OpenAI
OpenAI Group PBC is an American artificial intelligence research organization headquartered in San Francisco, operating as a for-profit public benefit corporation partially controlled by a nonprofit foundation.[^c1] Founded in December 2015 as a nonprofit by Elon Musk, Sam Altman, Ilya Sutskever, Greg Brockman, and others,[^c2] the organization's founding mission was "to ensure that artificial general intelligence (AGI) benefits all of humanity."[^c3]
The November 2022 release of ChatGPT is widely regarded as triggering the generative AI boom and widespread public interest in artificial intelligence.[^c4] The company's product portfolio includes the GPT family of large language models (including GPT-5, released in 2025-2026), DALL-E for image generation, and Sora for video generation. Sora was discontinued as a standalone product in April 2026 after OpenAI announced it would shut down the service;[^c17] a reduced video generation capability (5-second clips, no audio) remains available to ChatGPT Plus and Pro subscribers.[^c18] By late 2025, ChatGPT had grown to over 800 million weekly active users across finance, healthcare, education, and media;[^c19] by May 2026 this had grown to over 900 million, with CFO Sarah Friar describing the company as facing "a vertical wall of demand" for its products.[^c41][^c42]
OpenAI has undergone significant structural evolution. Originally a nonprofit, it adopted a capped-profit model in 2019 with a 100-times return limit on investments.[^c6] In October 2025, the organization converted to a public benefit corporation (PBC),[^c7] with the OpenAI Foundation holding 26% equity, [[Microsoft]] holding 27%, and employees and other investors holding 47%.[^c8] Under the restructuring, OpenAI committed to purchasing $250 billion in Azure cloud services from Microsoft, while Microsoft's intellectual property rights were extended through 2032 but excluded OpenAI's consumer hardware.[^c21][^c22] In April 2026, the partnership was amended again: Microsoft's license to OpenAI's models became non-exclusive (while remaining in effect through 2032), and Microsoft's obligation to pay OpenAI a revenue share was eliminated, while OpenAI continued paying Microsoft through 2030.[^c28][^c29] On April 28, 2026, OpenAI models were deployed on Amazon Bedrock in a limited preview, two days after the amendment took effect, marking the first availability of OpenAI models through a competing cloud provider.[^c48] The renegotiation capped OpenAI's revenue-sharing payments to Microsoft at $38 billion through 2030, down from a potential $135 billion under the original arrangement.[^c32] The company raised $122 billion in committed capital at an $852 billion valuation, closing the round on March 31, 2026.[^c9][^c50][^c51] CFO Sarah Friar stated in May 2026 that OpenAI may pursue additional fundraising beyond this round, describing the company as operating in a "compute remains scarce" environment where computing power was a "huge competitive advantage."[^c36][^c37] A $100 billion investment and data-center partnership with [[Nvidia]] announced in September 2025 was ultimately abandoned; Nvidia finalized a $30 billion equity investment in OpenAI, and CEO Jensen Huang cited the company's plans for an initial public offering as the reason the larger deal could not proceed.[^c31] Annualized revenue reached approximately $20 billion in 2025, though the company continued to operate at a significant loss, with HSBC projecting losses of $23.5 billion in 2025 and potentially $60 billion by 2027;[^c20] a 2026 estimate projected a $14 billion loss for that year.[^c23]
In the first quarter of 2026, OpenAI reported approximately $5.7 billion in revenue, roughly $900 million ahead of rival Anthropic's $4.8 billion for the same period. The company posted an adjusted operating margin of negative 122 percent, meaning it lost $1.22 for every dollar earned.[^c61][^c62] Anthropic, however, had overtaken OpenAI in annualized revenue run rate by May 2026, reporting approximately $45 billion compared to OpenAI's $25 billion.
