The AI Dividend: When Government Starts Producing Wealth

Every generation produces politicians who promise to shrink government. Few succeed, not because voters reject the idea, but because the cost of governing has remained stubbornly human. Bureaucracies scale through people, and people are expensive. Governments can digitize forms, outsource departments, or eliminate programs, but the fundamental economics have remained unchanged: public administration is labor intensive. Artificial intelligence may be the first technology capable of breaking that relationship.

Around the world, a new political movement is emerging around this possibility. Argentina's Javier Milei has built his economic agenda around dramatically reducing the size and cost of the state. In the United States, Elon Musk's Department of Government Efficiency (DOGE) demonstrated both the appetite for radical government reform and the limits of pursuing it with traditional tools. Identifying waste is one challenge. Permanently lowering the cost of governing is another. Until now, governments simply lacked the technology to achieve lasting productivity gains at scale.

Imagine a government where permitting, procurement, benefits administration, compliance, fraud detection, legal review, and much of the machinery of bureaucracy are handled by autonomous systems rather than layers of administrators and contractors. The objective is not fewer public services. It is delivering the same or better services at a permanently lower cost. That distinction transforms efficiency from a one-time budget exercise into a recurring source of economic value.

The question is no longer whether government can become more productive. The question becomes who owns that productivity.

Private companies already have an answer. When firms become more efficient, the gains are distributed through higher profits, lower prices, increased investment, or higher wages. Governments have no equivalent mechanism. When a public agency becomes more efficient, the savings usually disappear into larger budgets, expanding mandates, or the natural growth of public spending. Productivity gains exist, but citizens rarely experience them directly.

Governments should instead establish what might be called an AI fiscal dividend.

Every verified, recurring dollar saved through AI-driven productivity becomes a fiscal dividend that is automatically allocated between two objectives: reducing taxes today and reducing public debt for tomorrow. The ratio between those two uses becomes a transparent fiscal policy rule. During recessions, more of the dividend could be returned to taxpayers, supporting investment and consumption without increasing deficits. During periods of elevated debt or inflation, more could be directed toward retiring government debt, reducing future interest payments and strengthening the public balance sheet.

The emphasis on debt is particularly important because most developed economies have run out of politically painless options. Raising taxes slows growth. Cutting benefits is politically toxic. Inflating debt away erodes trust in the currency. Yet debt-to-GDP ratios continue to climb while interest payments consume an ever-larger share of public budgets. AI offers a path that none of those alternatives provide: expanding fiscal capacity by lowering the cost of government itself. Every dollar of recurring productivity gains used to retire debt permanently reduces future interest obligations, creating a compounding fiscal benefit. Instead of borrowing to finance yesterday's bureaucracy, governments could allow tomorrow's productivity to repair today's balance sheet.

This is fundamentally different from austerity. Austerity improves public finances by asking citizens to accept fewer services or higher taxes. An AI dividend improves public finances because the state itself has become more productive. The source of fiscal capacity is no longer sacrifice. It is productivity.

For centuries, governments have financed themselves through three primary mechanisms: taxation, borrowing, and, in a handful of countries, natural resource revenues. AI introduces a fourth. Governments can generate wealth through institutional productivity. They become capable of producing economic value by continuously lowering the cost of delivering public services. Traditionally, governments have been viewed as institutions that redistribute wealth created elsewhere in the economy. A highly productive state becomes something different: an institution capable of creating fiscal wealth on its own.

Of course, this creates an entirely new institutional challenge. Government spending is unlikely to fall simply because administration becomes cheaper. Healthcare costs will continue rising, populations will age, and political demand for public services will evolve. The relevant measure is therefore not whether total spending declines, but whether AI lowers the cost of delivering an equivalent level of service. Those savings must be independently verified, adjusted for implementation costs, infrastructure, cybersecurity, and ongoing maintenance, and measured against a credible counterfactual. Without rigorous accounting, every administration will claim enormous AI savings regardless of whether they actually exist.

Equally important, without a formal allocation rule, the productivity gains will simply disappear into bureaucracy. Agencies will argue they should retain the savings. Legislators will redirect them toward new spending. Temporary efficiencies will become permanent commitments. The public will once again finance technological progress without ever receiving its benefits.

There is another development that makes this moment unique. Governments are no longer merely regulating the AI revolution. Increasingly, they are becoming investors in it. In the United States, industrial policy has already shifted toward supporting domestic semiconductor manufacturing, AI infrastructure, and strategic technology through programs such as the CHIPS Act and other public investments. Whether through direct funding, subsidies, procurement, or equity-like exposure, the state is becoming financially intertwined with the very companies building the infrastructure of artificial intelligence. If governments are helping finance the technologies that will transform public administration, it becomes even more reasonable to ask whether citizens should receive a direct fiscal return when those technologies permanently reduce the cost of governing.

The political significance extends well beyond any single administration or ideology. Whether it is Milei in Argentina, fiscal reform efforts across Europe, or experiments like DOGE in the United States, the underlying objective is the same: governments are searching for ways to become permanently less expensive without becoming less capable. Previous generations lacked the technology to make that ambition realistic. Ours may not.

The defining economic question of the coming decade may not be how much productivity artificial intelligence creates. It may be whether governments build institutions capable of capturing that productivity for the public. If they succeed, the most important legacy of AI may not be smarter bureaucracies or faster public services. It may be something far more enduring: a fiscal system where technological progress steadily lowers national debt, expands economic freedom through lower taxes, and transforms government from a perpetual consumer of wealth into a long-term producer of it.