Senator Bernie Sanders introduced the American AI Sovereign Wealth Fund Act on June 5, 2026, a bill that would transfer a 50% ownership stake in the country’s largest artificial intelligence companies to the American public. The proposal is the most aggressive effort yet to treat frontier AI as public infrastructure rather than private property.
The 50% Equity Levy
The centerpiece of the bill is a one-time tax paid in stock. Rather than writing a check, covered companies would surrender shares equal to half of their outstanding equity to a newly created public trust.
- 50% of outstanding equity transferred to the public
- One-time levy paid in shares, not cash
- Valuation threshold determines which labs are covered
- Named targets include OpenAI, Anthropic and xAI
The levy is designed to capture value at the moment a company reaches scale, before founders and early investors can lock in generational fortunes.
Voting Rights and Dividends
The American AI Sovereign Wealth Fund would be governed by a board of trustees and would hold voting shares on behalf of citizens. Supporters describe two benefits. First, the public would gain a formal voice in decisions about safety, deployment and access. Second, the fund would pay dividends, with Sanders proposing that payouts be directed into Social Security to shore up the program.
The Argument for Public Ownership
Sanders argues that frontier models were built on taxpayer-funded research, public datasets and decades of public investment in computing. In his framing, the American people supplied the foundation, so they should share in the returns rather than watch a handful of firms capture trillions of dollars in value.
The Opposition
Tech trade groups and venture investors quickly attacked the plan as expropriation. Critics warn that it would trigger constitutional takings litigation, chill investment in US labs and push founders to incorporate abroad. Industry lobbyists also argue that a government stake would politicize safety decisions.
What Happens Next
The bill has been referred to committee, where it faces near-certain gridlock in the current Congress. Even so, the proposal has already done political work: it has forced presidential candidates to take a position on AI ownership and prompted several states to study their own public-equity models.
What This Means
Whether or not the Act becomes law, it marks a shift in the AI debate. The question is no longer only how to regulate frontier models, but who should own them, and what the public is owed for the data and research that made them possible.