The timing was not accidental. Senator Bernie Sanders introduced the American AI Sovereign Wealth Fund Act as Anthropic — the AI safety company behind Claude, widely expected to be one of the most significant technology IPOs in years — has been the subject of intense speculation about when it will go public and at what valuation. The juxtaposition was deliberate: here is a company that may soon be worth hundreds of billions of dollars, trained on data produced by millions of Americans, built with research that drew on decades of publicly funded academic work. Who should own it?
Sanders's answer is: the American people, at least in part. His bill would establish a sovereign wealth fund — a government-owned investment vehicle — that would acquire equity stakes in major artificial intelligence companies, with the returns distributed as dividends to American citizens. It is the most significant proposal yet for public ownership of AI infrastructure, and it has landed in a Washington that is simultaneously fascinated by AI's potential and deeply divided about how to govern it.
What the Bill Would Do
The American AI Sovereign Wealth Fund Act, as introduced, would create a federal fund capitalized through a combination of a tax on AI company revenues above a certain threshold and through the issuance of government bonds. The fund would use this capital to acquire equity stakes in qualifying AI companies — defined by their scale, their revenue, and their significance to American AI infrastructure.
The returns from those equity stakes — dividends, capital gains from share appreciation — would be distributed to American citizens as an annual payment, modeled loosely on Alaska's Permanent Fund dividend, which distributes a share of the state's oil revenues to every Alaskan resident. The explicit parallel is pointed: oil was the transformative resource of the twentieth century, and states that owned their oil resources were able to share the wealth broadly rather than allowing it to accumulate entirely to private capital. AI, Sanders argues, is the transformative resource of the twenty-first century. The same logic applies.
The bill includes provisions for the fund to exercise the governance rights that come with equity ownership — voting on corporate matters, participating in boards, using ownership to push for labor standards and safety practices. This dimension of the proposal has attracted as much attention as the dividend provision: it would give the federal government a direct voice in how major AI companies are operated.
The Anthropic IPO Context
Anthropic has not confirmed IPO plans. But the company, which has raised billions of dollars in investment from Amazon, Google, and other technology players, has a valuation that analysts estimate in the range of $60 to $100 billion and a product — Claude, its AI assistant — that has achieved widespread enterprise and consumer adoption. The conditions for a public offering are present.
When Anthropic does go public, it will join a category of AI companies whose collective market capitalization is expected to reach into the trillions over the next decade. The wealth creation will be extraordinary. Who captures that wealth — the founders, the venture capital investors, the institutional shareholders who buy in at IPO — is a distributional question with profound implications for American inequality.
Sanders's bill intervenes in that question explicitly. By establishing a mechanism for public ownership before the wealth is fully created and concentrated, it attempts to solve a distributional problem before it becomes a political crisis rather than after. The argument is that waiting until AI wealth has already been created and concentrated, and then trying to redistribute it through taxation, is both politically harder and practically less effective than establishing ownership stakes at the beginning.
The Opposition and Its Arguments
The bill faces opposition from multiple directions. Technology industry groups have argued that government equity ownership would introduce political considerations into corporate governance that would distort decision-making and slow innovation. Libertarian-leaning economists have argued that the fund would represent an inappropriate extension of government power into private markets. Even some Democrats who agree with the distributional goals of the bill have expressed concern about the governance implications of the federal government holding voting shares in the country's most significant technology companies.
The constitutional questions are also real. Federal ownership of equity stakes in private companies is not unprecedented — the government held equity in banks and auto companies during the 2008 financial crisis — but a permanent, ongoing equity acquisition program of the scale Sanders envisions would be novel and would face legal challenges.
The political path for the bill in the current Congress is, to put it gently, narrow. Republicans control both chambers and have shown no appetite for measures that expand government ownership of private enterprise. The bill is unlikely to pass in its current form. That is not, by itself, a reason to dismiss it — legislation that fails in one Congress sometimes shapes the debate in ways that produce results in a later one, and Sanders's proposals have a track record of moving from the margins of Democratic politics to the mainstream over time.
The Real Debate Underneath the Bill
The American AI Sovereign Wealth Fund Act is, on one level, a specific legislative proposal. On another level, it is an opening bid in a debate that Washington has been avoiding: the question of who owns the AI future and what obligations AI companies have to the public whose data, whose creative work, and whose publicly funded research made their products possible.
That debate is coming regardless of whether Sanders's specific bill advances. The Anthropic IPO, when it happens, will make the wealth creation visible in a way that valuations and private funding rounds do not. The concentration of AI capability in a small number of companies, and the concentration of AI wealth in a small number of investors, is going to produce political pressure for some kind of public claim on the technology. Whether that claim takes the form Sanders proposes, or some other form, or no form at all, is the question that the next several years of Washington politics will have to answer.