Trump Opens Union Equity Discussion in OpenAI and xAI, Leaving Anthropic Out

Donald Trump said on Friday, June 6, in remarks to reporters at the White House, that the U.S. government might take direct equity stakes in AI companies, specifically naming OpenAI, Anthropic, and xAI. The President's statement that making the government "a partner in this revolution" would be "a beautiful thing" opened a public debate that had been conducted quietly among the Department of Commerce, the Treasury, and the three largest frontier labs in the country.

The model under discussion originates from OpenAI itself. Sam Altman submitted a proposal for a Public Wealth Fund in April 2026: the company would donate, rather than sell, equity to the federal government. This equity would fund a sovereign investment fund. It does not involve taxpayer money at the outset. In return, OpenAI would gain regulatory clarity, potential preferential treatment in public procurement, and protection against the possibility of forced nationalization that Bernie Sanders openly supports.

Sanders proposed a much more aggressive version the same weekend. He wants 50% government equity in the three companies and an additional 50% tax on their shares. This amounts to partial nationalization. The coincidence that MAGA and the progressive left have reached the same conclusion regarding the private capture of frontier technology did not go unnoticed on Capitol Hill.

The Detail That Changes Everything: Anthropic Is Not at the Table

The operational difference among the three labs is what ultimately defines where this will end up. According to Bloomberg, Anthropic is not in equity discussions with the White House. In February 2026, Defense Secretary Pete Hegseth designated Anthropic as a supply chain risk after Dario Amodei refused to relax two red lines: no mass surveillance of Americans and no lethal autonomous weapons.

The practical consequence began in March. The Pentagon initiated tests with models from OpenAI and Google to replace Claude in the Maven Smart System, a classified platform that, according to Bloomberg, was used in recent operations against Iran. The internal goal of the Department of Defense is to phase out Anthropic within six months. Sam Altman, according to NOTUS, met with Howard Lutnick at the Treasury this week to outline the structure of the Public Wealth Fund. Elon Musk's xAI, which signed a OneGov contract with the GSA in September for $0.42 per federal agency for 18 months, is more politically aligned and is likely to receive favorable treatment.

The Message for Companies Outside the U.S.

The takeaway for CIOs and innovation directors outside the United States is stark. If OpenAI becomes a company with 10% to 30% U.S. government equity, decisions regarding exports, chip restrictions, and model blockages will cease to be commercial and will instead depend on political agendas in Washington. For Capgemini, which has standardized part of its operations on GPT-5.5 for regulated European clients, this presents a new problem of sovereign jurisdiction sourcing.

In the United Kingdom, the Treasury began a consultation in May on how the National Wealth Fund could, in theory, take stakes in British labs. The movement had been stalled due to a lack of a replicable model. If OpenAI moves forward, London gains a political reference to justify the step within Parliament. For suppliers like Stability AI and ElevenLabs, based in London, this discussion changes the captable equation of any upcoming funding round.

In the European Union, the transatlantic divergence opens a window for Mistral in Paris to gain political attractiveness as an alternative outside the perimeter of the U.S. Public Wealth Fund. The text of the AI Act does not regulate supplier ownership, but European governments have already signaled in recent speeches that the sovereign origin of the model should weigh in strategic public procurement. For any bank in the eurozone under DORA, this becomes a criterion to document in supplier evaluation panels.

In Brazil, BNDES does not have a mandate for equity stakes in foreign AI labs, and the local discussion on computational sovereignty still revolves around the Brazilian Artificial Intelligence Plan and the use of Sovrana, the national infrastructure announced in 2024. The question that remains for any technology procurement committee in 2026 is simpler: if a contract with OpenAI comes with a partner called "the United States government," what changes in the procurement process of a private company globally?

What Is Not Yet Confirmed

OpenAI had not publicly commented on the timeline of the proposal by the time this article was published. Anthropic and xAI also did not comment. The White House did not confirm a timeline. The feasible schedule, according to two sources close to the Treasury quoted by NOTUS, depends on the approval of the defense budget for 2027 and is unlikely to materialize before the fourth quarter.