Senator Bernie Sanders announced plans for legislation this week that would force the largest AI companies to hand 50 percent of their equity to the American public, naming OpenAI, Anthropic, and xAI as primary targets in a proposal that landed the same day Anthropic confidentially filed for an IPO.
The timing could not have been more pointed. On June 1, Sanders published a New York Times op-ed outlining what he is calling the American A.I. Sovereign Wealth Fund Act, a proposal that would impose a one-time 50 percent tax paid not in cash but in stock on the country's biggest AI companies. The federal government would receive those shares, hold them through a newly created sovereign wealth fund, seat equal representation on each company's board, and retain voting power to block corporate decisions deemed harmful to the public. The proceeds, Sanders argues, should eventually flow back to Americans as direct payments and expanded public benefits. That same day, Anthropic confidentially submitted a draft S-1 registration statement to the Securities and Exchange Commission.
The intellectual foundation Sanders is working from is straightforward, even if the policy instrument is aggressive. AI systems are trained on text, code, images, and creative work that humanity collectively produced. The productivity gains those systems unlock belong, in his framing, to the people whose labor generated the training data, not exclusively to the venture firms and founders who assembled the compute and wrote the architectures. That argument is not new, but it has rarely been attached to a proposed federal ownership transfer with this level of specificity.
What makes the proposal unusual, and worth taking seriously despite the long legislative odds, is that Sanders acknowledges it echoes ideas already floated from within the industry itself. OpenAI had previously proposed creating a public wealth fund that gives every citizen a stake in AI-driven economic growth. Anthropic's CEO Dario Amodei had similarly endorsed the concept of national sovereign wealth funds holding AI stakes. Sanders noted in his op-ed that even President Trump's executive orders have gestured toward an American sovereign wealth fund. The senator is not working from the fringe. He is taking language the AI industry has used as rhetorical cover and turning it into a legal obligation.
For venture capital firms, the proposal immediately raises the question of how to model legislative risk into late-stage AI deals. With Democrats in the minority in both the Senate and the House, the American A.I. Sovereign Wealth Fund Act is unlikely to pass in its current form once introduced. But the legislative risk calculation is no longer zero. Senators Elizabeth Warren and Representative Greg Casar separately floated proposals last week to tax AI productivity gains for jobs programs and universal healthcare. The policy pressure is accumulating and each new proposal raises the probability that some version of AI profit-sharing, even a diluted one, survives eventual negotiations.
For founders raising their Series B or C right now, the practical effect may already be showing up in term sheets. Investors pricing a 50 percent forced equity transfer into a downside scenario do not need the proposal to pass to adjust their return models. They need only to believe the conversation is real. When a sitting senator names your company in a New York Times op-ed and outlines board seats and voting blocks, the conversation is real.
The Anthropic IPO filing sharpens this considerably. Going public while a prominent Senate proposal argues for public ownership of half your equity is not a comfortable situation for underwriters or institutional investors. The company may need to address the risk in future public filings if the legislation is formally introduced and considered material. That kind of disclosure, buried deep in an S-1, would be the first time many mainstream investors encounter the Sanders framework, and it would not be in the context of an op-ed but in a legal document warning them that Congress could compel a 50 percent stock transfer to the federal government.
The xAI situation adds another layer. Following SpaceX's reported $1.25 trillion combination with xAI earlier this year, the entity that Sanders named in his proposal may no longer be as simple to define as a standalone startup. That structural complexity alone illustrates how fast the AI landscape is moving relative to the legislative drafting process, and why precise threshold definitions in any final bill text would matter enormously. Whether coverage is determined by valuation, revenue, compute spend, or market influence will determine which companies actually fall under the law.
The broader market implication here is not whether this specific proposal passes. It almost certainly will not in its current form. The implication is that public ownership of AI infrastructure is now a mainstream policy debate, not a fringe position, and that the venture industry has not yet built consensus on how to respond to it. The firms that model this seriously, think through the structural scenarios, and engage with policymakers rather than dismissing the conversation will be better positioned than those waiting for the proposal to die in committee. It may die. But the argument behind it is not going anywhere.
Also read: GoPro is running out of road as the AI memory crunch turns a camera company into collateral damage • Tencent's plan to embed a full AI agent inside WeChat puts 1.4 billion users at the center of China's platform war • Luma AI opens its physical AI lab to the public as the video generation startup bets its world model advantage translates to robotics
