Sen. Bernie Sanders is moving the AI wealth-sharing argument from essays into legislation. His American AI Sovereign Wealth Fund Act would create a public fund financed by a one-time 50% stock tax on large AI companies. The threshold in the bill summary is $200 million or more in related annual gross receipts from AI activities.
The proposal is deliberately aggressive. It is not a normal corporate tax, a windfall-profits tax, or a voluntary benefit plan. It would require qualifying companies to transfer equity interests, which the Treasury would move into a new sovereign wealth fund. Sanders argues that because AI is built on public knowledge and creative labor, the wealth it creates should be shared by the public.
The bill is unlikely to become law in its current form. That does not make it irrelevant. It gives a concrete shape to a policy debate that has been floating around AI for years: if a small number of companies capture extraordinary AI-driven economic gains, should the public receive direct upside?
The mechanism is equity, not cash
The bill summary says the tax would apply to companies engaged in AI activities related to AI data centers, AI computing infrastructure, AI services, and advanced robotics. Covered companies must have at least $200 million in related annual gross receipts. The one-time tax would be imposed 90 days after enactment.
That design is the core controversy. A cash tax would reduce liquidity. A stock tax changes ownership. If the government receives 50% of a company’s equity, it becomes a major shareholder with economic and governance consequences.
Sanders’ June 1 op-ed described the goal plainly: public ownership of half of the largest AI companies. He argues the federal government should have voting power and board representation to block decisions that hurt citizens and push policies that help them. AP reported that Sanders estimates the fund could be worth nearly $7 trillion and generate money for direct payments and public programs such as health care, education, and housing.
Those estimates should be treated as the proposal’s claims, not as an outcome. A sovereign wealth fund’s value would depend on which companies qualify, how their valuations hold up, how the stock transfer is implemented, and whether courts and Congress allow the structure to survive.
This is a sharper version of a familiar idea
Public participation in AI upside is not only a Sanders idea. OpenAI has talked about a public wealth fund concept. Anthropic has written about national sovereign wealth funds with stakes in AI. Sanders also points to Norway’s oil fund and Alaska’s Permanent Fund as examples of public wealth-sharing mechanisms.
The difference is compulsion and control. A voluntary public-benefit plan can be designed around a company’s preferred structure. Sanders’ proposal would use tax law to force equity transfer from qualifying AI companies, including firms where AI is only part of a larger business.
That raises hard implementation questions. How would the law separate AI and non-AI revenue inside diversified companies? How would it treat subsidiaries, private companies, foreign ownership, employee stock, and future financing rounds? What happens to startups that cross the threshold? Would a forced equity transfer survive constitutional challenge? The bill summary answers some mechanics, but not the full economic system around them.
The political signal matters
The timing is not accidental. AI companies are racing toward extreme valuations while workers, creators, and public officials worry about job displacement, data use, energy demand, and concentrated power. Sanders is giving those anxieties a concrete target: ownership.
That will draw predictable opposition from investors and founders who see forced public equity as a startup deterrent. It will also draw interest from people who think existing tax and labor policy cannot keep up with AI-driven concentration. Both reactions are part of the story.
The more useful policy question is smaller than the slogan. If AI creates large productivity gains, which mechanisms distribute those gains without destroying the incentive to build? Options include ordinary taxes, worker ownership, public funds, licensing revenue, compute levies, public-option models, procurement rules, and direct equity. Sanders has chosen the most forceful version.
What to watch next
Watch the bill text, co-sponsors, committee path, and legal analysis. The details will matter more than the headline. The definition of “applicable AI company” could decide whether the bill targets only frontier labs and AI infrastructure firms or sweeps in large software, cloud, robotics, and semiconductor businesses.
Also watch how AI companies respond. Some may reject the proposal outright while still supporting softer public-benefit mechanisms. Others may use the bill as evidence that the industry needs to offer credible sharing plans before lawmakers write harsher ones.
For readers tracking how AI company structure is becoming a policy issue, see our AI company tracker and AI model leaderboard.
