Sam AltmanEverything you need to know about Sam Altman.

ref/sam-altman structure --capped-profit

The Capped-Profit Bet

A nonprofit that needed billions invented a new corporate form — and spent the next six years arguing about what it really was.

The problem

By 2019, OpenAI’s leadership — Altman newly CEO — had concluded that the nonprofit donation model could not fund frontier AI. Training runs were heading toward tens and then hundreds of millions of dollars; the reported $1 billion pledge from 2015 had not fully materialized. The lab needed to raise venture-scale capital without, in Altman’s telling, becoming a normal profit-maximizing company.

The invention

In March 2019, OpenAI announced OpenAI LP: a for-profit subsidiary of the nonprofit, in which investors’ returns were capped at 100x their investment, with any value beyond the cap flowing back to the nonprofit. Employees and investors could hold profit interests; the nonprofit board retained ultimate control, including the power to fire the CEO — the mechanism used in November 2023.

Altman’s pitch: the cap let OpenAI pay market-competitive returns (attracting talent and capital) while ensuring that if AGI created astronomical value, the excess would serve the mission, not shareholders. He also said he personally held no equity in the for-profit — a claim he has repeated since.

The money it unlocked

The criticisms

1. The cap is so high it barely binds. At 100x, critics argue, the cap constrains almost nothing in practice — investors behave like normal venture investors because, for all realistic outcomes, they are normal venture investors.

2. Control proved fragile. The November 2023 crisis showed the nonprofit board could fire the CEO — but also that doing so nearly destroyed the company, and the board was then replaced with friendlier directors. Critics say the episode revealed the governance as theater; defenders say it proved the mechanism worked.

3. The for-profit tail wags the nonprofit dog. Elon Musk’s 2024 lawsuit alleges OpenAI abandoned its nonprofit mission — OpenAI calls the claims meritless; the case is ongoing. Others argue more quietly that commercial pressures (shipping cadence, enterprise deals) now drive decisions the charter was meant to insulate.

4. Altman’s no-equity claim cuts both ways. Supporters read it as proof of mission-alignment; skeptics note it also insulates him from shareholder accountability while he exercises founder-like control.

The defense

Altman and OpenAI argue the structure did exactly what was promised: it raised the capital that built GPT-3, ChatGPT, and GPT-4 — systems a pure nonprofit could never have funded — while keeping a mission-bound board with real firing power. The 2023 crisis, in this telling, is evidence the board’s power was real, not decorative.

Why it matters

The capped-profit structure is Altman’s signature institutional invention — an attempt to square the circle of mission and money. Whether it is a genuine governance innovation or a clever wrapper for a normal startup is one of the central unresolved questions of the AI era.

Source notes