OpenAI's 5% Government Stake: What Altman Proposed

OpenAI is in early-stage talks to give the U.S. government an equity stake in the company, a proposal that would tie Washington’s balance sheet directly to the fortunes of the world’s best-known AI developer. The plan, first reported by the Financial Times on July 2 and confirmed in outline by multiple outlets since, would hand the government roughly 5% of OpenAI — a holding worth about $42.6 billion at the company’s most recent $852 billion valuation.

The discussions are described by people familiar with them as “conceptual” and preliminary. No terms are settled, and executives on both sides caution that any final structure could look very different — or never happen at all. But the idea itself is a striking one, and it arrives at a charged moment for the industry as leading labs move toward the public markets and Washington sharpens its interest in how frontier models get built and released.

What OpenAI is actually proposing

According to the reporting, chief executive Sam Altman has floated an arrangement in which the U.S. government would take a 5% stake not just in OpenAI but in each of the leading American AI developers — a group that would include Anthropic, Google, and Meta — with the shares pooled into a government-held investment vehicle.

Crucially, OpenAI would be donating the equity rather than selling it. Unlike Intel, which earlier traded shares to the government in exchange for a large slug of federal funding, OpenAI’s proposal envisions handing over stock without receiving cash in return. The pitch is framed less as a financing transaction than as a way to give the public a direct claim on the wealth these companies are generating.

Altman and his executives have repeatedly pointed to the Alaska Permanent Fund as their model. Alaska voters created that fund in 1976, amending the state constitution to require that at least 25% of the state’s oil and mineral royalties be set aside and invested rather than spent. Starting in 1982, the fund began paying every Alaska resident an annual dividend. The analogy OpenAI is drawing is direct: if oil was the resource that generated Alaska’s windfall, artificial intelligence is the resource generating this one, and a permanent fund could spread the gains across the population instead of concentrating them among shareholders.

Why now

The timing is not incidental. The proposal surfaced just days after Washington delayed the release of OpenAI’s GPT-5.6 model, part of a broader push by officials toward voluntary standards governing how new frontier models are shipped. The same week, the Commerce Department lifted export curbs it had briefly imposed on Anthropic’s Fable and Mythos models — a reminder that the government’s posture toward the leading labs has grown markedly more hands-on.

OpenAI is also preparing for a planned IPO, reportedly targeting a public debut later in 2026. A company heading into public markets while under intensifying political scrutiny has obvious incentives to get ahead of the “who benefits from AI” question. Offering the government — and by extension the public — a stake is one way to reframe a debate that has increasingly cast frontier labs as concentrating extraordinary wealth and power in a handful of private hands.

That dynamic is part of a larger picture. As venture funding hit record highs in the first half of 2026, OpenAI and Anthropic alone accounted for a striking share of all global startup capital — a concentration that has drawn attention from lawmakers across the political spectrum.

The political reaction

The response has split along unusual lines. On the left, Senator Bernie Sanders has argued the proposal doesn’t go nearly far enough. In June, the Vermont senator introduced the American AI Sovereign Wealth Fund Act, which would impose a one-time 50% tax — paid in stock — on AI companies with more than $200 million in annual revenue. Against that backdrop, a voluntary 5% donation looks modest.

On the right, some critics have dismissed the concept outright as a form of creeping government ownership of private industry. And a distinct set of governance scholars have raised a structural objection that cuts across partisan lines: a government that holds equity in the companies it regulates faces an obvious conflict. A regulator with a financial stake in a firm’s success, the argument goes, cannot credibly enforce rules impartially against that firm — particularly rules about safety, model releases, or market competition.

There is also the question of mechanics. The FT characterized the talks as early and conceptual, and reported that implementing any deal on the scale Altman describes might require an act of Congress. Pooling stakes from multiple private companies into a federal investment vehicle, deciding how the shares are governed and voted, and determining how any dividends would flow back to the public are not details that get resolved in a boardroom.

Whether the other labs would go along

The proposal only works at scale if OpenAI’s rivals participate, and there is no indication yet that they will. Anthropic, Google, and Meta have not publicly committed to ceding equity, and each has its own reasons for caution. Anthropic, which has confidentially filed to go public, is navigating its own path to the markets and may be reluctant to hand a fixed slice of a fast-appreciating asset to the government. Google and Meta are already public companies with existing shareholders who would have to be persuaded that donating billions in equity serves their interests.

That leaves a real possibility that OpenAI’s gesture — however sincere — ends up as a solo move rather than an industry standard, which would undercut the “spread the gains broadly” rationale that gives it its moral force.

What it means

Strip away the novelty and this is a story about who captures the value of the AI boom, and it lands squarely at the intersection of markets and politics.

For OpenAI, the proposal is a bet that giving up a slice of equity now buys goodwill — and regulatory breathing room — that is worth more than the shares. Heading into an IPO under a government that has already shown it will delay model releases and impose export controls, aligning its own upside with Washington’s is a way to convert a potential adversary into a stakeholder. The risk is that it sets a precedent competitors resent and regulators exploit.

For the government, a 5% stake worth tens of billions is genuinely tempting fiscal firepower, but the conflict-of-interest problem is not hand-waving. If Washington owns a piece of the companies it polices, every enforcement decision — every safety rule, every antitrust question, every export restriction — becomes entangled with the value of its own holdings. The cleaner the regulator, the harder that is to justify.

For the industry, the episode signals how quickly the relationship between frontier labs and the state is being redrawn. A year ago the debate was about how regulators like the EU would enforce rules on general-purpose AI. Now a leading American lab is voluntarily offering the government equity. That is a very different model of oversight — ownership rather than rules — and once one company puts it on the table, it becomes a template others will be pressed to answer for.

What to watch next: whether the talks move from “conceptual” to a concrete term sheet; whether any rival lab publicly backs the idea or explicitly rejects it; how OpenAI’s IPO timeline interacts with the proposal; and whether Congress treats a voluntary donation as sufficient or, following Sanders, pushes for something mandatory and far larger. For now, the number to remember is 5% — and the fact that, for the first time, one of the most valuable companies in the world is the one proposing to give it away.