AI productivity gains could cut U.S. deficit by half of $2.2T

AI's fiscal promise meets economic reality

Artificial intelligence could deliver $2.2 trillion in deficit reduction by 2036 through productivity gains, but more than half of those savings may evaporate through unintended consequences, according to a working paper published Wednesday by economists at the Brookings Institution and the Federal Reserve.

With the U.S. national debt surpassing $39 trillion in May and trust funds for Social Security and Medicare projected to deplete by 2032 and 2033 respectively, policymakers have looked to AI as a potential solution to fiscal pressures. The new analysis suggests that optimism requires significant qualification.

The researchers project AI-driven productivity increases could shrink the annual budget deficit from roughly 6% of GDP to as low as 2% within a decade. Higher output per worker would expand the tax base without rate increases, while AI could eliminate inefficiencies in health programs where administrative costs consume one quarter of spending.

Why it matters

This research provides the first comprehensive accounting of how AI's economic disruption creates offsetting fiscal costs that policymakers must plan for. The findings suggest AI may buy time but cannot substitute for difficult decisions on taxes and entitlements—a reality that challenges narratives positioning technology as a budget fix.

Five ways AI claws back savings

The Brookings and Fed economists identify specific mechanisms through which AI productivity gains generate new fiscal burdens:

Extended lifespans increase entitlement costs. AI-powered medical diagnostics and treatment could significantly reduce mortality rates. One sepsis-detection algorithm has been associated with a 17% decrease in patient deaths. But longer lives mean more years of Social Security and Medicare benefits. The researchers estimate a highly disruptive scenario could add 3 million retirement-age Americans by 2036.

Income shifts from wages to capital. If productivity gains flow primarily to asset owners rather than workers—as typically occurs—the government's revenue mix changes unfavorably. Individual income taxes currently generate 52% of federal revenue compared to 6% from corporate taxes. Capital gains face even lower effective rates around 5%, according to IRS data. More income earned as profits and returns rather than paychecks means lower average tax rates despite GDP growth.

Labor force participation drops. The analysis projects AI disruption could reduce workforce participation by 3 percentage points—equivalent to 6 million fewer workers by 2036, comparable to pandemic-era declines but likely permanent. Fewer workers means reduced payroll and income tax receipts while increasing enrollment in programs like SNAP and disability benefits.

Interest rates rise from infrastructure investment. Massive spending on chips, data centers, and supporting infrastructure may elevate the neutral interest rate, lifting market rates and federal debt-service costs. The researchers estimate this could add $60 billion annually to interest payments by 2036.

Defense spending accelerates. International competition in AI capabilities may trigger an arms race requiring increased military investment. Maintaining strategic advantage could add over $350 billion in cumulative defense spending over the next decade.

Together, these five factors could recapture more than half the fiscal gains from AI productivity, reducing the effective deficit reduction from $2.2 trillion to $1 trillion or less. The 1990s internet boom offers a cautionary precedent: tax revenues increased 2.2% of GDP and the deficit fell 60% between 1992 and 2002, but those gains eroded within a decade.

The findings were first reported by Fortune based on the Brookings Institution and Federal Reserve working paper.