The 5% Solution: When the Regulator Becomes the Shareholder
OpenAI has reportedly proposed handing the US government a 5% stake — worth roughly $42.6 billion at current valuations — as part of a broader arrangement where each leading AI developer contributes equity to a sovereign wealth fund vehicle. Sam Altman has pitched the idea to Trump, Commerce Secretary Howard Lutnick, and Treasury Secretary Scott Bessent, framing it as the best way to “share the benefits of AI” with the public. It sounds magnanimous. It isn’t.
The proposal lands in a context that makes its framing almost surreal. Last month, the Trump administration ordered Anthropic to curtail access to its newest model for foreign nationals on national security grounds, forcing a suspension. This week’s GPT-5.6 release required explicit government approval before shipping. As one Hacker News commenter put it bluntly: “The administration is not building an oversight framework. It is a toll booth.” The government already controls the gate. Now it wants a cut of the toll revenue too. This is the structural feature that the “public wealth” rhetoric obscures: when the entity deciding whether your product can ship also owns 5% of your company, the line between regulation and rent-seeking doesn’t blur — it evaporates.
The deeper problem is incentive alignment, but not in the direction Altman suggests. Bruce Schneier and Nathan Sanders, writing in the Guardian last month, made the counterargument crisply: if the sovereign wealth fund’s value depends on Nvidia growing from $5 trillion to $10 trillion, fund managers will advocate for chip sales, deregulation, and AI adoption “with the same zeal as the company’s private investors.” Norway’s sovereign wealth fund — the world’s largest — holds substantial stakes in oil companies, and that dependence has demonstrably inhibited Norwegian climate action. Public ownership doesn’t give government leverage over corporations; it gives corporations leverage over government. The 5% stake doesn’t democratize AI. It transforms the regulator into a silent partner with a fiduciary duty to maximize the regulated’s share price.
What makes this moment particularly sharp is the contrast arriving the same week from Spain. The Spanish government has instructed companies controlled by SEPI — its own sovereign wealth fund — to blacklist Palantir from new contracts across defense, critical communications, and public infrastructure. This follows France’s domestic intelligence agency terminating its Palantir contract and London Mayor Sadiq Khan blocking a £50 million Metropolitan Police deal. Spain’s model is the inverse of America’s: rather than taking equity in the company it worries about, it excludes it entirely. The Spanish government decided that Palantir’s integration of classified data, its work with the Israeli military, and its deep embedding in the Trump administration’s immigration enforcement created an unacceptable sovereignty risk. Taking 5% of Palantir would have been absurd; the answer was zero.
The US approach reveals something about how this administration understands power. Trump’s “presidential portfolio” now includes equity stakes of 5-15% in at least 20 companies, spanning semiconductors, steel, nuclear, rare earths, and quantum computing. The Cato Institute has tracked this as an improvisational sovereign wealth fund built through “deals, shakedowns, equity stakes, warrants, and revenue-sharing arrangements” — none of it authorized by Congress. The OpenAI proposal extends this pattern to the most strategically consequential sector yet. Bernie Sanders has pushed for a far more aggressive version — 50% stakes across all major AI companies — but even his framework shares the same flaw: it conflates democratic accountability with shareholder participation.
The honest version of Altman’s pitch is this: OpenAI is buying regulatory insurance. A government that owns 5% of your company is a government that needs your IPO to succeed. It’s a government whose sovereign wealth fund reports to the same treasury secretary who already approved your model releases. The “public benefit” framing is the Alaska Permanent Fund analogy — but Alaska doesn’t also regulate oil extraction permits on a per-shipment basis. The combination of pre-release model approval and equity ownership creates something unprecedented: a state that both gates and profits from a private technology’s deployment. Spain chose exclusion. The US is choosing entanglement. The latter is far more dangerous, because a government that profits from you never has an incentive to stop you.
Sources
- OpenAI ‘in early talks to give 5% stake to US government’ — The Guardian
- Bernie Sanders’ AI sovereign wealth fund plan is good. But we think this is better — The Guardian (Schneier & Sanders)
- Spain tells state-backed firms to avoid new Palantir contracts — Anadolu Agency
- Trump’s Presidential Portfolio Goes Quantum — Cato Institute
- U.S. could take stakes in more companies, Trump adviser says — NBC News
- HN discussion: OpenAI ‘in early talks to give 5% stake to US government’
- HN comment: “The administration is building a toll booth”
- HN comment: “The government owning a stake is different from the people owning a stake”