The News
Sam Altman has floated giving the US government a 5% stake in OpenAI, worth roughly $42.6 billion at the $852 billion valuation from OpenAI's March round, per CNBC. The larger idea behind it is to have the leading American AI companies each route 5% of their equity into a public vehicle modeled on the Alaska Permanent Fund, per TechCrunch. Talks are conceptual, Altman has carried them to President Trump, Commerce Secretary Lutnick, and Treasury Secretary Bessent, and any real version would likely require an act of Congress.
The View
The Alaska comparison is doing a lot of work, and it does not hold.
The Alaska Permanent Fund works because Alaska taxed something it already owned. The oil sits under public land, the state has a sovereign claim to it, and the fund converts a depleting public resource into a permanent financial one. AI equity has no such public claim behind it. It is stock in private companies that a handful of investors and employees own. Routing 5% of it to Washington does not monetize an asset the public already holds. Private firms would be handing the government a stake, and the question worth asking is what they expect back for it.
The honest name for 5% given to your regulator is an insurance premium. OpenAI, Anthropic, and the rest are staring at antitrust exposure, an unsettled copyright picture, export controls that already reached in and switched a model off this spring, and an administration that can make life hard. An equity gift buys goodwill no lobbying budget can, and it buys it from the exact people who write the rules. Whether you file that under public benefit or regulatory capture depends on where you stand. The mechanism is the same either way.
Then there is the conflict the arrangement builds in. Make the government a shareholder and you have asked one entity to both regulate the labs and profit from them. When Commerce ordered a frontier model shut down this spring, it cost the vendor real money. Would a Commerce Department holding 5% of that vendor move as fast the next time? A state that owns a slice of what it polices has a standing reason to police it gently, which is the opposite of what oversight is for.
It also entrenches the incumbents. Only the biggest labs are large enough to make a 5% donation meaningful, and only they collect the implied protection that comes with it. The startup trying to unseat them gets no seat at that table. A public stake in the winners, structured by the winners, is a moat with a flag planted on top.
Room for Disagreement
The serious case is real. If AI automates a large share of human labor, the gains concentrate in whoever owns the models, and a broad public claim on that upside is a reasonable hedge against an ugly distribution. Better to negotiate the stake while the wealth is still forming than to claw it back after it hardens. The Alaska fund is genuinely popular and genuinely well run, it pays every resident a dividend, and it has survived fifty years of politicians who would have loved to raid it. A 5% sovereign stake might distort less than the heavy-handed regulation that is the realistic alternative. I hold that case in one hand. In the other I hold the fact that the people proposing it are the people it protects.
Notable
- 5% of OpenAI runs to about $42.6 billion at the $852 billion March valuation.
- The proposal envisions Anthropic, Google, and Meta ceding similar stakes; none has agreed.
- A real fund would likely need Congress, not just an executive handshake.
Bottom Line
A public stake in the AI labs is either the public capturing AI's upside or the labs buying protection with equity they minted, and right now the same 5% does both. Watch who writes the enabling law, because the details decide which one it turns out to be.
— Hank