Bernie Sanders’ AI Sovereign Wealth Fund Plan
Schneier argues Bernie Sanders is right about AI concentrating power, but says taking stock in AI firms is the wrong fix. He favors taxes and publicly run AI models instead.
Intelligence analysis by GPT-5.4 Mini
The post agrees that AI is concentrating wealth and power in a few companies, but rejects Sanders’ idea of a U.S. sovereign wealth fund built from AI company shares. The authors argue public ownership would blur the line between regulation and profit, and instead prefer taxation plus a public AI option.
The article says AI is like a giant money machine that could help a few rich people control too much. It thinks the government should tax that money and build its own safe AI tool, instead of owning part of the private companies like buying a piece of the machine.
Analysis
The core disagreement
The post starts by endorsing Sanders’ central warning: AI could leave democracy shaped by a small group of wealthy tech owners with little public input. It also agrees that the economic gains from AI should not stay entirely with a tiny class of founders and investors.
Where it parts ways with Sanders is the mechanism. Sanders proposed a U.S. sovereign wealth fund that would take a 50% stock stake in companies such as Anthropic, OpenAI, and xAI. The idea is to give the public voting power and a share of the upside. The authors argue that this would create a conflict: once the government owns part of the upside, it has incentives to help the companies grow even when growth clashes with public interest.
Why ownership can distort policy
The post says public ownership can make regulators softer, not tougher. If state returns depend on higher AI valuations, the government may be pulled toward approving more data center buildout, chip sales, workforce exploitation, and rapid deployment even when those choices are risky or inappropriate. The authors point to Norway’s oil-linked sovereign wealth fund and U.S. public pension funds as examples where financial dependence can blunt public-interest action.
A different split between profit and control
Instead of mixing ownership and oversight, the authors argue for separating the goals. To share AI wealth more broadly, they favor taxes, including ideas like an energy tax on data centers or an AI token tax. To shape AI in the public interest, they propose a public option: government-run AI models under democratic control that can compete with private offerings.
They cite Switzerland’s Apertus as a proof of concept: a public-sector large language model built with licensed data and public computing infrastructure. The post says Apertus is not yet a top benchmark performer, but it is stronger on transparency, sustainability, and regulatory compliance. The larger point is that public AI can pressure private firms without making the state financially dependent on their profits.
Key points
- The post agrees that AI is concentrating wealth and influence in the hands of a few companies and billionaires.
- It rejects Sanders’ proposal for a U.S. sovereign wealth fund funded by large equity stakes in AI companies.
- The authors argue public ownership would create a conflict between public oversight and the government’s financial interest.
- They favor taxation to redistribute AI-generated wealth and a public AI option to set a democratic baseline.
- The post points to Switzerland’s Apertus as an example of public AI built with licensed data and public infrastructure.
If the authors’ approach wins out, governments could use taxes to spread AI profits more broadly while building public AI systems as a check on private firms. That could create more transparency and give people an alternative that is designed around public rules, not just profits.
If the state ties its finances to AI company stock, it may become reluctant to slow those firms down even when safety or fairness demands it. The result could be weaker oversight, faster deployment, and more pressure to favor corporate growth over public interest.
