OpenAI proposes to cede 5% of its shares to the US government: the $42.6 billion redefining AI policy
🔎 5% of OpenAI to the US government: a historic deal or a diversion maneuver?
On July 2, 2026, Sam Altman offered the US government something no one would have imagined two years ago: ceding 5% of OpenAI's capital to the federal government. At the current valuation of $852 billion, this represents $42.6 billion in shares transferred to an American sovereign fund modeled on the Alaska Permanent Fund.
The concept is appealing on paper. Every major US AI company would contribute 5% of its shares to a public vehicle. The dividends would be distributed directly to American citizens. All of this, without any new taxes.
But the timing is just as revealing as the proposal itself. This announcement comes as Anthropic reaches a $47 billion revenue run-rate and surpasses OpenAI, as OpenAI is weeks away from a confidential IPO filing, and as it has just emerged from a lawsuit with Apple over the massive poaching of 400 employees.
Altman did not make this proposal in a political vacuum. He pitched it in person to Donald Trump, Secretary of Commerce Howard Lutnick, and Secretary of the Treasury Scott Bessent. It is a calculated gesture, not an act of generosity.
The Essentials
- OpenAI is proposing to cede 5% of its equity ($42.6 billion) to the US federal government via a sovereign fund inspired by the Alaska Permanent Fund.
- Sam Altman presented the plan directly to Trump, Lutnick, and Bessent in early July 2026, with additional details communicated on July 14.
- 69% of American workers surveyed support the idea of forcing AI firms to cede 50% of their shares to a public fund — well beyond the 5% proposed by Altman.
- The proposal would require an act of Congress to be implemented, making it politically uncertain despite genuine popular support.
- The deal comes at a critical time for OpenAI: pre-IPO, the Apple lawsuit, and a valuation race where Anthropic is aiming for $900 billion.
Recommended Tools
| Hostinger | Web hosting to keep up with AI news | Starting at 2,99 € (July 2026, check on hostinger.com) | Creators and tech media |
|---|---|---|---|
The numbers behind the deal — 42.6 billion on an $852 billion table
The amount is colossal, but it needs to be put into context. OpenAI is valued at $852 billion after its latest funding round. Five percent of this amount equates to exactly $42.6 billion in shares.
This is not a check. It is a transfer of title. The government does not pay anything. It receives shares and would then collect dividends on OpenAI's future profits.
The reference model is clear: the Alaska Permanent Fund, created in 1976, which invests 25% of the state's oil revenues and pays an annual dividend to every Alaskan resident. In 2025, this dividend amounted to approximately $1,700 per person.
Transposed to the federal level with AI revenues, the calculation becomes dizzying. If OpenAI generates $10 billion in annual net profit, 5% represents $500 million in dividends per year. Distributed among 330 million Americans, this comes out to about $1.50 per person per year. Not enough to change your life.
The real potential emerges if the fund applies to all major AI firms. With Google, Microsoft, Anthropic, and xAI included, the annual amount could reach several billion. But there, we leave Altman's proposal and enter the realm of political speculation.
Sam Altman's pitch to the White House
Altman didn't send a memo. He physically went to present his idea to the three key decision-makers in the Trump administration.
First, Trump himself. The president has always been drawn to spectacular real estate and financial deals. A sovereign fund that brings in money for citizens without raising taxes is a politically perfect narrative for him.
Next, Howard Lutnick, Secretary of Commerce, who directly oversees technology policy and AI chip exports. Lutnick is a financier by training — he immediately understands the logic of a sovereign fund.
Finally, Scott Bessent, Secretary of the Treasury, who would structurally manage this fund. He would have the concrete task of actually bringing it to life.
According to The Guardian, Altman presented the proposal as an "unprecedented public-private partnership" rather than as regulation. The choice of words is not insignificant. "Partnership" sounds better than "forced divestment" or "partial nationalization".
BuildFastWithAI notes that on July 14, additional details were shared: the fund would be managed by an independent board of directors, with representatives from the Treasury, Commerce, and the private sector. Dividends would be paid out annually via a mechanism similar to stimulus checks.
The Alaska Precedent — why the model holds up
The Alaska Permanent Fund is the only American precedent of a sovereign wealth fund redistributing natural resource revenues to citizens. Created by a constitutional amendment in 1976, it is funded by at least 25% of the state's oil royalties.
In nearly 50 years, the fund has reached approximately $80 billion in assets. It pays an annual dividend to every Alaskan resident, which has ranged between $300 and $3,200 depending on the year.
The parallel with AI is explicitly drawn by Altman. The argument: American data, digital infrastructure, and government-funded research (DARPA, NSF) are the "natural resources" of the AI era. Companies exploit these resources. Citizens should receive a dividend from them.
