Bernie Sanders AI Sovereign Wealth Fund Bill

The core design choice in Sanders's bill is its strangest feature. Rather than simply taxing AI profits and cutting checks, the legislation imposes a one-time 50% tax payable in stock, seeding a sovereign wealth fund estimated at roughly $7 trillion, then mandates a 5% annual dividend that flows to citizens as payments exceeding $1,000 a year ^[1]. The choice is deliberate: Sanders is arguing the public should own a permanent equity stake in companies built on collective human knowledge, not merely collect a recurring levy ^[8]. But that distinction draws the sharpest critique from within his own camp. Stephanie Kelton, an economist and former Sanders adviser, argues the architecture is unnecessary, noting that 'the federal government doesn't need to seed an investment fund with assets in order to disburse dollars to Americans' ^[3]. Her point cuts to the bill's heart: if the goal is putting cash in pockets, Washington can tax AI firms directly and skip the fund entirely. By contrast, the fund's defenders see ownership itself as the point, granting the public both dividends and, through a seven-member Independent Commission for Democratic AI, voting power over the labs ^[1]. The mechanism, in other words, is not an implementation detail; it is the ideological argument.

The most surprising feature of this fight is who agrees with it. Public ownership of AI has produced a cross-spectrum alignment that scrambles the usual battle lines. Just weeks before the bill dropped, reports surfaced that the Trump administration was separately negotiating a government stake in OpenAI, and Trump has broadly embraced government investment in private firms ^[5]. OpenAI CEO Sam Altman met with Sanders and supports the general concept of public equity in AI, even offering to advocate for it, while objecting that the 50% figure is too high ^[2]. That overlap is striking given the venom from the right elsewhere: the Wall Street Journal editorial board branded the plan a 'road to AI state socialism,' and critics frame it as confiscation ^[3]. Yet the seam runs deeper than the labels suggest. The reaction on social platforms reflected the same split, with even an ideological opponent inside the administration conceding the proposal resonates across the spectrum while rejecting the socialism framing. The takeaway: the question is no longer whether the public should hold a stake in frontier AI, but how large the stake should be and who controls it, an argument now joined by figures who agree on almost nothing else.

There is a basic accounting problem at the center of the bill. The named targets, including OpenAI, Anthropic, and xAI, are companies Sanders himself acknowledged are currently unprofitable ^[2]. A fund built on equity in money-losing firms and a dividend that depends on stock appreciation invites an obvious question, raised across both expert commentary and online debate: is there any 'wealth' to seize yet? ^[6]. The proposal is widely described as unlikely to become law, though analysts note it advances the broader idea of offsetting AI's labor disruption through government stakes ^[7]. The dominant register of public reaction captured the tension precisely, with supporters embracing the principle while doubting both its political odds and its arithmetic. A frequent contrarian thread held that nationalizing unprofitable startups would damage the United States as a place to found or invest in a company. The feasibility gap is the bill's most concrete vulnerability: even sympathizers struggle to explain how a stake in firms that don't yet turn a profit funds a guaranteed annual check to every American.

The bill's two loudest framings sit on opposite poles. Critics led by the Wall Street Journal editorial board argue the 'one-time 50% tax' is really expropriation that would violate the Fifth Amendment's takings clause, a constitutional challenge that could stall the plan before any dividend is paid ^[3]. Samuel Hammond of the Foundation for American Innovation extends the warning to governance, cautioning that 'even if taking partial ownership of frontier AI companies can make sense on paper, in practice it's a recipe for political favoritism and corruption' ^[3]. Tosin Akintola of Reason adds that government veto power would throttle innovation, writing that 'if every advancement in AI is subject to government approval, as Sanders proposes, it's unlikely that breakthroughs like these would be achieved' ^[4]. Against the takings argument, the online left advances a mirror-image moral claim: that a 50% stake is not seizure but restitution, owed because the labs trained their models on scraped art, code, and books without consent, the same 'collective human knowledge' Sanders invokes ^[8]. That framing reframes the entire dispute, casting the government's stake as repayment of a debt the companies already incurred, and it is the argument doing the most to unify the bill's supporters even as legal experts question whether it can survive court.