Bernie Sanders wants to seize half the stock of major AI companies and hand them to a sovereign wealth fund. It's a bold move designed to answer a question he posed in the New York Times: Will the future of humanity be determined by a handful of billionaires with no democratic input?
That question matters. The concentration of power, wealth and control in the hands of tech oligarchs ranks among the most pressing threats to democratic governance worldwide. Sanders is right to flag it as urgent.
But the remedy he's proposing creates a different kind of problem altogether.
The senator's logic is straightforward on its surface. Government ownership of AI company stock would give the public a seat at the table through board representation and voting shares. It would also redirect trillions in AI-driven wealth back to everyday Americans instead of letting it flow entirely to Silicon Valley billionaires. Both goals sound reasonable.
Yet there's a fatal contradiction baked into the approach. Once government becomes a major shareholder in these companies, the incentives flip. A government fund manager watching Nvidia grow from five trillion to ten trillion in valuation will have every reason to support that growth, just as fiercely as a private investor would. Regulatory hurdles that slow AI deployment? Inconvenient. Worker protections that cut into margins? Problematic. Foreign markets opening up for sales? Excellent news for the fund.
History offers a cautionary lesson. Norway's sovereign wealth fund, the world's largest, owns substantial stakes in major oil companies. Yet Norway has struggled to convert that ownership into climate action. The Norwegian government's financial dependence on those same companies has actually inhibited climate policy. American public pension funds face the same dynamic: fiduciary duty to maximize returns overwhelms any intention to steer corporate behavior toward the public interest.
What Sanders is really trying to accomplish can be achieved more cleanly through two separate mechanisms.
First, wealth redistribution. Taxation works. Senator Elizabeth Warren has proposed an excise tax on data center energy consumption. Others have suggested an AI token tax. These approaches funnel AI-generated wealth back to the public without entangling government finances in corporate valuation.
Second, public influence over AI development. Here's where a public option becomes powerful. Instead of seizing private companies, governments can build their own AI systems operated by public institutions under democratic control. The goal isn't to eliminate corporate AI but to establish a competitive baseline that private offerings must meet or exceed to win business.
Switzerland is already experimenting with this model. Apertus, a large language model developed by Swiss public servants and university researchers, uses licensed training data and public supercomputing infrastructure powered by renewable energy. It won't win performance benchmarks against OpenAI or Anthropic today. But it demolishes them on transparency, sustainability and regulatory compliance including copyright adherence. That's not a marketing advantage now. It becomes one as regulation tightens and clients demand accountability.
This approach applies real competitive pressure on corporate actors without the perverse incentives of government shareholding. Private companies racing to match or beat public baselines face genuine pressure to behave responsibly. They can't ignore labor standards or safety concerns if a publicly operated alternative keeps improving. They can't cut corners on transparency if government-backed systems set a higher standard.
The political calculus around Sanders' proposal is worth considering too. The Trump administration and AI billionaires have signaled openness to it. That alignment should raise questions about why tech moguls would embrace such a dramatic intervention in their own companies. The answer is probably that they expect favorable government policies protecting that newfound investment will more than compensate for ceding equity.
Sanders is politically savvy enough to recognize these tradeoffs. The appeal of seizing a rare moment of bipartisan alignment on AI policy matters. But expedience shouldn't override structural risks that history demonstrates are real.
Author James Rodriguez: "Sanders identified the right problem but prescribed a remedy that solves it by corrupting the doctor."
Comments