This is intellectually coherent. But there is a fundamental difference. Alaska's oil legally belongs to the state. The data and research that feed OpenAI's models do not belong to the federal government. The legal foundation is therefore much more fragile.
The Alaskan model also works because it is anchored in a state constitution, not in a federal law. A federal-level AI sovereign fund would require an act of Congress — a massive obstacle in the current political climate.
Why now — the convergence of pressures on OpenAI
The proposal didn't come out of nowhere. Three factors converge to explain the exact timing of this announcement.
The imminent IPO
OpenAI is weeks away from a confidential IPO filing. An $852 billion stock market listing would be the largest in tech history. But an IPO of this magnitude attracts the attention of regulators, elected officials, and the media.
By proposing a sovereign fund before the IPO, Altman is sending a message: "We are not a company that takes and gives nothing back. We are ready to share the wealth with the American people." It's a trillion-scale public relations operation.
The Apple lawsuit and the 400 poached employees
Apple has sued OpenAI for the systematic poaching of over 400 employees. This lawsuit threatens to reveal internal practices, disputed non-compete clauses, and potentially secret agreements with other companies.
In this context, appearing as a responsible player who wants to give to the American people is a narrative strategy. The Apple lawsuit portrays OpenAI as a predator. The sovereign fund repositions it as a benefactor.
Popular pressure is real
A recent poll shows that 69% of American workers support the idea of forcing AI firms to cede 50% of their shares to a public fund. That is a significant figure. And it far exceeds the 5% proposed by Altman.
By voluntarily offering 5%, OpenAI is attempting to defuse a demand that is ten times more ambitious. It's the classic strategy of "better to give 5% now than risk losing 50% later."
Geopolitical implications — the AI war against China
The sovereign wealth fund is not just a domestic issue. It is part of the geopolitical competition with China for artificial intelligence supremacy.
Altman's argument, as reported by AIN, is that the sovereign wealth fund would create an "alignment of interests" between the US government and its AI champions. When OpenAI wins, America wins. This is an argument that particularly resonates within the Trump administration.
China, for its part, already has a model where the state and tech companies are closely linked. Chinese AI giants — Baidu, Alibaba, ByteDance — operate in an ecosystem where the Communist Party has permanent oversight. The American sovereign wealth fund would create a democratic equivalent: the state as a shareholder without authoritarian control.
It is also a signal sent to US allies. By showing that the benefits of AI will be shared with citizens, Washington can counter the Chinese narrative that American tech capitalism only benefits an elite.
But there is a risk. If the US government becomes a shareholder in OpenAI, European and Asian countries could see this as proof that American AI companies are extensions of the state. This would strengthen the arguments of those who want to develop their own sovereign models and reduce their dependence on American infrastructure.
Privatization of gains, socialization of risks — the crux of the problem
This is the most serious criticism against Altman's proposal. And it deserves an in-depth discussion.
When OpenAI loses money — and it has lost billions for years — those losses are borne by private investors. When OpenAI makes money, the proposal is to give a share to the government.
This is the reverse of the "socialize losses, privatize gains" principle that caused outrage during the 2008 financial crisis. Here, it is more subtle: the losses remain private, but the gains are partially socialized. The risk to the taxpayer is low (they pay nothing in capital), but the principle is debatable.
The government would receive 5% of a company whose initial losses it did not bear. The Series A, B, and C investors, those who took a risk when OpenAI was worth 10 billion, now see their stake diluted in favor of a shareholder who risked nothing.
However, the counterargument also exists. OpenAI has benefited from years of fundamental research funded by the US government. Transformers, the GPT architecture, advances in deep reinforcement learning — all of this rests on decades of research funded by the NSF, DARPA, and other federal agencies.
The reinforcement learning model that underpins the complex reasoning of tools like o1-preview, documented in the OpenAI o1 System Card, relies in part on publicly funded academic advances. Even the applications of these models, whether it be competitive programming or high-level educational reasoning, rely on this public foundation.
In this reading, the sovereign wealth fund is not a gift. It is a return on collective investment.
The Congressional Hurdle — why the deal is far from done
Despite popular support and interest from the White House, the proposal is running into a constitutional and political wall: the US Congress.
A federal sovereign fund of this magnitude cannot be created by executive order. It would require a law passed by the House of Representatives and the Senate. And the current Congress is far from unanimous on the issue.
Fiscally conservative Republicans will see this as a form of nationalization, even if limited to 5%. The federal government is not meant to be a shareholder in private companies — this is a principle deeply rooted in American conservative economic thought.
Progressive Democrats, paradoxically, might also oppose it. Not because they are against redistribution, but because they will consider 5% insufficient. If 69% of workers want 50%, accepting 5% could be perceived as a betrayal of popular expectations.
