The White House discusses AI profit sharing with major companies

Over the past two years, the AI market has cared about only one question: who can make the most money?

NVIDIA orders, cloud providers' capital expenditures, data center construction, model company valuations, and corporate adoption rates form the core narrative of this AI-driven market cycle. Investors are buying growth, betting on profit pools, and debating how much economic value AI can convert into corporate revenue.

But now, another issue is beginning to emerge:

If AI truly creates unprecedented wealth, should that money belong solely to the company, its employees, and shareholders?

This is truly what matters most about the OpenAI Public Wealth Fund discussion.

It is not an already implemented regulatory policy, nor is the U.S. government about to "seize equity in AI companies." More accurately, it is the first time the AI industry has brought the question of "how future excess returns should be allocated" onto the public policy table.

The counterintuitive aspect of this is that the market didn’t begin discussing allocation because people doubted AI’s ability to generate profits. On the contrary, it was precisely because more and more people believed AI would generate substantial excess profits that political systems began asking: Should these gains be monopolized by only a few companies and investors?

First, let’s clarify the boundaries of the facts.

According to NOTUS on June 4, senior White House officials have held preliminary discussions with leading AI companies about voluntarily ceding a portion of equity. This approach is similar to the Alaska Permanent Fund, in which the government or a public trust holds a portion of assets and distributes part of the returns to residents.

In its April white paper, OpenAI also proposed the establishment of a public wealth fund. Large model companies could contribute capital, equity, or other forms of investment, enabling ordinary households that do not directly hold technology stocks, venture capital assets, or private equity to share in the growth benefits of AI.

Sanders' version is more radical. He advocates for large AI companies to cede a higher proportion of equity to the public and grant the public some degree of governance rights. The "50% stock tax" and board seats mentioned in the material represent the most radical political proposals in this round of discussion.

But these three things cannot be viewed together.

Discussions at the White House remain preliminary reports in the media, with no formal proportions, legal structures, or timelines. The OpenAI white paper is a corporate policy proposal, not a government document. Sanders' proposal is impactful, but there is still a long way to go before it becomes actual policy.

Therefore, the most reasonable current assessment is not that "AI companies will be nationalized," but rather that a new variable—previously absent—is now appearing on AI valuation charts:

Will the most profitable AI company in the future need to give up a portion of its economic rights to gain acceptance from society and regulators?

This has limited short-term impact on the secondary market. Public market AI-related assets such as NVDA, MSFT, AMZN, GOOGL, and META are still primarily driven by demand for computing power, cloud capital expenditures, order expectations, and profit realization.

But for unlisted model companies, the impact is more direct.

If companies like OpenAI, Anthropic, and xAI go public in the future, investors will not only ask how much money they can make, but also: How much of that money must be surrendered to public funds, governments, or other public mechanisms?

This is not a realized valuation hit, but a new policy discount.

OpenAI's proactive proposal of a public wealth fund is essentially buying "social license" for future expansion.

Social license is not a formal permit but rather the degree of tolerance from the public, regulators, and the political system toward a company’s ongoing expansion. The more successful an AI company becomes, the more acute this issue becomes.

The stronger the model's capabilities, the more discussions arise about replacing human labor. The higher the valuation, the more likely ordinary people are to perceive AI as a wealth machine exclusively benefiting a few companies, employees, and shareholders.

OpenAI is not facing ordinary tech company issues, but a narrative pressure on the scale of the Industrial Revolution:

If AI truly boosts productivity, who shares in these gains?

OpenAI’s white paper emphasizes the need for the United States to maintain its leadership in AI, while acknowledging that automation may reshape a large number of jobs. The public wealth fund is one of the buffer solutions it proposes.

In market terms, OpenAI may seek to reduce more unpredictable political risks by leveraging a portion of controlled future economic rights.

If we completely ignore the narrative that "AI is taking jobs and profits are going to a few," we may face higher taxes, stricter regulation, antitrust pressures, or even be forced to disclose more complex policy risks during the listing process.

Proactively designing a gentle sharing mechanism may transform the risk from "unknown political shocks" into "estimable long-term costs."

It’s somewhat like a resource company entering a region by first designing local employment, infrastructure, and revenue-sharing plans. The difference is that AI companies are not dealing with residents around a mine, but with the entire labor market and electorate.

