On July 2, the Financial Times reported that OpenAI has proposed giving the United States government a 5% equity stake in the company. At OpenAI's $852 billion valuation, established in a $122 billion funding round three months earlier, the stake would be worth approximately $42.6 billion.
CEO Sam Altman discussed the plan directly with President Trump, Commerce Secretary Howard Lutnick, and Treasury Secretary Scott Bessent. He framed the equity as the best way to share the upside of AI with the public. The proposal extends beyond OpenAI: Altman suggested the mechanism apply to other major American AI companies, including Anthropic, Google, and Meta.
Six days earlier, Washington had grounded his product. On June 2, Trump signed an executive order asking AI companies to voluntarily submit frontier models for government review up to 30 days before release. On June 26, the administration exercised that provision for the first time, restricting the rollout of GPT-5.6 to a small group of "trusted partners." Sol, Terra, and Luna, the three models in the 5.6 series, remain unavailable to the general public.
The stated template is the Alaska Permanent Fund. Established in 1976, the fund invests Alaska's surplus oil revenues into a diversified portfolio and pays annual dividends to every state resident. In 2025, each eligible Alaskan received $1,000. The fund currently manages approximately $83 billion. It has outlasted every governor, every political cycle, and every attempt to redirect its revenues. Nobody votes to shut down the pipeline when the pipeline sends them a check.
Altman is proposing to do for artificial intelligence what Jay Hammond did for petroleum. Give the public a financial interest in the industry's growth, and the politics of restriction change. Regulation becomes a write-down on a national asset. Oversight becomes a drag on the Treasury's portfolio. The mechanism aligns the government's financial incentives with the industry's, and the regulatory apparatus bends toward accommodation.
The timing says more than the framing. Altman didn't propose the fund when GPT-4 launched. He didn't propose it after raising $122 billion in March. He proposed it six days after the government restricted his product. Five percent dilution is a small price for regulatory alignment.
Senator Bernie Sanders has already staked the other end of the range. His American AI Sovereign Wealth Fund Act, introduced in June, would impose a one-time 50% stock tax, paid in equity rather than cash, on any AI company earning more than $200 million annually. At current valuations, the fund would be worth approximately $7 trillion. It would distribute 5% of its market value each year to the Treasury and grant an independent commission voting shares with authority to block corporate decisions deemed harmful to the public.
The distance between 5% and 50% is the negotiation. Sanders' version includes voting rights, an independent oversight body, and intervention authority. Altman's includes none of those. Five percent is a dividend. Fifty percent is a board seat. Altman's number reads moderate only because Sanders set the anchor.
The most striking element of Altman's proposal is its scope. He isn't offering OpenAI alone. He's proposing that every frontier AI lab contribute 5%. If the government accepts and Google declines, Google looks extractive. If all companies comply, competitive dynamics stay unchanged because everyone dilutes equally, and the regulatory environment shifts in everyone's favor. An industry that cannot legally coordinate on pricing can coordinate through the architecture of a sovereign wealth fund.
The $122 billion round included $50 billion from Amazon, $30 billion from Nvidia, and $30 billion from SoftBank. OpenAI generates approximately $2 billion per month in revenue, up from $13.1 billion for all of 2025. The proposed 5% costs nothing in cash. It costs investors approximately 5% of the equity they purchased three months ago. In return, the government holds a position that appreciates when it doesn't intervene.
The FT characterized the discussions as "conceptual and early-stage." Implementation would require an act of Congress. Both facts will be cited as evidence the proposal isn't serious. They miss the function. Once the conversation is about how much of AI the public should own, the question of whether to regulate AI has already been answered. You don't regulate your portfolio. You grow it.