OpenAI

OpenAI is an American artificial intelligence company known for developing ChatGPT, DALL-E, and Whisper, tools that helped spark the global boom in generative AI. Founded in 2015 by Sam Altman, Greg Brockman, Ilya Sutskever, and Elon Musk, the company began as a nonprofit research lab before evolving into a hybrid structure and, by 2025, a public benefit corporation overseen by the nonprofit OpenAI Foundation. The shift gave OpenAI more freedom to raise money for ambitious AI projects while keeping its founding goal—to create artificial general intelligence that benefits humanity—at the core of its mission.

Open full sized image

Sam Altman is CEO of OpenAI, the artificial intelligence company that developed ChatGPT.

© Justin Sullivan/Getty Images

OpenAI’s ambitious pursuit is the development of artificial general intelligence (AGI)—a theoretical iteration of AI that gives machines the adaptive ability to learn, understand, and apply intelligence to tasks across multiple domains in a manner that’s similar to, or exceeds, human capacity.

Although OpenAI’s declared goal is to develop AGI technology that “benefits all of humanity,” its potential to “outperform humans at most economically valuable work” has sparked debate regarding the broad disruptive effects—including the benefits and the risks—that such technologies could have on society, the economy, and the political landscape.

OpenAI’s product suite is anchored by the pairing of large language models (LLMs)—with data sets so massive they’re measured in petabytes—and natural language processing. These allow the models to respond to user prompts in human language (as opposed to computer programming language).

ChatGPT is a conversational AI application that understands and generates natural language to engage in human-like interactions, specifically in its responses to user prompts. The most advanced iteration, based on OpenAI’s GPT-5 model, can retrieve and summarize publicly available information from the Web.

DALL-E is an AI system that can create and manipulate images and visual art from natural language descriptions. The latest versions, DALL-E 2 and DALL-E 3, vary in capacity and policy (for example, DALL-E 3 can decline requests to mimic the style of a living artist).

Whisper is an open source machine learning model designed to automatically transcribe audio files and speech.

Codex is an AI model designed to understand and write human-readable code. Most notably, it’s the AI system powering GitHub Copilot, a tool that developers can use to write code in various languages and frameworks.

OpenAI was founded in December 2015 as a nonprofit organization dedicated to researching and advancing AI technology. Early investors included entrepreneur Peter Thiel and business entities such as Amazon Web Services (AWS), Infosys, and YC Research. These early commitments allowed OpenAI to begin with a $1 billion war chest to fund its research and operations.

OpenAI’s mission was to advance AI development in a manner that would “benefit humanity as a whole.” The nonprofit status was designed such that its early stages would be free of financial expectations or the obligation to turn a profit.

Although the company was relatively new, OpenAI’s funding commitment gave it the remarkable ability to offer competitive salaries to attract and retain top AI talent from Silicon Valley and beyond.

In just a few years’ time, OpenAI’s research and experimentation led to the accelerated development of innovations and products in the AI space, starting with its 2016 release of OpenAI Gym—a Python toolkit for designing “reinforcement learning” algorithms.

In 2017, OpenAI invested an estimated $7.9 million in cloud computing, which represented a significant portion of the company’s operational expenses that year. This was a pivotal strategic move, as cloud technology provided the kind of computational power and virtual scalability necessary for the rapid development of advanced AI models.

Not only did this accelerate research and improve efficiency; OpenAI’s entry into cloud computing also paved the way for future collaborations, most notably with Microsoft, whose Azure cloud computing platform would prove critical in its future funding, development, and organizational reshuffling.

In 2018, Elon Musk departed from OpenAI’s board. Although Musk initially cited a potential conflict of interest with his company Tesla, for which he served as CEO, reports later surfaced alleging internal conflicts as the reason for his departure. Musk would later become a vocal critic of OpenAI and AI technology in general.

In the same year, OpenAI unveiled the generative pre-trained transformer (GPT)—the first iteration of its generative large language model trained on a large and diverse body of textual information. Four years and two iterations later, the GPT language model would be released to the general public, rapidly revolutionizing almost every industry that utilizes text and code.

As OpenAI continued developing its AI technologies, it learned that its anticipated cost of production would exceed the capital it was able to raise through standard nonprofit channels. In 2019, OpenAI set up an unconventional hybrid nonprofit and for-profit model with a commercial arm that would allow it to raise more capital.

The for-profit arm of OpenAI partnered with Microsoft, which initially invested $1 billion, provided cloud computing through its Azure platform, and integrated OpenAI’s software into its products. Microsoft remained a primary cloud provider, but the partnership later evolved to allow OpenAI to offer its products across multiple cloud platforms and to operate with greater independence.

In 2020, OpenAI released a beta version GPT-3, its conversational AI chatbot. This third iteration—among the most advanced AI models for processing language and generating human-like text—would later power OpenAI’s flagship ChatGPT platform.

