AI Hype Fails to Fix the Federal Budget

A Brookings Institution working paper published July 1, 2026 by Ben Harris (director of Brookings' Economic Studies program), Neil R. Mehrotra (Federal Reserve Bank of Minneapolis), and William Overcash models whether AI-driven productivity growth can resolve the fiscal trajectory that the Congressional Budget Office projects will push U.S. public debt to 175% of GDP by 2056. Separately, in a June 16, 2026 Reason analysis, Brookings-affiliated economist Jessica Riedl directly rebutted public claims - including Elon Musk's statement that AI and robotics are "the only thing that can solve for the debt situation" and Anthropic CEO Dario Amodei's comment that "the budget may balance without us doing anything because there's so much growth" - that AI alone will close the deficit.

The Brookings paper models a "once-in-a-generation" broad productivity shock that could cut annual deficits from roughly 6% to 2% of GDP, a meaningful improvement, but finds five AI-specific offsetting forces could claw back more than half of that gain: longer lifespans expanding entitlement spending, a shift of income from higher-taxed labor to lower-taxed capital, reduced labor-force participation raising income-support enrollment, higher interest rates raising debt-service costs, and a possible AI arms race increasing defense spending. The authors conclude the GDP growth assumption is the single largest driver of the fiscal outcome, making AI's macroeconomic impact the key uncertainty. Riedl's independent Reason analysis, using different assumptions, estimated a sustained 1-percentage-point productivity boost (comparable to the late-1990s IT boom) would raise tax revenue to $834 billion annually by 2036, but that worker-displacement costs (an estimated $250 billion annually) and higher interest rates from AI-driven investment demand (an estimated $240 billion annually) would offset much of that gain, netting to roughly $600 billion in annual savings, about one-eighth of the $4.4 trillion deficit Brookings projects under current policy. Riedl separately rebutted Sen. Bernie Sanders' proposal for the federal government to take a 50% equity stake in major AI companies, calling it "a fundamental accounting illusion" since unrealized equity gains generate no federal cash revenue unless shares are sold.

Teams pitching AI ROI to public-sector budget holders, or citing "AI will grow us out of the deficit" framing in funding or policy conversations, now have two independent, methodologically different analyses to point to that reach a similar conclusion: even optimistic productivity scenarios leave the large majority of the U.S. deficit unaddressed once offsetting costs are netted out. Both are also a reminder that productivity gains do not automatically flow to government revenue - the split between labor and capital income, and between government and private payers in healthcare, changes the fiscal outcome substantially.

Whether CBO's long-term budget projections begin incorporating explicit AI-productivity scenarios; realized productivity-growth data as AI adoption scales (a June 2026 CEPR study cited by Fortune found AI-attributed labor productivity growth of about 1.8% in 2026); and the fate of proposals like Sanders' AI equity-stake plan.

Riedl's piece is a single-author opinion column in Reason, a libertarian-leaning outlet, though she is herself a Brookings Institution fellow and her analysis is grounded in cited CBO, Brookings, White House OMB, and IMF figures. The subsequently published Brookings working paper is a more formally reviewed research product (acknowledging feedback from named economists and funded by the Peter G. Peterson Foundation) but represents the authors' own model and assumptions, not a CBO or government forecast; both should be read as serious but contestable analyses rather than settled consensus.