AI could cut deficits by half, Brookings says, but it won't fill the $40T hole

Fortune reports that SpaceX founder and CEO Elon Musk argues AI and robotics used at large scale are “pretty much the only thing that’s going to solve the U.S. debt crisis.” His figure, as Fortune frames it, points to a debt burden of $39.5 trillion at the time of writing, and the hope is straightforward: if AI boosts productivity enough, the U.S. can grow its way out of worsening federal deficits without the political pain of big spending cuts.

But new research from Brookings, authored by Ben Harris, Neil R. Mehrotra, and William Overcash, lands a reality check. The study says that even in more optimistic scenarios, AI-driven growth is still unlikely to bridge the gap “even in more optimistic scenarios fully.” In the best case, Brookings finds AI’s productivity channel could reduce deficits by about half. In the worst case, mitigating factors could knock two-thirds off any improvement. Translation: AI may help, but it is not a plug-and-play solution for a $40T-class fiscal problem.

To understand why this argument is getting so much airtime, you have to understand the debate Brookings is responding to. In the “rebalancing” argument about federal debt and economic growth, optimists often prefer expanding the economy over cutting federal spending. It is less painful politically and socially. Musk, a “debt hawk” in Fortune’s telling, has leaned into the same logic: AI and robotics productivity gains, deployed at scale, are the lever that could keep growth strong while shrinking the deficit.

Brookings acknowledges why the optimism exists. The study points to the already-heavy capital expenditure into transformative AI technologies and to “the unharnessed capacity of the technology to boost productivity.” And the evidence that AI spending is moving real-world assumptions is not just theoretical. Fortune notes that AI investment has continued at a rapid pace and has even surprised analysts. BNP Paribas lifted its near-term U.S. GDP growth estimates earlier this year based on capex announcements implying a larger impulse from the AI buildout than previously expected. BNP still estimated full-year 2026 growth at 2.6%, but it flagged that a Q4/Q4 comparison this year versus last would be 2.6% rather than the 2.1% it had estimated before.

There is also empirical work pointing to early productivity effects. Fortune cites a June study from the Centre for Economic Policy Research (CEPR), finding that the implied measure of AI-attributed labor productivity growth for 2026 is 1.8%. The gains are expected to be highest in high-skill services and finance, where they exceed 2%.

Where Brookings turns from “AI might help” to “AI might not save you” is in the second-order mechanics of how productivity gains flow through government budgets. The study highlights that healthcare is both expensive and inefficient. Outlays for Medicare and Medicaid in 2026 are expected to be $674 billion and $472 billion, respectively, according to Congressional Budget Office estimates cited by Fortune. In a positive scenario, Brookings argues AI could materially affect the fiscal outlook because the healthcare sector shows substantial misallocation and inefficiency that a productivity shock could reduce.

That sounds like a win, until the broader system responds. Brookings warns that efficiency gains that lower costs can also increase longevity, which means more demand for social security support later on. The study also flags labor market churn: the “much-feared labor market shift” could mean more unemployment and more reliance on income support payments as the dust settles. Then there is defense. If countries try to win an AI arms race, defense spending is likely to increase.

Finally, Brookings highlights how national income composition and financial rates can move the goalposts. A changing mix of income may shift the tax base away from highly taxed labor income toward less-taxed non-corporate capital and corporate profits. And higher investment demands can raise the neutral rate of interest, pushing up equilibrium interest rates. Higher equilibrium rates increase interest expenditures for the government.

So even though AI improves the budget outlook “somewhat,” the study’s conclusion is blunt: you cannot rely on AI alone to solve the U.S. fiscal problem. Fortune reports Brookings’ finding that, at best, these offsetting factors cut the deficit-reduction benefit by roughly half. At worst, they could reduce any improvement by two-thirds. That is the gap between techno-optimists and reality that decision-makers should care about.

For executives, investors, and boards, this is not just a macro story. It is an incentive story. AI capex is still a race, and if governments are not counting on AI to close the deficit completely, then the burden of fiscal adjustment does not disappear. It may shift in who pays, what gets prioritized, and how quickly budgets get squeezed. Brookings’ framework suggests that even if productivity shocks arrive, policymakers will still confront healthcare cost pressures, labor transitions, defense competition, tax base shifts, and higher interest-rate dynamics. In other words: AI is likely to be part of the solution, but it is also likely to create follow-on costs that policy has to manage, not ignore.