A hedge fund that manages money for the ultra-wealthy is now loudly arguing that AI companies should be taxed and the proceeds handed directly to the American public. Bridgewater Associates, Ray Dalio's famously secretive Connecticut firm, has been holding talks with Washington policymakers from both parties about proposals to stop AI from eating the economy whole. Nobody expected this call to come from a billionaire's house.
The Pitch: Tax the Robots, Pay the People
Here's what Bridgewater is actually proposing. A 35% tax on AI token usage, meaning every time an AI processes a request, the company running it owes a cut to the government. According to the New York Post, the revenue from that tax could reach $600 billion by 2030. That money would be used to buy shares in AI giants and distribute them directly to the public.
Not a monthly check. Not a government program you have to apply for. Actual equity. Ownership stakes in the companies hoovering up all the economic gains right now. The logic is pretty straightforward: if AI is going to make a handful of corporations obscenely wealthy while displacing millions of workers, maybe the workers should at least get a piece of the action.
Bridgewater's chief investment officer Greg Jensen and CEO Nir Bar Dea published a lengthy essay last month calling for 'immediate policy action' to realize AI's potential while avoiding what they described as 'widespread societal disruption and catastrophic safety accidents.' That essay has since caught the attention of lawmakers on Capitol Hill, the Post reports, and informal talks have begun.
Why This Is Different From UBI (And Why That Distinction Matters)
You might be thinking: isn't this basically universal basic income with extra steps? It's not, and the difference is politically important. UBI, championed by tech figures including Sam Altman and, yes, Elon Musk, gives everyone a regular government payment. Bridgewater's proposal gives everyone a stake in the companies themselves.
Jensen was blunt about why he prefers equity to a check. 'You're still dependent on a bureaucrat deciding how much and when you get your check,' he told the Post. 'If we actually distribute the equity to all citizens it once again takes the power away from the politicians to use that in ways that you might not want and it gets power directly to the citizens.'
That is, whether you agree with it or not, a coherent ideological argument. It's also one that sidesteps the endless Republican talking point about government dependency, because you can't cut someone's dividend check the way you can cut a welfare payment. Citizens would hold assets. That's different.
The Payroll Tax Problem Nobody Talks About
Bridgewater makes a second argument that has gotten less attention but deserves more. AI-generated work is increasingly replacing human labor. Human workers pay payroll taxes. AI does not. So right now, companies are being actively incentivized to replace people with machines, and the tax code has nothing to say about it.
'I think it just is common sense that we don't want to incentivize machine labor over human labor,' Jensen said. He's calling for a new IRS division to enforce the token tax, which is either a bold structural reform or a bureaucratic nightmare waiting to happen, depending on your disposition.
Jeremy Bearer-Friend, an associate law professor at George Washington University, told the Post he sees another underappreciated benefit here: public voice in corporate governance. 'An AI equity tax could give the public a role in boardroom decisions about public safety,' he said. Shareholders get to vote. Right now the public has no vote on what the most powerful technology in human history does next.
Monopoly Power and the 5% Threshold
Jensen didn't stop at the token tax. In comments to The Information this month, he argued that any AI company controlling more than 5% of US or global AI computing power should face heightened regulatory oversight, similar to how systemically important banks get extra scrutiny after the financial crisis.
His reasoning is hard to argue with on the merits. 'In two years, OpenAI and Anthropic are going to control 35% to 50% of the world's compute,' Jensen told The Information. 'That's a crazy outcome for a society to allow on something as powerful as compute. Would we let one entity control that much of some other form of energy or commodity?'
No. We would not. We have antitrust law specifically because we learned that lesson with oil, railroads, and telecommunications. The question is whether those lessons survive contact with an industry that has grown so fast, and donated so much money to so many politicians, that enforcement feels almost quaint.
The Catch: Follow the Money
Let's not be naive about what's happening here. Bridgewater is a massive investment firm. It has its own positions in AI. Jensen acknowledged as much in the firm's report, stating that Bridgewater would be 'disproportionately subject to the taxes and regulations we recommend' but supports them anyway because AI needs long-term public buy-in to thrive.
AI observers quoted by the Post were more direct: ensuring smooth adoption of AI limits public backlash and prevents the kind of heavy-handed regulation that might actually hurt Bridgewater's investments. A well-managed, publicly legitimate AI industry is a more stable investment environment. The altruism and the self-interest are not mutually exclusive.
There are also real implementation problems. Cyril Gorlla, founder of AI advisory firm CTGT, told the Post that a token tax faces a fundamental infrastructure challenge: there is no universal metering standard for AI the way there is for electricity. 'You have a meter box' for utilities, Gorlla explained. For AI, nobody has built that yet. Which means the plan that sounds elegant on paper requires inventing an entire enforcement mechanism from scratch.
The Dingo Take
$600 billion. That's the projected revenue from Bridgewater's token tax by 2030, assuming anyone in Washington actually does something about it, which is the part of the sentence that should make you nervous. Congress spent the better part of a decade failing to regulate Facebook after it burned democracy for ad revenue. The idea that they will move with urgency on a 35% AI tax, against an industry that has become one of the biggest lobbying forces in the country, requires a degree of optimism that years of watching American governance has made impossible to sustain.
But here's what's strange and worth sitting with: the most coherent, structurally serious proposal for sharing AI wealth currently on the table is coming from a hedge fund. Not labor unions. Not progressive think tanks. Not the Democratic Party. A firm founded by a billionaire. That tells you something about the state of the economic left in this country that should be embarrassing to everyone involved. When Ray Dalio's people are out-flanking you on worker equity, something has gone badly wrong with your message.
Jensen is right that time is running out. The essay he and Bar Dea published said it plainly: 'Absent intervention today, mitigation of that danger will become nearly impossible as diffusion of the technology accelerates and models themselves become capable of improving and acting autonomously.' That's not a think piece. That's a warning. The window for the public to get a seat at this table is closing fast, and the people most responsible for keeping it open are busy doing absolutely nothing.
Comments