The great AI fortune was never a solo project. It was built on decades of publicly funded research, data scraped from the entire internet, and a legal infrastructure that let companies capture the value while the public absorbed the costs. That’s the part Daniel J. Pilla leaves out in his National Review piece, The Plan to Confiscate AI Company Stock, which treats Bernie Sanders’ proposed AI sovereign wealth fund as a Marxist seizure of private property. Pilla argues that a 50 percent equity transfer to the government is theft, a dangerous precedent that will turn business decisions over to bureaucrats and smother the incentives that built America’s innovation economy.
Let me grant what he gets right, because conceding the true half is what earns the right to be unkind with the rest. The proposal is unusual — a direct equity transfer rather than a tax — and the governance design of a seven-member commission of political appointees deserves real scrutiny. If you hand board seats to people chosen by the president, you’re inviting the kind of political capture Pilla warns about. That part is legitimate, and it’s worth designing against. A properly designed governance firewall would look like an independent board with fixed, staggered terms, no removal authority by the president or Congress, and a statutory mandate limited to fiduciary duty — maximizing return for the fund’s beneficiaries — rather than advancing policy goals. That’s not a hard institutional problem. It’s been solved before.
Now for what he leaves out. Which is nearly everything that matters.
Pilla’s central premise is that AI companies produced their wealth from nothing but their own genius and risk-taking, and that any public claim on it is confiscation. That premise is false. Start with the dollar figures. The basic research that made large language models possible — neural networks, backpropagation, the transformer architecture — was funded overwhelmingly by the U.S. government. DARPA, the National Science Foundation, the Department of Energy wrote the checks. University labs on federal grants produced the breakthroughs. The companies that later turned that research into products didn’t pay for it. They inherited it.
Now the data. These models are trained on the internet — billions of text and images produced by the rest of us. Every Wikipedia entry, every news article, every blog post, every photo, every Reddit comment. The “creative work of hundreds of millions of people,” as the Sanders bill accurately calls it. AI companies scraped the commons and called it training data. They didn’t pay for that either.
Then there’s the legal architecture: limited liability, the fair-use doctrine’s contested application to training data, the absence of a federal data-privacy regime that would have constrained scraping, and a patent system that lets companies lock up publicly funded inventions. Every one of these is a government-created privilege that reduces risk and increases return. The risk Pilla celebrates — the entrepreneur who mortgages his house — is already heavily subsidized by the state.
Pilla writes that “the federal government stands in the unique position of an uninvested partner.” But the government was always an investor — just one that never asked for equity. It invested in the research, the infrastructure, the legal system, and the educated workforce. The only novelty here is that the proposal asks for a return on that investment. Pilla calls it “outright theft.” I’d call it a belated invoice for services rendered.
Now, what would the fund actually build? It would hold a 50 percent stake in the largest AI companies — not run them, hold the shares — and distribute the dividends to every American. Estimates suggest about $1,000 a year per person. Five percent of the fund’s annual return goes to direct payments; the rest funds health care, education, housing, and environmental programs — things every other wealthy country already provides without descending into serfdom.
We have working examples. Norway took its oil revenue and created a sovereign wealth fund now worth over $2 trillion — about $390,000 per citizen. The country didn’t become a bureaucratic nightmare; it’s one of the most dynamic economies on earth. Alaska mails a Permanent Fund dividend check to every resident every year, and has since 1982. The Bank of North Dakota, a state-owned bank, has been profitable every year since 1919. These are not radical experiments. They’re boring infrastructure that works.
Pilla fears that government ownership of shares will lead to political control over information. It’s a real concern, worth designing against. But the current arrangement — where eight men control the platforms that increasingly shape what Americans see, read, and believe — is not a model of democratic accountability. The question isn’t whether someone will have power over these companies. It’s who: a handful of billionaires, or the public through carefully designed institutions.
Pilla also warns that the same logic could be applied to pharmaceutical companies, energy producers, home builders, financial institutions — any “systemically important” sector. It could. The question is whether the triggering conditions that justify a public equity stake in AI exist in those sectors to the same degree. The AI industry was built on a specific combination: decades of direct government research funding, training data scraped from the public commons without compensation, and legal privileges that let companies capture value they did not create. That combination is not present in pharma, energy, or finance — at least not yet. The specificity of the conditions is the safeguard against the slippery slope. If and when those conditions arise elsewhere, the same logic would apply. That’s not a precedent to fear; it’s a principle of fairness.
The end of innovation? Ask a Norwegian entrepreneur whether the oil fund stopped people from starting companies. It didn’t. Good institutions make it easier to take risks, because failure isn’t fatal. When you own a piece of the country’s wealth, you’re more free, not less.
Pilla ends with a warning about total dependence on government. But we already have total dependence — on a job market that treats millions of willing workers as disposable, on a health care system that ties insurance to your employer, on a housing market that auctions shelter to the highest bidder. The AI fund is one proposal to build a small piece of independence on the other side: a direct stake in the wealth the country generates, not dependent on a paycheck or a stock portfolio.
Call the thing by its right name. It isn’t confiscation. It’s the public finally asking for its equity back, and building a shared stake in the next industrial revolution. Pass the fund, fix the governance, and give every American a direct stake in the wealth their taxes and their data made possible. That’s not the end of American innovation. It’s the beginning of making sure innovation benefits the Americans who made it possible.