In 2021, OpenAI released two of its more acclaimed products—DALL-E and Codex—both marking significant, if not pivotal, advances in image and code generation from natural language prompts. Although both platforms featured new and arguably disruptive functions within specific industries, it would be the following year when OpenAI would release a product that would disrupt nearly every industry, rattling the foundations of digital-age labor and heralding the call for more comprehensive regulation and oversight.

In November 2022, OpenAI launched its ChatGPT model, making it widely available to the general public. ChatGPT’s transformative impact on society at large was immediate. It was able to generate natural language in a wide variety of textual styles and genres using OpenAI’s most advanced models. ChatGPT also demonstrated some capacity to pass college-level and professional licensing exams in fields such as medicine, law, business, and other general subjects.

ChatGPT failed several exams, but its natural language processing (NLP) capabilities and vast knowledge base raised concerns that it could replace human responses—and, eventually, a large segment of the workforce.

The feared dystopian narrative of machines replacing humans remains a matter of speculation and debate. But many proponents argue that AI technology, if not a replacement for human labor, may at least create a divide between human workers capable of accessing and working with AI and those without it.

In November 2023, OpenAI’s board of directors briefly ousted Altman as CEO and stripped Brockman of his position as board chair. Mira Murati, the company’s chief technology officer (CTO), was named interim CEO. A swift backlash, both internally (among OpenAI employees) and externally (namely, from Microsoft, which owns 49% of OpenAI’s for-profit arm), led to the reinstatement of Altman as CEO and Brockman as president.

The board’s reconstitution in late 2023 resulted in:

The change in board makeup marked the next stage in OpenAI’s evolution, particularly as it relates to its bifurcated nonprofit/for-profit hybrid status. As the sophisticated research and computing power used to support the company’s large language models (LLMs) continues to grow, the emergence of a deep-pocketed commercial entity such as Microsoft was a natural step. The transition of the company mandate from nonprofit to for-profit comes with strings attached, including the desire for a return on investment, but also increased regulatory scrutiny.

Ilya Sutskever, cofounder and chief scientist who was involved in Altman’s ouster (but later expressed regret), resigned in May 2024 and was replaced by Jakub Pachocki, the company’s director of research. Murati, who had returned to her position as CTO after the November 2023 shake-up and reversal, announced her departure in September 2024. In their resignation announcements, both Sutskever and Murati indicated a desire to embark on their own projects.

Other departures included chief research officer Bob McGrew and vice president of research Barret Zoph. Both resignations coincided with Murati’s departure, and both were driven by interests to pursue other opportunities. Although these departures may have been decided independently and amicably, they coincided with the company’s planned restructuring into a for-profit entity, suggesting that the directional shift may have influenced the leadership changes.

In spring 2024, OpenAI entered into partnerships, further integrating its products with those offered by major technology and media companies:

In addition to these collaborations, OpenAI acquired Multi, a start-up that developed an AI-assisted video conferencing tool, in June and the Web domain Chat.com in November.

OpenAI also introduced several major releases in 2024, including:

In October, OpenAI transitioned to a public benefit corporation (PBC), marking the culmination of a multiyear effort to redefine its business model. The company converted its for-profit subsidiary, OpenAI Global LLC, into OpenAI Group PBC, a structure that formalized its shift to a more conventional corporate framework from a combined nonprofit and for-profit structure, while keeping overall control with its nonprofit parent, the OpenAI Foundation.

Under the arrangement, the foundation retained authority over the for-profit entity and holds a 26% stake, while Microsoft owns about 27%, with employees and other investors sharing the remainder. The foundation’s stake—valued at roughly $130 billion at the time—made it the most valuable charitable foundation in the United States, surpassing the endowment of the Gates Foundation. Microsoft’s position reinforced its central role in OpenAI’s financing and technology development; its market value briefly topped $4 trillion after the restructuring was finalized.

The shift followed nearly a year of negotiations with regulators in California and Delaware, who endorsed the plan after securing commitments that OpenAI would keep its headquarters in California, maintain nonprofit oversight of its commercial arm, and strengthen safeguards related to product safety and child protection, according to The Wall Street Journal. The foundation’s board retained authority to appoint and remove members, and it continued to include a safety committee empowered to delay or block the release of new AI models.

The streamlined structure enabled OpenAI to raise capital more easily and positioned it for a possible initial public offering (IPO) to support its global research and infrastructure expansion. The company also renegotiated its long-term agreement with Microsoft. Although the software giant retained commercial rights to OpenAI technologies and early access to its research through 2032, OpenAI was permitted to work with other cloud-computing providers.

In April 2026, OpenAI and Microsoft further revised their partnership, capping the revenue share OpenAI pays to Microsoft and removing exclusivity provisions tied to its technology. The changes allowed OpenAI to distribute its products across multiple cloud providers while maintaining Microsoft as a primary partner, underscoring the company’s broader push to reduce reliance on a single infrastructure provider.