It is not about a one-time compensation, but about how future excess profits will be accepted by society.

The phrase "transfer of equity" may sound intimidating, but different pathways have vastly different impacts on valuation.

The first option is for a company to voluntarily contribute a small percentage of economic rights, potentially without voting rights, to a public wealth fund.

If the ratio is limited and the rights are clear, it resembles a long-term policy cost. Assuming an AI company is valued at $1 trillion in the future and allocates 5% economic equity to a public fund, this would dilute existing shareholders, but the market could discount it as a well-defined adjustment.

The second is the government obtaining economic benefits through industrial policy.

For example, certain subsidies, loans, or industry support may come with warrants, which entitle the holder to a portion of equity returns under agreed-upon terms in the future. It is important to distinguish: warrants are not equivalent to direct ownership of common shares, and economic rights without voting rights do not equate to board seats.

The former is more like a fiscal sharing arrangement, while the latter enters corporate governance.

The third is a Sanders-style mandatory high public ownership ratio.

If large AI companies are required to surrender a high percentage of equity and allow public or government representatives to join their boards, the impact will no longer be about profit sharing, but rather about control, governance conflicts, and incentives for innovation.

When the government acts as both regulator and shareholder, it creates new conflicts of interest: is it protecting consumers and competition, or safeguarding the value of the companies it owns?

This is why, although the aggressive scenario has strong spread potential, it cannot currently be used as a high-probability pricing benchmark.

A more realistic scenario involves repeated discussions of small-scale, voluntary, economically incentivized solutions. While it may not be implemented immediately, it will remain an unavoidable issue in AI companies’ fundraising, IPOs, and policy communications.

For OpenAI, what’s truly sensitive isn’t “whether to share,” but whether the sharing mechanism will affect the governance structure.

Microsoft, venture capital firms, employee equity holders, and strategic investors will all be concerned: Does the public fund receive economic rights or voting rights? What is the proportion? Does it affect exit valuation? Will it alter the pricing logic for a future IPO?

Enterprise clients also ask: If the government becomes an economic beneficiary in some capacity, will procurement, data governance, and regulatory neutrality become more complex?

Therefore, the market implication is not that AI companies' profits are immediately being taken away, but rather that the AI profit pool is being discussed within a public allocation framework for the first time.

This line is still in its early stages.

The evidence chain is sufficient to indicate that the publicization of AI earnings is entering a phase of public policy experimentation; however, it is not sufficient to indicate that industry rules for AI have changed.

The next four most important points to observe are:

First, see whether other companies beyond OpenAI follow suit:

If Anthropic, xAI, or other leading model companies begin to support similar mechanisms, this could shift from OpenAI’s unilateral corporate strategy to an industry-wide negotiation framework. Conversely, if more companies publicly avoid or oppose it, the market will be more likely to view it as a unique approach by OpenAI.

Second, observe whether the White House and executive branches formalize it:

If departments such as the Ministry of Finance, the Ministry of Commerce, and the National Economic Commission begin proposing fund structures, tax arrangements, or warrant schemes, policy exploration enters a pricing phase. If it remains at the level of meetings and media leaks, the impact is primarily emotional risk.

Third, review the financing documents and future prospectus:

OpenAI, Anthropic—valuation discounts will move from discussion to transaction only if future fundraising materials or IPO filings add disclosures such as "public wealth fund, revenue sharing, government economic interests, special governance arrangements."

Fourth, check whether the market price has begun to reflect this:

Only when there is increased trading volume, rising volatility, or underperformance relative to the broader market in AI-themed ETFs, semiconductor ETFs, leading cloud companies, or related options—synchronized with policy news—will it indicate that capital is treating this factor as a trading theme. Currently, there is no such evidence.

Therefore, there is no need to interpret this as a valuation collapse in the AI industry.

A more accurate statement is:

The AI market used to price only for growth, but now it is beginning to price for allocation.

If the final solution is merely a small, non-voting, clearly disclosed economic interest, it resembles an insurance premium paid by an AI company for long-term expansion. The cost exists, but it is estimable, tradable, and acceptable.

However, if voluntary sharing is pushed by political pressure into mandatory ownership, and even into board membership and governance arrangements, the valuation logic changes significantly.

At that time, the market was offering a discount—not just on profits, but on control of the company and long-term growth flexibility.