OpenAI has disbanded its standalone robotics team for the second time in six years, folding the group into its broader research and applied organisations. While the move echoes the company’s decision to wind down its original robotics division in 2020, OpenAI insists the latest reorganisation is not a retreat from robotics but rather a sign of its commitment to bringing general-purpose robots into the world, according to a Fortune report. The restructuring move from the ChatGPT-maker was confirmed when contacted by Fortune. As part of the new structure, the robotics researchers at OpenAI have been reassigned to the main research at the organisation headed by chief research officer Mark Chen, whereas applied robotics and hardware teams will be functioning under the broader Applied division at the company which is led by president Greg Brockman.
Second robotics reboot ai OpenAI ends
OpenAI had a robotics team in its early days but wound it down in 2020. Co-founder Wojciech Zaremba later cited limited robot data and a more efficient route to artificial general intelligence through other methods, notably training very large language models. That approach produced ChatGPT in late 2022.The company rebuilt a stand-alone team in 2024 and announced a partnership with Figure to develop AI models for humanoid robots. The partnership ended in February 2025, when Figure said it would build its own robot AI. OpenAI says it is developing its own humanoid robot, with CEO Sam Altman targeting a first public demonstration in 2027.
Why OpenAI believes robotics no longer needs a separate team
According to Fortune, OpenAI’s decision reflects a broader belief that advanced general-purpose AI models are becoming capable of handling robotics tasks without requiring specialised robotics-focused AI systems. The company said its researchers initially believed progress in robotics would come from technologies developed for projects such as Sora, OpenAI’s video-generation model, which was designed to understand physical-world interactions in order to generate realistic videos.Since then, OpenAI says its latest frontier models, including GPT-6 Astra, have demonstrated capabilities that can be applied directly to robots. The company said this discovery motivated it to move robotics researchers onto teams developing more general AI models instead of maintaining a separate robotics division. A company spokesperson told Fortune that OpenAI remains committed to accelerating progress toward general-purpose robots despite the organisational changes.
Experts divided
As reported by Fortune, Guanya Shi, the assistant professor at Carnegie Mellon University’s Robotics Institutem said that he believes that this move of OpenAI signals that robotics is “no longer a special problem,” but is a downstream application of general intelligence, same as power coding, legal and financial tools. He further added that the general models are already equipped to handle tasks which once required dedicated training, such as coping with a bottle in a different position or changed lighting. However, he still feels that there’s a need for smaller specialised models that run on a robot’s onboard computer without a constant internet connection.Russ Tedrake, an MIT professor and CEO of Walden Robotics, said robotics companies keep an advantage that AI labs lack: their robots are already gaining experience in real places such as factory floors. Better foundation models make that site-specific data more valuable, he said.Pete Florence, co-founder of Generalist, a startup training general-purpose AI models for robots, said there may be different starting points for growing intelligence and it is unclear that any current model will dominate robotics in the long run. The idea that general models can control robots directly would seem to challenge the premise of many robotics start-ups, which have raised billions on the belief that robots need brains of their own.
Cost pressure
OpenAI is also under pressure to cut costs and focus on a few core products. Last year Altman told the company to abandon “side quests,” and it has dropped some efforts, including Sora, to free up computing resources. The company is under pressure to reduce heavy operating losses, which reportedly totalled $10.9 billion in the latest quarter, as it seeks a new funding round and prepares for a possible IPO in early 2027.
Robotics expertise still matters, researchers say
Not everyone believes increasingly capable foundation models will eliminate the need for specialised robotics expertise. Russ Tedrake, an MIT professor and co-founder of Walden Robotics, told Fortune that robotics companies still possess a major advantage: real-world deployment data collected from robots already operating in factories and workplaces.That data allows robots to adapt to specific environments and tasks, helping bridge the gap between powerful general AI models and practical real-world applications. For now, OpenAI maintains that moving robotics researchers into other teams will speed up progress toward its long-term goal of building general-purpose robots, even as the company once again shutters a standalone robotics operation.