Few economic debates have attracted as much optimism as the link between AI and public debt. The hope is simple: if artificial intelligence delivers a large jump in productivity, economies will grow faster, tax revenues will rise and the heavy debt burdens carried by advanced economies will become easier to manage. Kenneth Rogoff, the Harvard economist and former chief economist of the International Monetary Fund, is not convinced. In a recent commentary he argues that AI could make deficits harder, not easier, to contain.
Rogoff is one of the best-known voices on sovereign debt. His book with Carmen Reinhart, This Time Is Different, traced centuries of financial crises and became a reference point for debates about how much borrowing governments can safely carry. His scepticism therefore carries weight with investors and finance ministries alike.
The Optimistic Case for AI-Driven Growth
The optimistic scenario has a clear logic. Public debt in many rich countries has climbed sharply since the 2008 financial crisis and again during the pandemic. In the United States, federal debt held by the public is now roughly equal to the size of the economy, while Japan’s gross debt is more than twice its GDP. Several large European economies also carry debt above 100 percent of output.
Economists often point out that the most painless way to reduce a debt ratio is to grow the denominator. If AI lifts productivity growth even modestly for a sustained period, the cumulative effect on output and tax receipts could be large. Supporters argue this could do for the 2020s and 2030s what rapid post-war growth did for countries that emerged from the Second World War with enormous debts.
Why AI and Public Debt May Not Mix So Easily
Rogoff’s argument is that this narrative overlooks several forces that could push in the opposite direction. The first is income distribution. If the gains from AI flow mainly to the owners of capital and a small number of technology companies, rather than to workers broadly, the tax base may not expand as much as headline growth suggests. Capital income is often more mobile and more lightly taxed than wages, and large technology firms have a long record of shifting profits to low-tax jurisdictions.
The second is spending pressure. Advanced economies already face rising costs from ageing populations, pensions and healthcare. On top of that, governments are under pressure to raise defence budgets, finance the energy transition and invest in their own AI capabilities. Even if revenues rise, outlays may rise just as fast.
Labour Disruption and the Cost of Adjustment
The third concern is disruption in the labour market. If AI displaces workers faster than new roles appear, governments may face higher spending on unemployment support, retraining and social programmes. Political pressure to cushion the transition could be intense, especially if job losses are concentrated in particular regions or professions.
History offers some comfort, as past technological revolutions eventually created more jobs than they destroyed. But the adjustment periods were often long and painful, and the fiscal costs of managing them fell largely on the state. A rapid AI transition could compress that disruption into a shorter period, raising the budgetary strain just when governments are hoping for relief.
Regulation and Geopolitical Risk
Rogoff also points to weak regulation and geopolitical risk as reasons for caution. Poorly governed AI could create new kinds of financial instability, fraud or cyber threats, which may ultimately require public money to contain. In past crises, from banking collapses to pandemic responses, governments have repeatedly ended up absorbing losses that began in the private sector.
Geopolitics adds another layer. The race for leadership in AI has become a strategic contest, particularly between the United States and China. That competition encourages subsidies, industrial policy and defence-related spending on technology, all of which add to government outlays. Export controls and supply chain tensions around advanced chips also create uncertainty about how widely and how quickly the gains from AI will spread.
Interest Rates: The Hidden Variable
An important part of any debt debate is the cost of borrowing. For years after the financial crisis, extremely low interest rates allowed governments to carry rising debt without much pain. That era ended with the inflation surge of 2021 and 2022, and higher rates have since pushed up interest bills in many countries.
Faster productivity growth could, in principle, raise the natural rate of interest, because stronger investment demand competes for savings. If borrowing costs rise alongside growth, the benefit to debt dynamics would be smaller than optimists expect. Rogoff has argued in the past that markets and policymakers were too quick to assume that ultra-low rates were permanent, and the same caution applies here. His full reasoning can be read in Rogoff’s analysis in World Finance.
What Governments Should Take From This
The practical lesson is that fiscal planning should not depend on an AI dividend that may arrive late, be smaller than expected or be unevenly shared. Governments that treat future productivity gains as a reason to delay hard choices on spending and taxation risk being caught out. A more prudent approach would be to stabilise debt now and treat any AI-driven windfall as a bonus.
Tax policy may also need updating. If AI shifts income towards capital and intangible assets, systems that rely heavily on taxing wages could see their base erode. International cooperation on corporate taxation, such as the OECD-led minimum tax agreement, may become more important as AI-driven profits grow.
Bond investors will be watching closely. Markets tend to tolerate high borrowing when they trust that governments have a credible plan. If the debate over AI and public debt encourages complacency in fiscal policy, that trust could weaken, pushing up risk premiums on government bonds and making the debt burden even more expensive to carry.
Conclusion
The relationship between AI and public debt is far less straightforward than the optimistic story implies. Rogoff’s warning is not that AI will fail to boost growth, but that the fiscal benefits could be offset by distributional shifts, new spending demands, labour disruption and geopolitical competition. For heavily indebted rich countries, the safest assumption is that technology will not solve their debt problem for them.
