OpenAI formalizes its PBC: the governance playbook every founder negotiating with a foreign investor should read

In October 2025, OpenAI closed the recapitalization that establishes its definitive corporate structure: on one side, the OpenAI Foundation (the former nonprofit organization); on the other, OpenAI Group PBC, a public benefit corporation that now holds the commercial operation. The official Our Structure page, published by the company itself, describes the design that had been under negotiation for nearly a year with the attorneys general of California and Delaware.

The fact itself isn't breaking news: more than a year has passed since the announcement, on October 28, 2025. What makes it worth reading now is the legal and financial design that came out of it, because it solves a problem that every Brazilian founder negotiating with an American or European fund also faces: how to raise heavy capital without handing over control of the company along with the check.

The problem the structure solves

Since 2019, OpenAI had already operated with a for-profit subsidiary controlled by the original nonprofit. The problem was that this old model limited its ability to raise capital: a serious institutional investor wants stock that appreciates predictably, not a stake capped by a return ceiling (the old "capped profit" model). The solution, according to OpenAI itself, was to replace the regular for-profit subsidiary with a PBC, which by legal definition is required to pursue its stated mission and consider the interests of all stakeholders, not just the shareholder.

In practice, this gives OpenAI Group a capital instrument equal to that of any ordinary company, common stock that shares proportionally in growth, but with a layer of statutory mission obligation built into the corporate form itself. It's different from relying solely on a well-intentioned board: the obligation sits in PBC law, not in the directors' goodwill.

Who's in charge and who makes money are different legal entities

The core of the design is separating two questions that usually go together in a funding round: who controls and who profits. The OpenAI Foundation doesn't manage day-to-day operations, but it holds exclusive voting and governance rights that give it the power to appoint the entire board of OpenAI Group PBC and replace directors at any time.

This separation is reinforced by physical overlap: every current Foundation director also sits on the Group board, except Dr. Zico Kolter, who chairs the Safety and Security Committee and now serves only on the Foundation board, as a non-voting observer on the commercial side. The stated idea is to avoid what OpenAI itself calls "fragmented governance," when the side that defends the mission and the side that closes commercial deals stop talking to each other.

The numbers behind the control

What makes this arrangement more than a symbolic gesture is the sheer size of the stake the Foundation retained. According to the official page, at the close of the recapitalization OpenAI Foundation held a 26% equity stake in OpenAI Group, valued at approximately $130 billion based on the company's valuation at the time.

In addition to the direct stake, the Foundation received a warrant that guarantees it additional shares should OpenAI Group's share price rise more than tenfold within 15 years. Combining both pieces, OpenAI itself states that the Foundation is positioned to be the largest long-term beneficiary of the company's success, which ties the nonprofit entity's financial incentive to the business's actual growth, rather than leaving it dependent solely on donations or a fixed budget.

The counterpoint: concentrated power has a price

The more critical reading of this engineering is straightforward: the Foundation concentrates veto power and board-appointment power without being accountable to minority shareholders the way a publicly traded company is. Whoever buys OpenAI Group stock is, in practice, buying a cash flow whose long-term control runs through an entity that doesn't answer to a commercial shareholders' meeting, only to its own board.

The argument in favor is that this is exactly the stated function: preventing short-term investor pressure from overriding the long-term mission. But this only works if the investor agrees to come in knowing it will never hold a voting majority, a concession that growth equity funds rarely make without a risk premium built into the share price or specific protective clauses.

For anyone negotiating a term sheet in Brazil, that's the point to watch: no international investor will accept an asymmetric control structure without understanding exactly where the residual power sits, and without negotiating information rights, veto rights on material matters, or exit rights (drag-along, tag-along) that compensate for the lack of majority control.

What this teaches founders raising capital in Brazil

Few Brazilian startups will replicate a foundation holding 26% of a $500 billion company, but the engineering principle is reusable on a smaller scale. A founder who needs to raise heavy capital from a foreign fund without losing control of the product can negotiate:

• Differentiated share classes, with multiple voting rights retained by the founder, separating decision-making power from the size of the economic stake.
• Contractual "mission lock" clauses, which condition certain strategic decisions (sale, change of purpose) on qualified approval, not just a simple majority.
• Upside instruments tied to valuation milestones (like the Foundation's warrant), useful for aligning whoever brought in seed capital or key talent without immediately diluting the cap table.
• Independent committees reporting directly to the controlling board, replicating the role of the Safety and Security Committee for sensitive topics like data security or regulatory compliance.

The risk, on the Brazilian founder's side, is using this kind of structure as an excuse to refuse any dilution of control. A serious foreign investor has already seen OpenAI's playbook and will ask, directly, what concrete problem the voting asymmetry solves in your company, not just whether it exists. A complex governance structure without a mission or safety rationale to justify it tends to be read as a red flag, not as sophistication.