The AI Bubble #24,736: Consumption Is Outrunning Wages

Stock market gains are driving consumption

I have already written several pieces on the AI bubble. It sure looks like the late 1990s tech bubble. I know we have a big “this time is different crowd,” but Sam Altman and Mark Zuckerberg don’t seem any sharper than the wizards of the universe who dominated the business pages in Internet bubble days.

The point I emphasized in my prior posts is that to make sense of the market capitalization of the magnificent seven dominating the stock market, over the next five years they would need a pace of profit growth that would imply either radically more rapid economic growth than almost any one is predicting, or shrinking profits for the rest of corporate America. Neither of those seem likely, but sure, anything can happen.

Another part of the story that many of us have emphasized in talking about the risks of a collapse of the bubble is that consumption is being driven by the bubble. This point can be shown in different ways, but a simple one is simply to take the ratio of labor compensation to consumption.

Source: Bureau of Economic Analysis.

This gives a very clear picture. From 2013 to just before the pandemic, the ratio of labor compensation (wages plus benefits) to consumption was mostly between 75 and 76 percent. The pandemic led to a big jump in the ratio due to a fall in wages, associated with the jump in unemployment when the economy shut down, and the big stimulus checks the government sent out in 2020 and 2021.

After the economy returned to more or less normal in 2022, the ratio started to fall rapidly. It hit 71.6 percent in the third quarter of 2025, the most recent period for which we have data.

While this drop of 3.4 percentage points might seem small, it corresponds to roughly $1 trillion in annual consumption. That is the amount of additional consumption that can be attributed to the AI bubble and the amount of consumption that would be lost if the bubble bursts. That is equal to 3.0 percent of GDP, which is certainly enough to give us a recession, especially when added to the collapse of spending on data centers and other AI investments.

The extent to which consumption outpaces wages is likely to increase as the bubble grows further. That will make the recession following the collapse worse and the adjustment process longer. The sooner the AI bubble bursts, the better it will be for almost all of us, except the AI whizzes.