Overview
OpenAI is the company that turned large language models from a research topic into a consumer product. Its chatbot, ChatGPT, launched in November 2022 and had about 1.2 billion weekly users by September 2026, according to the company's DevDay 2026 recap. Behind it sit the GPT family of models, which developers and businesses rent through OpenAI's API, along with image, video, voice and coding products.
It is also one of the oddest corporate structures in technology. OpenAI began as a charity, added a profit-making arm to raise money, and in October 2025 converted that arm into a public-benefit corporation that a nonprofit still controls. Most arguments about the company come back to that structure and to its chief executive, Sam Altman.
History
OpenAI was founded in December 2015 as a nonprofit research lab with a stated mission to ensure that artificial general intelligence benefits all of humanity. Its founding group included Altman, Elon Musk, Greg Brockman, Ilya Sutskever, Wojciech Zaremba and John Schulman. Musk left the board in 2018.
Frontier AI research turned out to need far more computing power than donations could pay for. In 2019 OpenAI created a "capped-profit" company beneath the nonprofit, Altman became chief executive, and Microsoft invested $1 billion, the start of a partnership that later grew to many billions more. GPT-3 followed in 2020. Then ChatGPT, released as a research preview at the end of November 2022, reached a mass audience within weeks and set off the commercial AI race. Several researchers had already left over how fast and how commercially the lab was moving; one group, led by Dario Amodei, founded Anthropic in 2021, a story told in Anthropic: The Safety Company That Had to Win.
In November 2023 the nonprofit board removed Altman, saying he had not been consistently candid with it. Within five days, after more than 700 employees threatened to leave, he was back and most of the board had been replaced. HustleMemo's profile, Sam Altman: The Operator of Trust, examines what the episode revealed about governance.
The company then spent much of 2025 negotiating with the attorneys general of California and Delaware over its structure. The result, completed on 28 October 2025, is described below. In March 2026 OpenAI closed a funding round of $122 billion in commitments at an $852 billion post-money valuation, the largest private raise on record at the time.
How the business works
OpenAI earns money in two main ways. The first is subscriptions: individuals pay monthly for ChatGPT Plus or Pro, and companies pay per seat for business and enterprise versions. The second is usage-based pricing through its API, where developers pay according to how much text, code, audio or imagery the models process. In its March 2026 funding announcement the company said it was earning about $2 billion a month and had more than 50 million subscribers.
The costs are as large as the revenue. Training and serving frontier models requires enormous data-centre capacity, and OpenAI has signed long-term computing commitments with Microsoft, Oracle, chip makers and others worth hundreds of billions of dollars. The company has not reported a profit. Its bet is that usage, and the price businesses will pay for AI that does real work, will grow faster than its computing bill.
Ownership and leadership
Under the October 2025 recapitalisation, the business is OpenAI Group PBC. The nonprofit, renamed the OpenAI Foundation, holds 26% of it, plus a warrant for more if the company reaches a valuation milestone, and it appoints the company's board, according to OpenAI's description of its structure. Microsoft holds about 27%. The remaining 47% belongs to current and former employees and other investors. Altman has said he holds no equity in OpenAI.
Altman is chief executive and Bret Taylor chairs the board. Greg Brockman, a co-founder, is president, and Fidji Simo, the former Instacart chief, runs the consumer and business products as CEO of Applications.
Controversies
DISMISSED, under appeal: Musk's lawsuit. Musk sued in 2024, arguing that OpenAI had abandoned its nonprofit mission. On 18 May 2026 a jury in Oakland found his claims had been filed too late, and the judge dismissed them. The jury did not rule on whether OpenAI broke its founding promises. Musk said he would appeal.
ALLEGED: copyright. The New York Times and other publishers and authors have sued OpenAI and Microsoft, alleging their work was used to train models without permission. OpenAI argues that training is fair use. These cases are unresolved.
ALLEGED: harm to users. Families in the United States have sued OpenAI alleging that ChatGPT contributed to self-harm. OpenAI denies responsibility and has added parental controls and other safeguards. No court has found it liable.
PROVEN: a governance failure. Whatever the merits on each side, the November 2023 crisis showed that the nonprofit board meant to restrain the company could not make its decision stick.
Why it matters
OpenAI matters because it set the terms of the AI era: the chatbot as the main way people use AI, subscription and usage pricing, and a willingness to raise and spend capital on a scale no software company had attempted. Its structure is an experiment in whether a nonprofit can keep control of one of the world's most valuable companies. The answer will shape how other AI developers are governed.