Should the US government take a stake in OpenAI and Anthropic?

Key points

• OpenAI and Anthropic each filed for IPOs (initial public offerings, where a private company sells shares to the public for the first time) in June 2025.
• Both companies were founded explicitly to prevent the harms of unchecked, profit-driven AI development.
• Critics now say both have become the kind of corporate giants their founders warned against.
• Some policy observers are calling for the US federal government to claim a share of the companies' stock and use the proceeds for a national fund or taxpayer dividends.
• No legislation to do this currently exists; the proposals remain at the opinion stage.

OpenAI was founded in 2015, and Anthropic in 2021, by people who genuinely worried that Google and Meta would build powerful AI with profit as the only compass. Their founders said their organisations were different: structured as safety-focused labs rather than pure businesses, working for humanity rather than shareholders.

That pitch has worn thin.

What changed?

Both labs have drifted toward the very model they were created to resist. Market money followed the technology, and with it came market pressures. Today, OpenAI and Anthropic guard investor value with the same energy as any Silicon Valley heavyweight. Both filed for IPOs in June 2025, and analysts immediately began floating trillion-dollar valuations.

The gap between the founding promise and the current reality is now wide enough that it has sparked a serious policy debate, first surfaced in an opinion piece in The Guardian.

So what are people actually proposing?

The idea is straightforward. If OpenAI or Anthropic fail to attract enough investor appetite at their IPOs, or even if they succeed, the federal government could step in and claim equity, a formal ownership stake, in the companies. That stake would sit inside a US sovereign wealth fund, a government-managed investment pool of the kind Norway has used for decades to share oil revenues with citizens.

Alternatively, the government could redirect a portion of the companies' revenues directly to taxpayers as a dividend, a regular cash payment, similar to the annual Alaska Permanent Fund cheque that Alaskans receive from oil royalties.

The logic: AI was built on publicly funded research, runs on infrastructure the public subsidised, and will reshape every part of the economy. If the gains flow only to a small group of shareholders, that represents a transfer of public wealth into private hands on a scale rarely seen.

Does this have any realistic chance of happening?

Not soon. No bill is before Congress. The proposals exist as arguments, not legislation. The US does not currently have a sovereign wealth fund, and creating one would require political will that does not obviously exist right now.

But the debate itself matters. The framing is shifting from "how do we regulate AI" to "who actually owns the value AI creates". That is a bigger question, and it will not disappear when the IPO buzz fades.

For ordinary people, the practical takeaway is simple: the companies building the tools that will affect your job, your healthcare, and your news feed are about to become publicly traded corporations. The decisions that shape those tools will increasingly answer to shareholders. Whether government has any role in balancing that is now, at least, part of the conversation.

Common questions

Could the government actually force a company to hand over stock?

In theory, yes, through legislation, though it would face significant legal challenges. In practice, no government agency currently has that authority over private technology companies.

What is a sovereign wealth fund?

It is a pool of money owned and managed by a national government, usually funded by natural resource revenues or trade surpluses, and invested to benefit citizens over the long term. Norway's fund, backed by oil money, is the world's largest.