Washington Was Offered 5% of OpenAI. When the Regulator Becomes a Shareholder.
In a nutshell
everything on the web starts with the domain
A month ago, gafam.ai reported that Donald Trump, Bernie Sanders and Sam Altman had converged, from three different directions, on a single strange idea: that the US government should own a piece of the companies building artificial intelligence. At the time, it was a principle without a price. This month it acquired both a number and a mechanism.
gafam.ai reads what that means — and why it matters more in Europe than the American coverage acknowledges.
What OpenAI Proposed — Stated Carefully
The reporting must be framed precisely, because this is a proposal, not a done deal. According to the Financial Times, which broke the story on July 2 citing two people familiar with the talks, OpenAI has proposed handing the US government a 5% equity stake in the company. Based on OpenAI's $852 billion valuation from its March 2026 funding round, that stake would be worth roughly $42.6 billion.
The structure is the interesting part. Altman's pitch is not about OpenAI alone: the proposal envisions each of the largest US AI developers — Anthropic, Google and Meta named among them — ceding a matching 5% into a shared government investment vehicle, explicitly modeled on the Alaska Permanent Fund, the 1976 sovereign fund that turns the state's oil revenue into annual dividends for residents. Altman's framing is that ordinary Americans should share in the AI upside rather than watching the gains pool at the top.
Now the essential caveats, because they are load-bearing. Reuters, while corroborating the FT report, explicitly stated it could not independently verify it, and neither OpenAI nor the White House responded — leaving the proposal on single-source footing at publication. It remains conceptual, not agreed. It would require an act of Congress. And it depends on voluntary participation from rivals: none of Anthropic, Google or Meta has signaled agreement, and a source said the Trump administration and Anthropic have not even discussed the idea. This is an opening bid in what would be a multi-year negotiation, not a settled arrangement.
Why This Is Not Coming From Nowhere
The proposal is easy to dismiss as a thought experiment. It should not be, because it fits an established and accelerating pattern of US state ownership in technology.
This is the same administration that took a 10% stake in Intel, and that negotiated revenue-sharing arrangements on Nvidia's and AMD's AI chip sales to China — reportedly 15%, later rising to 25% on Nvidia's H200. It has taken positions in IBM, quantum and critical-minerals companies. OpenAI itself has been developing this idea since an April 2026 white paper titled "Industrial Policy for the Intelligence Age," which proposed a "Public Wealth Fund" giving citizens an automatic stake in AI companies. Altman first pitched the concept to the Trump administration in early 2025.
In other words: government equity in frontier AI is arriving piecemeal — Intel, chip-revenue shares, now a proposed OpenAI stake — rather than through a single deliberate policy. That pattern is worth tracking independent of whether this specific 5% figure survives, because it signals a broader shift toward American state capitalism in the AI sector.
The Conflict at the Heart of It
Here is the structural problem, and it is not subtle. The US government is currently OpenAI's regulator. It vets OpenAI's models before release — the GPT-5.6 rollout was staggered for exactly this reason. It controls export access, as the Fable 5 saga showed. If it now also becomes OpenAI's shareholder, those two roles collide.
Governance scholars have named the concern directly: a regulator that holds an equity stake in the company it regulates cannot enforce rules impartially against that company. Would the government's willingness to sanction OpenAI — for a safety failure, an antitrust violation, a consumer harm — soften if the Treasury owned 5% of the upside? The proposal that is framed as sharing AI's benefits with the public is also, viewed from another angle, the most complete regulatory capture arrangement in the history of technology: not the regulated capturing the regulator through lobbying, but the regulated making the regulator a financial partner in its success.
This is the point gafam.ai made a month ago, when we wrote that a government invested in OpenAI's success has a structural disincentive to regulate it in value-reducing ways. The proposal now on the table is that dynamic, formalized and priced.
The European Perspective
The OpenAI stake proposal is, for Europe, the clearest sign yet that the United States and the European Union are building fundamentally incompatible models of AI governance — and that the gap is widening at the structural level, not merely the regulatory one. Europe governs AI through law: the EU AI Act, the DMA, enforcement actions, fines.
It is an arms-length model in which the state stands apart from the companies and constrains them from outside. The American trajectory, visible in the Intel stake, the chip-revenue shares and now this proposal, is ownership: the state as shareholder, its interests financially aligned with the companies' success. These are not two flavors of the same approach. They are opposites. And the consequence for Europe is direct and uncomfortable. If the US Treasury holds equity in OpenAI, then every European enforcement action against OpenAI — every EU AI Act investigation, every DMA proceeding, every fine — becomes a threat to a US government financial asset, and therefore a transatlantic diplomatic conflict rather than a routine regulatory matter.
European regulators would find that enforcing European law against an American AI company means picking a fight with the American state as co-owner. This is the mechanism by which the OpenAI proposal, even if it never fully closes, reshapes Europe's position: it raises the diplomatic cost of European AI enforcement at exactly the moment the EU AI Act's powers activate on August 2. There is also a quieter sting.
The "public wealth fund" idea — citizens sharing directly in AI's upside — is a genuinely progressive concept that Europe, with its stronger social-democratic tradition, might have been expected to pioneer. Instead it is being operationalized, however imperfectly, by a Trump administration, while Europe has no sovereign AI stake, no public wealth mechanism, and no frontier lab to attach one to. Europe may end up regulating an industry it neither owns nor builds, while others capture both its value and its governance. gafam.ai will be watching.
We are not first. We are right.
SOURCES
— Financial Times / CNBC: OpenAI proposes U.S. government own 5% stake to address political blowback
— Forbes: OpenAI Reportedly Pitches Granting U.S. Government 5% Stake
— Let's Data Science: OpenAI Offers 5% Stake to U.S. Government — the regulator-shareholder conflict
— Intellectia: OpenAI Proposes 5% US Government Stake: Public Wealth Fund Explained
— AI Weekly: OpenAI pitches 5% US government stake worth about $42.6B (verification caveats)