More Alarm Bells: The AI Bubble Warning Signs Keep Coming

If five warning signs were not enough, the evidence continues to pile up. The AI investment boom of 2026 is generating alarm from directions few anticipated — including from the very people building and financing it.



The Infrastructure Overbuild


AI infrastructure spending in 2026 is on track to exceed, in both scale and duration, the telecom capital expenditure that defined the dot-com bubble. The largest technology companies are expected to represent 40% of total Russell 1000 capital expenditure between 2026 and 2028 — a figure exceeding $2 trillion. The sheer weight of this buildout assumes a level of future demand that has never been independently verified, only enthusiastically projected.


Semiconductor Valuations in Uncharted Territory


Nvidia has reached a market capitalisation of $5.45 trillion, with a price-to-earnings ratio of 43. Some AI firms are raising capital at valuations of 150 times revenue. The entire semiconductor sector climbed 65% in a matter of months. These are not the numbers of a maturing industry pricing in steady growth — they are the numbers of a market betting everything on a single outcome and refusing to price in any alternative.


The Industry Is Starting to Say It Itself


Perhaps the most telling alarm bell is the one coming from inside the house. In August 2025, the Nasdaq fell 1.4% in a single morning after OpenAI’s own chief executive acknowledged that investors might be overexcited about AI. Shortly after, Goldman Sachs’ CEO admitted publicly that a large portion of the capital being deployed in AI would likely produce no returns at all. When the architects and financiers of a boom begin hedging their language, it is rarely a coincidence.


A Pattern History Knows Well

The AI boom is not the first time Wall Street has convinced itself that a transformative technology justified any price. The internet was transformative too — and it was. But between 2000 and 2002, the Nasdaq lost nearly 80% of its value.


The technology survived; the valuations did not. Today’s AI rally shares the same defining characteristic of every bubble that preceded it: the price has run so far ahead of the earnings that only perfection will do.

Any disappointment — a single weak earnings quarter, a Federal Reserve pivot, a geopolitical shock disrupting chip supply chains — could be enough to tip the balance.


Several alarm bells ringing at once does not guarantee a crash is imminent. Markets can remain irrational far longer than logic suggests they should. But history is consistent on one point: bubbles do not announce themselves on the way up. They are only obvious in the wreckage.