Why the tax code doesn’t need an AI apocalypse – a data‑driven rebuttal

The rise of generative AI has sparked a frenzy of tax‑code proposals, yet the data suggest modest adjustments, not a sweeping rewrite. Historical tax reforms have favoured broader bases rather than niche levies, even as technology reshaped industry.

Tax code fundamentals amid AI hype

Labor‑saving tools have altered job structures, but net‑income shares for workers and capital owners have remained roughly stable. In 2025 the U.S. recorded ↑ 63 million hires alongside a similar number of separations, underscoring market fluidity.

“Technology changes the rules of the game, not the principles of taxation,” said a senior Treasury analyst.

Proposals to tax tokens, compute cycles, or impose AI‑specific surcharges would penalise adoption and reverse the long‑term trend toward wider tax bases that emerged after the Industrial Revolution. Property taxes, for example, have already financed data‑center growth in Loudoun County, Virginia, without special carve‑outs.

Revenue implications are modest. Capital‑gains receipts from AI‑related equity sales are projected at ↑ $45 billion this year, while the federal deficit sits at ↓ $1.5 trillion. Redirecting existing streams to targeted assistance is far more efficient than creating a new AI‑tax fund.

Existing programs such as Trade Adjustment Assistance illustrate the pitfalls of narrowly‑tailored aid—low participation and attribution challenges. Enhancing the universal unemployment insurance system offers a proven, scalable alternative.

In short, the tax code’s core principles—simple rules, low rates, broad bases—remain sound. Policymakers should resist reactionary measures and focus on fiscal discipline, leveraging proven mechanisms to cushion any labour market shifts.

For further context see Reuters and Bloomberg.

Dispatch from Arthur Sterling (Macroeconomics Editor).