Can AI Become the Savior of U.S. Treasury Bonds? — The Paradox of Tech Dividends and Fiscal Dilemma
Introduction
By end of 2025, U.S. national debt approached $38 trillion, with net interest expenses near $1 trillion. The rising interest deficit ratio puts significant pressure on fiscal space and economic vitality. With monetary policy unlikely to ease and fiscal revenues under pressure, the market pins hopes on AI-driven productivity gains to dilute debt. However, this expectation faces numerous structural challenges in practice. This article draws on historical debt resolution experience, analyzes AI's real effectiveness in the current U.S. fiscal system, and explores whether it can truly become the 'savior' of U.S. Treasuries.
Three Paths to Reduce Debt Ratio and Historical Lessons
In the framework of macroeconomic and fiscal sustainability analysis, the evolution of the debt ratio is jointly determined by interest rates, GDP growth, and the primary deficit. Accordingly, reducing the debt ratio can be attributed to three core paths: first, lowering interest rates on the monetary side to ease interest burdens; second, relying on technological innovation on the economic side to boost economic growth, achieving growth outpacing debt growth; third, compressing the primary deficit through fiscal spending cuts and revenue increases to control new debt from the source. Among these, only the second and third paths have the fundamental ability to resolve debt risks.
The two phases in U.S. history when debt problems were substantially resolved align well with this framework. From 1946 to 1974, the U.S. debt ratio fell from over 100% to around 20%, driven by a combination of interest rate controls, technology conversion, and fiscal surpluses. The Clinton era in the 1990s relied on the Internet technology revolution and fiscal discipline to turn deficits into surpluses. A common feature of these two successful episodes: technological innovation not only expanded the denominator of the economy but also generated significant tax revenue increments, while the fiscal side converted dividends into surpluses through structural reforms.
Current Realistic Dilemma of AI Effectiveness
However, the 'jobless prosperity' driven by AI today fundamentally differs from traditional technology revolutions. On the productivity dimension, there is a significant time lag in the transmission of AI tech dividends to the real economy. According to estimates by the University of Pennsylvania, AI will only boost total factor productivity by 0.05 to 0.1 percentage points from 2026 to 2027, and only by around 0.2 percentage points by the early 2030s. This scale and speed are far from offsetting the current fiscal and debt pressures on the U.S.
More notably, the imbalance in factor income distribution caused by AI is eroding the tax base structurally. In U.S. federal revenue, personal income tax accounts for about 51%, payroll tax about 33%, together contributing about 85%, while corporate income tax only about 10%. AI's labor substitution effect is pushing high-income white-collar groups into permanent unemployment or lower-paying jobs, directly shrinking the base for personal income and payroll taxes. Meanwhile, corporate profit growth, due to the strong tax avoidance capabilities of multinational tech giants, can hardly compensate for fiscal revenue. Under the current tax system that heavily relies on 'taxing people' rather than 'taxing capital,' the more technology booms, the more the tax base shrinks—a paradox that has become the core contradiction in U.S. fiscal policy.
Solution: Structural Adjustment of Fiscal and Tax System
The experience of new economy governance during the Clinton era provides important references. In 1993, the Clinton administration passed the Omnibus Budget Reconciliation Act, raising the top marginal personal income tax rate from 31% to 39.6%, and slightly adjusting the large corporate income tax rate from 34% to 35%, while cutting defense and other public spending. History shows that when technological change widens the wealth gap, efficient fiscal transfer payments serve as a 'safety valve' to maintain system stability.
In the AI era, the traditional 'capital-labor' distribution pattern may break down at a faster pace. Current policy discussions mainly point in three directions: first, demanding dividends from the 'capital side' by raising capital gains taxes and imposing higher marginal rates on the wealthy; second, taxing the 'factor side' by levying 'digital factor taxes' on commercial gains from large models; third, pushing transfer payments via 'automation' by exploring 'robot taxes' or 'AI taxes' on companies that massively replace labor. These approaches essentially involve structurally taxing the 'factor beneficiaries,' redistributing tech dividends to society as a whole through secondary fiscal allocation.
Governance Dilemmas and Real-World Challenges
However, from theoretical conception to actual policy, the U.S. faces multiple structural obstacles. First, the intangibility and cross-border flow of AI factors render traditional tax systems based on permanent establishments ineffective, eroding the tax base. Second, in the context of global AI competition, any unilateral tightening of micro-taxation may suppress domestic innovation momentum, triggering reverse outflow of tech capital and high-end factors. Third, multinational tech giants that control core computing power and digital discourse have strong bargaining power in political games, making secondary redistribution of wealth stock difficult.
Conclusion
In summary, the current 'jobless prosperity' driven by AI has limited effectiveness in solving the debt problem under the U.S. fiscal system. In the short term, if AI merely replaces existing labor and services with lower prices without creating substantial new demand, it will not only fail to alleviate the deficit but may worsen the debt problem. To resolve this dilemma, the U.S. needs to push for transmission of tech dividends to the real economy while undertaking decisive fiscal and tax reforms. However, the fundamental conflict between 'digital productivity knows no borders' and 'fiscal sovereignty knows borders' means that the fiscal adjustment in the AI era cannot be achieved overnight, and the U.S. debt problem will remain a major challenge for the U.S. economy in the near term.