Libertarians, finally, will see this as an open door to state interventionism in technology. Today 5%, tomorrow 10%, and then what?
SquaredTech notes that senators from both parties have already expressed reservations, without outright rejecting the idea. The public debate is underway, but the legislative path is long.
One must also consider the issue of other companies. If the fund applies only to OpenAI, it is discriminatory. If it applies to all AI firms, Google, Microsoft, and Meta will not stand idly by. Their lobbyists are among the most powerful in Washington.
The race for valuations — why OpenAI needs this narrative
The competitive context is essential to understanding this proposal. OpenAI is no longer the undisputed leader of AI.
Anthropic and OpenAI each launch their enterprise JV: $10 billion to deploy AI in SMBs and large corporations, which shows that competition is now direct in the enterprise space. And with models like Claude Opus 4.7 (Adaptive) reaching 94.3 in the agentic ranking and 90 overall, Anthropic is a serious competitor.
OpenAI's $852 billion valuation relies partly on expectations of future growth. If the market begins to doubt OpenAI's ability to maintain its lead, the valuation could drop even before the IPO.
The sovereign fund serves as a narrative of stability. It tells institutional investors: "Even the US government is on our side." It is a signal of confidence that can support the valuation in the weeks leading up to the IPO.
What this means for developers and businesses
Beyond high-level politics, this proposal has concrete implications for the tech ecosystem.
If the sovereign fund comes to fruition, it will set a precedent for other sectors. Biotech, clean energy, or space tech companies could face similar demands. The principle of a "citizen dividend on technological externalities" could become a lasting policy framework.
For developers working with OpenAI's models — whether it's GPT-5.5 (agentic score of 98.2), GPT-5.4 Pro or o1-preview — the change would likely be invisible in the short term. A 5% state shareholder would not interfere with product development.
In the medium term, however, the dynamic could change. A government shareholder could demand audits on model bias, their environmental impact, or their use in critical systems. Nothing dramatic, but an additional layer of accountability.
For businesses hosting AI solutions, the structure changes very little. A host like Hostinger would not be directly affected by a sovereign fund. But companies integrating models via API might see costs increase slightly if OpenAI adjusts its prices to compensate for the dilution.
❌ Common mistakes
Mistake 1: Confusing the sale of shares with nationalization
This is not a nationalization. The government would receive 5% of the shares, not control. It would have no significant voting rights and no ability to run the company. It is a passive investment, not a power grab. Confusing the two weakens the debate.
Mistake 2: Believing the citizen dividend would be substantial
With 5% of a single company, the dividend per American would be in the range of $1 to $2 per year. Even if extended to several companies, it would not replace an income. The issue is symbolic and structural, not immediately material.
Mistake 3: Thinking the deal is already done
Altman made a proposal. The government has not signed. Congress has not voted. Nothing is legally binding. The probability that an AI sovereign wealth fund will see the light of day before 2028 is low, even if the political debate it generates is real and lasting.
Mistake 4: Ignoring the role of other companies
If the fund only applies to OpenAI, it is unfair and probably unconstitutional. If it applies to the entire sector, Google, Microsoft, Meta, and xAI will have their say. Their lobbyists can block the project in Congress. OpenAI alone cannot push this deal through.
❓ Frequently Asked Questions
How much would each American actually receive?
With 5% of OpenAI alone, the annual dividend would be about $1.50 per person. Extended to the entire AI sector, it could reach $10 to $30 per year. Symbolic, not transformative.
Would the government have voting rights at OpenAI?
Normally no. The Alaska Permanent Fund model grants the fund shares without specific voting rights. But the legal details remain to be defined and could be the subject of intense negotiations in Congress.
Why 5% and not 10% or 50%?
It's a calculated compromise. Enough to be media-significant. Little enough not to scare off existing investors. And well below the 50% supported by 69% of those polled to appear as a "reasonable" offer in the public debate.
Does China have an equivalent model?
Not exactly. China exercises direct control through the Communist Party, not through a redistributive sovereign wealth fund. But the idea of aligning the interests of the state and AI companies is common to both approaches, with radically different methods.
Would this affect the performance of models like GPT-5.5 or o1-preview?
Not directly. A passive 5% shareholder does not intervene in research or engineering. But additional political pressure could slow down certain deployments or impose transparency constraints that would indirectly affect development cycles.
✅ Conclusion
OpenAI's proposal to cede 5% of its equity to the US government is Sam Altman's boldest political move. At $42.6 billion, it is also the most expensive. But above all, it is a strategic maneuver — defusing regulatory pressure ahead of a historic IPO, countering the rise of Anthropic, and turning a political risk into a narrative asset. Will the deal pass Congress? Probably not in its current form. But the debate it opens up about who should profit from AI is here to stay.