AI Infrastructure Boom Is Fueling US Inflation, With $10.3 Trillion in Projected Spending Through 2032
Summary
- • AI infrastructure spending expected to hit ~$1T in 2026 — more than the US military budget
- • Columbia economist projects $10.3T total AI infrastructure investment through 2032
- • AI capex already a 'demand shock': inflating chip, energy, and construction prices
- • AI buildout on track to be the largest US investment cycle in American history
Details
AI infra spending ~$1T in 2026, exceeding US military budget
JPMorgan estimates AI infrastructure spending — data centers, chips, and servers — will hit around $1 trillion in 2026, surpassing annual US federal military spending.
Brookings paper: $10.3T cumulative AI infra spend through 2032
Columbia University economist Stijn Van Nieuwerburgh published a Brookings Institution paper projecting cumulative AI infrastructure investment of $10.3 trillion through 2032, equivalent to spending the entire 2021 bipartisan infrastructure law every year for a decade.
AI infra = 1.9% of US GDP in 2026; rises to avg 3.6% annually through 2032
Goldman Sachs estimates AI infrastructure represents 1.9% of total US economic output in 2026. Van Nieuwerburgh projects this will effectively double, averaging 3.6% of GDP per year through 2032.
Largest US investment cycle in history — bigger than railroads or the electrical grid
Van Nieuwerburgh projects the AI buildout will make up a larger share of total US GDP than any previous American investment boom: canals, railroads, the electrical grid, highway systems, and telecommunications networks combined.
Three simultaneous inflation drivers: energy shock, trade war, and AI capex
Economists identify three compounding sources of current US inflation: the global energy shock from wars in Iran and Ukraine, an escalating trade war with Canada and others, and the unprecedented Big Tech AI spending spree — all hitting simultaneously.
Still in 'Field of Dreams' phase — productivity gains not yet offsetting inflation
The AI buildout is still in a pre-revenue construction phase. Companies are spending trillions on infrastructure before AI delivers the productivity gains that economists say could eventually suppress inflation. The economy bears the cost today for gains that are years away.
Data center demand already inflating chips, energy, and skilled trade labor costs
Off-the-charts demand for memory and storage chips, building supplies, electricity, and skilled tradespeople — construction workers, plumbers, electricians — is already causing price increases in these sectors, per Georgia Tech economist Daniel Yue.
Hot economy compounds problem: low unemployment, strong consumer spending
The AI spending shock arrives when the US economy is already running hot — unemployment is low and consumer spending is strong — leaving limited economic slack to absorb the demand shock without feeding broader inflation.
Source: CNN analysis citing JPMorgan, Goldman Sachs, Columbia University/Brookings Institution research, and Georgia Tech economist Daniel Yue. Published September 28, 2026.
What This Means
AI infrastructure spending has grown large enough to function as its own macroeconomic variable, contributing to inflation at a time when the US economy is already running hot. The problem is structural: companies are spending trillions on capacity before AI yields the productivity gains that would offset those costs, creating a sustained demand shock with no near-term relief valve. Combined with energy price shocks from geopolitical conflict and trade-war tariffs, this makes the Federal Reserve's task substantially harder. Policymakers and economists are only beginning to model AI capex as a distinct inflationary force — a framing that will likely shape monetary policy and fiscal debate for years to come.