest examples of transformative technology triggering speculative investment cycles. Both reshaped the global economy in lasting ways. Both also produced manias, widespread company failures, and sharp market corrections before the dust settled.

What made those cycles survivable for the broader financial system was durability. The tracks, cables, and networks built during those booms held their value even as the companies that built them went under, providing a floor for the debt that had financed them. That floor is harder to identify in the current AI investment cycle.

There are also practical constraints that current valuations may not fully reflect. AI data centers require a significant and consistent electricity supply, and the economist notes that stiff regulation is likely to follow as governments weigh the technology’s broader effects on jobs and society.

The Bigger Risk May Already Be Priced In

The geopolitical risks shaping the current economic outlook are real and well-documented. Energy price pressures, supply chain disruptions, and sovereign debt concerns across Western nations are all legitimate vulnerabilities. Halligan does not dispute any of that.

The economist argues that markets priced for perfection leave little room for the kinds of disruptions that have historically accompanied every major technological transition, and that AI valuations may not fully account for that risk.

Whether AI ultimately justifies what investors are paying for it today remains genuinely uncertain. The technology’s potential is not in serious dispute. What is less clear is whether the financial structures built around that potential are as stable as current market prices suggest.