U.S. would likely still be growing without AI boom, UHERO analysis finds

AI has helped the U.S. economy grow, but that is not the same as saving it from recession. An October 21, 2026 analysis from the University of Hawaiʻi Economic Research Organization estimates that growth would have been weaker without AI investment and stock-market gains—yet probably still positive. A sudden reversal presents a different risk.

Economist Byron Gangnes estimates that removing those AI-related forces would subtract a little over half a percentage point to just over one percentage point from growth over the past year. At the upper end, growth would have been about half the actual 2.2% expansion.

The buildout is bigger than its domestic contribution

Gangnes puts annual real data-center construction spending at about $59.5 billion recently, up from $11.5 billion at the start of 2023, in inflation-adjusted 2017 dollars. Spending on information-processing equipment and software is running about half a trillion dollars above its 2022 level.

But much of the physical equipment is imported, so buying it adds little to domestic output. Software data do not isolate AI. Gangnes also finds evidence that AI spending has displaced other investment, potentially through higher interest rates, equipment prices and competition for electricity or water.

Household wealth does more of the work

Gangnes estimates that AI-related stock wealth added $255 billion–$509 billion to real consumer spending since late 2022, representing 17–34% of its total increase. He puts this spending channel’s contribution to economic growth at 0.4–0.8 percentage points—larger than the net investment contribution.

Two assumptions shape the estimate

• Which gains count: Gangnes attributes 75% of S&P 500 gains to AI, drawing on a J.P. Morgan estimate, and assumes 50% for smaller U.S. companies and foreign stocks.
• How much households spend: He assumes two to four cents of extra annual spending per dollar of stock wealth, with adjustment completed within a year.

Using a lower, one-cent spending response reduces AI’s total estimated growth contribution to 0.35–0.45 percentage points. UHERO labels the article preliminary material intended to stimulate discussion and critical comment.

A reversal is not the same as no boom

Gangnes warns that an abrupt stock reversal could quickly cut household spending and halt planned AI investment. Damage to confidence and credit-market turmoil could reinforce the pullback, potentially causing recession. His analysis assumes little growth benefit so far from AI making work more productive; it does not measure AI’s eventual economic value.