On June 8, 2026, OpenAI confidentially filed a draft IPO prospectus with the SEC and preemptively announced the filing, stating: "We expect it to leak so we're just announcing it."[^c63] The company joined SpaceX and Anthropic in an AI IPO pipeline valued at approximately $3.6 trillion.[^c18] On June 3, Altman visited Washington D.C. to meet with congressional leaders and White House officials, advocating against mandatory government pre-approval for AI models and instead pushing for increased funding for voluntary AI safety testing at the Department of Commerce.[^c64][^c65] He also met with Senator Bernie Sanders to discuss the concept of a sovereign wealth fund in which the public would hold equity stakes in AI companies. The same week, Altman joined rival AI CEOs in signing a joint letter to Congress demanding stricter regulation to prevent AI-assisted biological weapons development.[^c59]
In June 2026, OpenAI began the most significant overhaul of ChatGPT since its launch, transforming it into a "superapp" that integrates coding tools, AI agents, and third-party applications.[^c60] The redesign centered on [[Codex]], which had grown to over five million weekly active users, and reflected a conviction within the company that "chat is dead" and autonomous agents represent the future of AI products. On June 4, OpenAI launched Dreaming V3, a major upgrade to ChatGPT's memory architecture that reduced compute requirements by fivefold and doubled memory capacity for paid users. Also in early June, OpenAI announced its entry into robotics, with the OpenAI Robotics division led by DALL-E inventor Aditya Ramesh focused on assistive robots for skilled infrastructure workers.[^c57] Altman acknowledged at an event that customers were reporting they had spent their entire 2026 AI budgets in the first quarter, with concerns about AI spending emerging as a major issue.[^c58]
At its Dev Day 2025, OpenAI announced the Sora 2 API, the GPT-5 Pro model for enterprise use, a smaller voice model called GPT-Realtime Mini, and an agent-building platform.[^c24] The company's late-2025 pivot toward enterprise and agent-oriented computing reflected a recognition that "infrastructure, not algorithms alone, became the primary bottleneck" in AI development,[^c25] with CEO Sam Altman articulating a goal of "a world of abundant and cheap AI."[^c26] GPT-5.5, the company's first full model retraining since GPT-4.5, was released in April 2026 and positioned as an "agent runtime" for enterprise automation. In May 2026, an OpenAI general-purpose reasoning model autonomously disproved the planar unit distance conjecture, a problem open since 1946, in what external mathematicians described as "a milestone in AI mathematics."[^c54][^c55] That same month, OpenAI launched the [[OpenAI Deployment Company|OpenAI Deployment Company (DeployCo)]], a majority-owned subsidiary with over $4 billion in initial funding to help businesses build and deploy AI systems within their operations.[^c27] Also in May 2026, the company restructured its product organization, merging ChatGPT, Codex, and the developer API into a single division under co-founder [[Greg Brockman]], who became permanent head of product strategy.[^c34] Codex had surpassed 4 million users by May 2026.[^c38]
In May 2026, OpenAI began preparing potential legal action against Apple over the ChatGPT-Siri integration partnership, alleging that Apple had not fulfilled its contractual obligations after the arrangement failed to generate the billions of dollars in annual subscription revenue that OpenAI had expected.[^c30]
The organization has been marked by several major controversies. In November 2023, the board removed CEO Sam Altman, reinstating him five days later after nearly all employees threatened to resign.[^c12] The Superalignment safety team was dissolved in May 2024 after its co-leaders resigned, with departing researcher Jan Leike stating that "safety culture and processes have taken a backseat to shiny products."[^c13] By late 2024, only one of the four senior executives from a March 2023 photo remained at the company.[^c14] Elon Musk's lawsuit against OpenAI concluded in May 2026 when a federal jury returned a unanimous verdict for OpenAI and Altman, finding that Musk's claims exceeded the statute of limitations.[^c33] During the trial, testimony revealed that both Brockman and Altman held shares in the AI chip company Cerebras, with which OpenAI had signed over $20 billion in chip procurement agreements.[^c39][^c40] On June 1, 2026, Florida Attorney General James Uthmeier filed the first state-led lawsuit against OpenAI and Sam Altman, alleging the company knowingly concealed safety risks.
The company has also pursued a massive infrastructure expansion through the [[Stargate project|Stargate Project]], a $500 billion initiative in partnership with Oracle and SoftBank,[^c15] though the joint venture entity never became operational due to governance disputes and was replaced by bilateral agreements. By May 2026, OpenAI faced intensifying competitive pressure from [[Anthropic]], which had overtaken it in enterprise AI spending and valuation, reaching $965 billion compared to OpenAI's $852 billion.