Sam Altman said declining to take a direct equity stake in OpenAI was a mistake that later created confusion about his relationship with the company, adding a rare personal governance detail to a wide-ranging discussion of OpenAI's rise and its push toward more capable AI systems.
In a nearly 70-minute interview with Indian podcast host Raj Shamani published on YouTube on Nov. 3, 2026, the OpenAI chief executive said he had held no direct ownership in the company from the beginning. Altman described that choice as a practical error rather than an attempt to distance himself from OpenAI, saying the absence of a conventional stake made it harder for outsiders to interpret his incentives and authority.
The comments arrive after years of scrutiny around OpenAI's unusual corporate structure, which combines a nonprofit governing body with a capped-profit operating business. Altman has long been associated with the organization's commercial expansion, from the release of ChatGPT to partnerships that financed increasingly expensive model training. His admission adds another layer to questions about how OpenAI balances mission-driven governance with the capital demands of frontier AI development.
Scaling remained OpenAI's central technical bet
Altman used much of the interview to explain why OpenAI continued to spend aggressively on larger models and computing infrastructure even as critics argued that large language models were approaching diminishing returns.
He said the company had repeatedly observed that techniques producing useful results at a smaller scale could generate much larger gains when expanded by factors of several times or even tens of times. That pattern, commonly called a scaling law, describes how model performance can improve predictably as developers add computing power, training data, and model parameters.
Altman traced OpenAI's confidence in that approach back to the period following AlexNet, the influential neural network that helped demonstrate the potential of deep learning in image recognition. The industry could already see a relationship between larger neural networks, greater data availability, and stronger performance, he said.
The commercial consequences of that view have become more visible as advanced AI companies compete for access to data centers, high-end chips, electricity, and cloud capacity. Frontier-model development increasingly depends on infrastructure that few companies can finance or operate alone, giving large technology groups and their strategic partners an advantage over smaller AI labs.
GPT-4 tests shifted OpenAI's planning
Altman identified August 2022 as a major internal turning point for OpenAI. GPT-4 had completed training months before its eventual public release in March 2023, and the company was evaluating what it could do through internal testing.
Those tests persuaded OpenAI that the model showed a route toward more advanced reasoning behavior, Altman said. He referred to chain-of-thought reasoning, a technique in which a model breaks a problem into intermediate steps rather than immediately producing a final answer. The approach has become central to efforts to improve AI performance in mathematics, programming, science, and other tasks where an answer requires multiple stages of inference.
OpenAI subsequently made a heavier commitment to computing capacity, according to Altman. The decision came while public discussion still included frequent claims that language models had reached a technical wall. OpenAI's experience with GPT-4 led its leadership to take the opposite position: larger-scale training and inference systems remained capable of delivering meaningful improvements.
That bet has also changed the economics surrounding AI. Training top-tier models requires enormous upfront spending, while running them for millions of users creates a separate and ongoing demand for chips and data-center capacity. The pressure to secure those resources has placed AI infrastructure alongside semiconductors and cloud computing as a major destination for private technology capital.
ChatGPT was released despite internal resistance
Altman also described a more uncertain moment before the November 2022 release of ChatGPT. According to his account, the early version recorded second-week user retention of 5% during testing. On its face, that was a modest figure for a consumer internet product.
OpenAI's internal reviews found that the smaller group of returning users was engaging with the prototype unusually heavily, Altman said. That behavior convinced him and others that the product had enough promise to be released as a research preview, despite broad resistance within the organization.
The launch quickly reshaped the public AI debate. ChatGPT gave users a direct way to test conversational AI for writing, coding, research, and creative tasks, exposing both the technology's practical appeal and its tendency to produce errors. It also accelerated competition among major technology companies to release comparable products and build AI features into search engines, office software, developer tools, and consumer applications.
Altman's account suggests OpenAI treated the launch less as a polished consumer-product rollout than as a decision to expose a compelling but imperfect research system to a much larger audience. The 5% retention figure mattered because it indicated that a small group had already found recurring use cases, even before the product's capabilities were widely understood.
Departures reflected disputes over control
The OpenAI chief executive also addressed the departures of several prominent figures linked to the organization at different stages, including co-founder Ilya Sutskever, former chief technology officer Mira Murati, former research vice president Dario Amodei, and early backer Elon Musk.
Altman connected those splits to disagreements over direction, governance, and who should control work approaching superintelligence-level systems. As AI capabilities advance, he said, senior leaders often want a stronger role in setting the path rather than remaining within another executive's structure.
The list covers departures across OpenAI's history and includes people who went on to pursue other AI projects. Amodei co-founded Anthropic, which has become one of OpenAI's most prominent competitors in frontier-model development. Musk has built xAI, while Sutskever later launched Safe Superintelligence Inc. Murati announced a new AI venture after leaving OpenAI.
Those exits illustrate how disagreements about AI safety, commercial strategy, research priorities, and governance can become difficult to separate when the same people are building systems expected to influence major parts of the economy.
Scientific discovery is the next target
Looking ahead, Altman said OpenAI wants to push AI beyond its current role as an engineering and productivity tool toward generating new scientific discoveries. He pointed to early work involving the Navier-Stokes equations and difficult mathematics problems.
The Navier-Stokes equations describe how fluids such as air and water move. They are fundamental to physics and engineering, yet major questions about their mathematical behavior remain unresolved. Altman said he felt acute tension shortly before OpenAI planned to announce results connected to the equations, suggesting the company sees scientific applications as a more consequential benchmark than consumer chatbot adoption.
Altman also discussed meditation and career development, describing a past period in which he meditated for two hours a day. On careers, he argued that progress compounds when people continue to take on new challenges instead of repeating the same year of experience over decades.
His remarks presented OpenAI's strategy as a long-running commitment to scale, rapid deployment, and scientific ambition, an approach that has produced major advances, intense competition, and persistent disputes over who gets to direct the systems being built.
Want deeper insight into AI and finance’s future? Explore our guide on how AI complements blockchain in real-world innovation.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