Altman characterized the restructuring as giving OpenAI “the flexibility to scale safely and sustainably.” Board chair Bret Taylor called it “a direct path to major resources before AGI arrives,” reaffirming OpenAI’s goal of developing artificial general intelligence that benefits humanity.

Earlier in October, OpenAI signed a chip-supply agreement with Advanced Micro Devices (AMD) to expand its AI infrastructure. Under the terms of the deal, AMD would provide its next generation of graphics chips for use in large-scale data centers, giving OpenAI access to up to six gigawatts of computing capacity. The first gigawatt is scheduled for deployment in the second half of 2026. The pact also gives OpenAI the option to acquire up to a 10% stake in AMD, tied to the rollout of the computing capacity. Analysts viewed the arrangement as an effort to reduce OpenAI’s dependence on hardware from leading AI chipmaker NVIDIA (NVDA).

Also in October, OpenAI introduced Atlas, a free web browser for Apple computers using ChatGPT as its interface, replacing the traditional address bar with natural language queries. The launch marked the company’s entry into the browser market and was seen as an effort to make ChatGPT a central tool for searching the web and managing online tasks. Atlas also gave OpenAI a new way to collect data and refine tools that let ChatGPT handle routine online tasks.

In November, OpenAI agreed to pay Amazon Web Services (AWS) $38 billion over seven years for computing infrastructure powered primarily by NVIDIA (NVDA) graphics-processing units (GPUs), along with supporting central processing units (CPUs). The deal gave OpenAI capacity to train and run advanced AI models, including agentic AI, systems capable of completing tasks autonomously.

The agreement advanced OpenAI’s strategy to spread workloads among multiple providers. It also strengthened Amazon’s position in the growing market for AI infrastructure and services.

Before the company’s restructuring was completed, OpenAI faced mounting regulatory scrutiny and investor pressure that helped shape the changes to its corporate structure later in the year. In February, the U.S. Securities and Exchange Commission opened an investigation into Altman to determine whether the CEO had misled investors during the temporary leadership shake-up in late 2023. That same month, Elon Musk filed suit accusing the company of anticompetitive practices in its transition to for-profit status, arguing that OpenAI had shifted its priorities toward profit and away from its original mission.

Weeks later, Musk and a consortium of investors submitted a $97.4 billion bid to take control of OpenAI, adding another chapter to the rivalry between Musk and Altman. The proposal, formally submitted in February, was publicly rejected by OpenAI’s board soon after. At about the same time, OpenAI closed a $40 billion funding round led by SoftBank, one of the largest in history, valuing the company at roughly $300 billion.

In May, OpenAI agreed to acquire IO, an AI hardware start-up founded by former Apple design chief Jony Ive, for $6.5 billion. The deal was completed in July 2025, bringing Ive and his design team into key product-development roles at OpenAI. Altman said the acquisition was intended to help create “a new category of personal computing devices,” with product details expected in 2026.

That same month, the company released ChatGPT-4o (“omni”), a multimodal model capable of processing and generating text, images, and audio in real time.

In June, a group of current and former employees issued an open letter calling for greater transparency about AI risks and stronger protections for whistleblowers and others who raised internal concerns. The letter drew renewed attention to debates about governance and safety that had intensified since the company’s leadership crisis in 2023.

OpenAI released its ChatGPT-5 model in July, describing it as a major advance in reasoning and autonomy. The launch proved difficult after users reported an increase in factual errors, privacy breaches, and misleading responses linked to the model’s expanded capabilities. Several organizations stopped using it when sensitive data were exposed through third-party apps.

The breaches prompted regulators in the United States and European Union to open inquiries into data protection and transparency. Altman later apologized and temporarily halted sign-ups while software fixes were made. The episode deepened scrutiny from investors and lawmakers about the company’s rapid pace of product development and oversight.

By 2026 OpenAI had begun operating more like a mature technology company than the research lab it had been at its founding. Following its 2025 restructuring, the company expanded the commercial reach of ChatGPT and other platforms and tools as companies continued integrating them into software platforms, search tools, and enterprise applications. As demand surged, so did the company’s need for capital to support the increased computing capacity and processing power required to train and operate increasingly sophisticated AI systems, fueling speculation that OpenAI could eventually seek access to public capital markets.

OpenAI’s rapid commercial expansion also continued to draw scrutiny over the company’s evolving mission and governance structure. In 2026 a federal trial in California focused on a lawsuit brought by Musk, who alleged that the company had abandoned its original nonprofit mission in favor of commercial interests tied to its partnership with Microsoft and its restructuring efforts. A jury ultimately ruled against Musk, finding that he had waited too long to file his claims, and the court dismissed the lawsuit.

In June 2026, the company announced that it had submitted an S-1 filing to the U.S. Securities and Exchange Commission, taking its first formal step toward an IPO. Because the filing was submitted under the SEC’s confidential review process, detailed financial information was not immediately made public. The move didn’t guarantee an IPO, but it marked the company’s clearest indication yet that it intended to seek a public listing.