U.S. policy makers and parts of the investment community are pinning hopes on artificial intelligence (AI) to accelerate economic growth and thereby ease the federal debt burden. That expectation faces significant uncertainty: historical data show U.S. per‑capita growth has hovered around roughly 2 percent annually since 1870, and no past technological wave has permanently lifted that long‑run pace.
What’s at stake?
The federal budget is projected to close 2026 with a deficit of about $1,900 billion, roughly 6 percent of GDP. Market‑held federal debt will reach about 101 percent of GDP this year; the Congressional Budget Office (CBO) projects debt could rise to 120 percent by 2036 and to 175 percent by 2056. For many policymakers the least painful route to dealing with this trajectory is faster economic growth.
Conventional remedies — tax increases, spending cuts, revising social and health programs, or tolerating higher inflation to erode the real value of debt — are politically difficult or have other costs. That makes growth, and the possibility that AI could produce it, particularly attractive as a policy goal.
Official growth assumptions and AI’s expected contribution
The CBO assumes roughly 1.8 percent annual real growth after 2027 and already includes a modest positive AI effect in that outlook. The office estimates generative AI could lift productivity growth by about 0.1 percentage point per year on average and raise non‑farm business output by roughly 1 percent by 2036. In a sensitivity case, the CBO finds that if productivity growth falls 0.1 percentage point below the baseline each year, the cumulative deficit over 2027–2036 would rise by around $317 billion.
A larger, sustained productivity increase would have a correspondingly larger fiscal impact, but the central question is whether the U.S. economy can move materially above its long‑run roughly 2 percent growth path.
Government support for AI investments
AI spending is not purely a private market story: the federal government actively shapes the environment for these investments. A 2025 presidential executive order sped up permitting for large data centers and associated energy infrastructure, and enabled financial supports, loans and tax incentives for priority projects, including data centers, transmission networks, power plants, semiconductor fabs, networking equipment and storage infrastructure. In 2026 the government reached separate arrangements with major technology firms to avoid passing the full costs of new power generation and grid upgrades for data centers onto residential consumers.
Officials justify public involvement on grounds of technological competitiveness, national security and energy considerations. From a fiscal-policy perspective, the objective is also to accelerate any productivity gains so they convert into higher tax revenues.
What is already visible in the data?
AI‑related spending has started to show up in macro data: technology firms are building data centers, purchasing large volumes of chips and servers, expanding network capacity and signing long‑term energy contracts. The Federal Reserve finds that software, computing, data‑center and energy investments linked to AI contributed noticeably to U.S. GDP in some quarters beginning in 2025; in the most recent quarter the contribution was 0.9 percentage point. That effect so far is largely investment driven — building a data center raises GDP while construction and equipment spending take place — but does not automatically translate into higher economy‑wide efficiency.
Historical perspective and expert estimates
Long‑run data indicate U.S. per‑capita growth has clung to a roughly 2 percent trend since 1870. Major technological revolutions (steam, electrification, the internet) produced visible but ultimately temporary productivity bumps; over the longer term the growth rate returned to its prior trend.
Estimates for generative AI’s annual productivity uplift vary: McKinsey’s range is about 0.1–0.6 percentage points; stricter academic models suggest 0.07–0.2 percentage points. Those figures could be meaningful within a typical business cycle but are well below the most optimistic scenarios and the magnitudes implied by current market pricing. They also look small relative to what would be needed to materially change the federal debt outlook.
Proponents argue AI is different because it could accelerate the innovation process itself — for example by speeding discovery of new materials, drugs or technologies — which in theory could shift the long‑run trend. Similar arguments were made about the internet, however, and the historical trend remained intact.
Potential consequences
If AI fails to deliver a sustained productivity breakthrough, the current investment boom will have generated costs without creating the fiscal space policymakers hope for. That could force difficult trade‑offs sooner: tax increases, spending cuts, tolerating higher inflation, or more intrusive forms of financial repression and the institutional apparatus that would entail. Given the political reluctance to impose austerity, failure of the AI scenario would remove one of the more palatable options for addressing debt dynamics.
In sum, AI investments are already affecting GDP through higher investment and the federal government is actively supporting the build‑out to maximize any productivity gains. Yet the 150‑year U.S. growth record and relatively modest central estimates of AI’s productivity effect leave substantial doubt about whether AI alone can resolve the long‑term fiscal challenge.
Tags: budget, economic growth, artificial intelligence, productivity, budget deficit, public debt, data center, U.S. economy, AI, United States
