Policy Snapshot
Giving citizens a direct ownership stakes in AI infrastructure via equity stakes
Scenario
Gradual
Augmentation
All Scenarios
Rapid
Automation
Scope
Near Term
(Volatility Risks)
Medium Term
(Transition Risks)
Long Term
(Structural Risks)
Governance Level
Local
National
International
Target
Entrepreneurs
Displaced Workers
Primary Actor
Governments
Private Actors
Consumption Taxes
Broad-based taxes on spending, such as value-added taxes or goods and services taxes, that maintain government revenue even as labor income declines.
What it is:
Consumption taxes, including value-added taxes (VATs), goods and services taxes (GSTs), and sales taxes, are levied at the point of purchase, creating a revenue base tied to spending rather than reported income or payroll. As AI reshapes the economy, labor income may decline or become less reliable as a tax base. Consumption taxes remain effective because automated systems and AI-generated wealth still ultimately drive consumption. While traditionally viewed as regressive, modern VAT systems can be paired with progressive rebates or "prebates" to offset burdens on lower-income households, ensuring revenue resilience without exacerbating inequality.
Recommended Reading:
Anton Korinek and Lee Lockwood
The future of tax policy: A public finance framework for the age of AI
January 2026
Korinek and Lockwood provide a public finance framework distinguishing AI consumption taxes (sound policy) from AI capital taxes (harmful to innovation). They support token taxes, digital services taxes, and robot services taxes applied at the retail level, but argue against taxing compute infrastructure or robot ownership in the near term. They propose that taxing AGI systems' resource accumulation may become necessary only if AI entities become autonomous economic actors — a contingency they flag but do not recommend implementing now.
Convergence Analysis
Funding Government in the Age of AI
August 2025
Huynh, Mittal and Frank identify consumption taxes as one of the two most resilient revenue instruments for transformative AI futures (alongside Land Value Taxes). They have widespread implementation and mature enforcement infrastructure, and they remain resilient even when labor income collapses, since they apply to spending (which would increasingly come from firms) rather than earnings. However they mention one major caveat is that their incidence is flat or regressive, and may place disproportionate burdens on low-income households without corrective design.
Dario Amodei
May 2025
Amodei has suggested a "token tax" — a percentage of revenue per AI model use — which would function as a specialized consumption tax on AI services. This framing treats AI usage as a taxable base analogous to other taxed transactions (like sales/VAT).
Real-world precedents:
According to OECD data, 175 countries operate some form of VAT, which generates on average 20.8% of total tax revenue across OECD nations.
The United States remains the notable exception without a federal VAT, relying instead on state and local sales taxes (approx. 32% of state collections).
To address digital enforcement challenges, the European Union adopted the "VAT in the Digital Age" (ViDA) package in 2025, introducing real-time e-invoicing and platform-based collection rules.
© 2026 Windfall Trust. All rights reserved.