The US economy’s problem may be that it’s too strong. Rising prices and interest rates stem from three forces: an energy shock from the Iran and Ukraine wars, an escalating trade war, and an unprecedented corporate AI spending spree. While the first two are issues that can be addressed through political will, economists are increasingly worried about the last one.

According to an article published in CNN, AI infrastructure spending on data centers, chips, and servers is expected to hit about $1 trillion this year, according to JPMorgan, more than the federal military budget. Columbia economist Stijn Van Nieuwerburgh projects $10.3 trillion through 2032, averaging 3.6% of GDP annually, up from roughly 1.9% this year per Goldman Sachs. That would exceed the GDP share of earlier booms like railroads, canals, highways and telecom networks.

The inflation risk arises because this is a demand shock arriving before AI’s promised productivity gains. Demand for memory and storage chips, building materials, electricity and skilled tradespeople is outrunning supply, which is constrained by regulations, permitting and the immigration crackdown. These costs are spilling into other industries through labor shortages and pricier supplies.

All of this lands on an already hot economy. Manufacturing activity is at its highest since July 2021, unemployment is 4.1%, and August retail sales rose 1.2%. An AI-driven stock boom, with the eight largest S&P 500 stocks all effectively AI companies making up 38.8% of market value, keeps higher earners spending. They account for 70% of consumer spending, which lets companies pass on higher costs from record diesel and shipping prices and tariffs. Inflation has sat above the Fed’s 2% target for over five years and would remain a problem even if the war ended and tariffs vanished. Meanwhile, Main Street absorbs the costs without sharing the gains.

The article describes a bond market meltdown driven by an overheating economy rather than a weak one, with yields rising alongside inflation. The logic for further Fed hikes is straightforward: demand is strong across the board, inflation has been stuck well above target for years, and AI capex adds fresh demand for scarce goods and labor. Chicago Fed president Austan Goolsbee said that if AI construction pushes demand past what the economy can absorb, there is “no ambiguity” about the Fed’s response: it would have to raise its target rate further. The usual brake, higher borrowing costs, may also fail to work. Since AI investment is rate-insensitive, the buildout is unlikely to slow soon, even with higher interest rates that may force the Fed to push rates higher to cool everything else.

The joker in the pack is therefore the returns on these investments. Once the returns kick in, the actual efficacy of these investments will show up on balance sheets, which may lead to some rationalization of future spending. However, since most of these investments will already be made, the RoI numbers will also take a few quarters to impact the economy.

End Notes

Source: https://edition.cnn.com/2026/09/28/economy/ai-economy-inflation