Bridgewater Proposed 35% Tax on AI Tokens

The investment firm argued a new levy on AI usage could fund public equity and mitigate machine-labor disruption.

Updated on Oct. 5, 2026 in Artificial Intelligence

Isometric editorial illustration of a high-density server rack stack, representing the infrastructure behind AI token usage taxation. — Bridgewater Associates has proposed a 35% tax on artificial intelligence token usage, aiming to redistribute economic gains and mitigate the labor displacement caused by AI. AI Illustration. Upload story photo >

Live Poll

Should AI companies be taxed to provide direct financial equity to everyday citizens?

Bridgewater Associates has proposed a 35% token tax on artificial intelligence usage to fund a citizen equity program. The firm suggests this mechanism would generate $600 billion in revenue by 2030 and distribute shares of AI companies to the public.

Why it matters

The proposal aims to curb incentives favoring machine labor over human workers while addressing societal disruption. By redistributing wealth, the firm seeks to ensure broad public participation in the economic gains generated by major AI developers.

The firm projects that two major AI companies, OpenAI and Anthropic, will control 35% to 50% of the world's compute power within two years. Bridgewater suggests a 5% threshold of total compute control as a trigger for heightened regulatory oversight.

The players

Bridgewater Associates

A major investment firm that manages large-scale institutional assets and provides research on global economic trends.

Greg Jensen

A top executive at Bridgewater Associates who has engaged in policy discussions regarding the oversight of leading AI entities.

OpenAI

A prominent developer of large-scale AI models projected to hold significant control over global compute capacity.

Anthropic

A research-focused AI company developing large language models identified as a primary player in future compute concentration.

The details

The proposal calls for creating a specialized division within the Internal Revenue Service to manage the token tax. To address safety, the firm recommends that regulators conduct mandatory sworn interviews with AI lab staff to assess potential risks. The policy explicitly targets the imbalance between human and machine labor productivity by making high-intensity AI operations more costly.

September 2026: Bridgewater published its essay outlining the AI policy.
October 2026: Greg Jensen discussed regulatory oversight of major AI companies.
2030: Target date for reaching the projected $600 billion in tax revenue.

The Tech Race

This proposal marks a departure from existing tech-regulation trends by shifting the focus from data privacy to the concentrated control of compute infrastructure. It follows a pattern set by the Inflation Reduction Act’s drug-price negotiation provisions by attempting to leverage government oversight to capture economic value from dominant private firms.

The proposal currently exists as a policy recommendation under discussion on Capitol Hill, with no immediate changes to existing tax or software costs. If enacted, the policy would fundamentally alter the cost structure for AI-heavy workflows and introduce new equity dividends for citizens.

The takeaway

Bridgewater is signaling that the concentration of compute power is the next major hurdle for macroeconomic policy. Readers should monitor upcoming congressional testimony or formal legislative drafts related to AI compute usage to see if this 35% tax threshold gains traction in formal committee debates.