NewsStocksOpenAI Disbands Stand-Alone Robotics Team as Its General Models Start Running Robots

OpenAI Disbands Stand-Alone Robotics Team as Its General Models Start Running Robots

Author: Fortune Crypto·

Key Takeaways

  • •OpenAI disbanded its stand-alone robotics team, moving the robotics research group under Chief Research Officer Mark Chen and the applied robotics and hardware teams under the Applied organization led by President Greg Brockman.
  • •The company said the restructuring reflects its discovery that general-purpose frontier models, including GPT-6 Astra, can be applied directly to robotics without dedicated teams.
  • •In an Nvidia study, an Astra-powered agent achieved a 74.8% success rate on a simulated manipulation benchmark, compared with 51.0% for the best specialized robotic models.
  • •The reorganization comes as OpenAI reportedly posted $10.9 billion in operating losses in its most recent quarter while pursuing a new funding round and a possible early-2027 IPO.
  • •Robotics experts cautioned that specialized models remain necessary, citing robots' need for real-world deployment data and smaller onboard models that operate without constant cloud connectivity.
OpenAI Disbands Stand-Alone Robotics Team as Its General Models Start Running Robots

OpenAI has disbanded its stand-alone robotics team and reassigned its researchers to other groups across the company. The reorganization was confirmed after Fortune inquired about the move. A company spokesperson said robotics remains a core part of OpenAI's long-term vision and framed the restructuring as a sign of the company's commitment to the field rather than a retreat from it.

Under the new structure, OpenAI has reassigned its robotics research team to the company's main research team, where it reports to Chief Research Officer Mark Chen. The applied robotics and hardware teams have been moved into the company's larger Applied organization, which reports to OpenAI President Greg Brockman.

The change came recently enough that OpenAI had already registered a booth at CoRL (the Conference on Robot Learning), a leading academic robotics conference held in Austin this week. Although the conference app and floor plan still marked OpenAI's spot, Fortune found booth #208 sitting empty, with the company's logo nowhere to be seen.

A Winding Road for Robotics at OpenAI

OpenAI maintained a robotics team in its early days but wound it down in 2020, with cofounder Wojciech Zaremba later citing limited robot data and a more efficient path toward artificial general intelligence (AGI)—the human-like AI that has been OpenAI's goal since its founding—through other methods. In particular, the company became increasingly devoted to training ultra-large language models on vast datasets of human-written text, the approach that eventually produced its breakthrough product, ChatGPT, in late 2022.

OpenAI rebuilt a stand-alone robotics team in 2024, announcing a partnership with the startup Figure to develop AI models for humanoid robots. That partnership ended in February 2025, when Figure said it would develop its own robot AI. OpenAI has said it is now developing its own humanoid robot, with CEO Sam Altman targeting a first public demonstration in 2027.

Despite the reorganization, the spokesperson said OpenAI remains committed to bringing general-purpose robots into the world.

A Wager on Embodied AI

The restructuring reflects a bet gaining traction across the AI industry: that general-purpose models built for language, code, and vision can also be used for "embodied AI," in which AI models control devices that act in the physical world, such as robots, driverless cars, or drones. If that thesis holds, it would eliminate the need for separate robotics teams building specialized AI models solely to serve as the digital "brains" of robots.

Researchers have already observed surprising performance when connecting GPT-6 Astra, OpenAI's general-purpose frontier model, to robots, directly using the model to perform certain robotic tasks.

Guanya Shi, an assistant professor at Carnegie Mellon University's Robotics Institute who had previously heard about OpenAI's reorganization, told Fortune that he would interpret the move as a signal that OpenAI sees robotics as "no longer a special problem." Instead, he said, the company likely views robotics as a challenge that can be addressed as a downstream application of general intelligence—similar to the way the same underlying frontier AI models now power specialized coding tools, or tools for legal and financial work.

OpenAI appeared to confirm this view in its response to Fortune. The company said that at the start of the year, its researchers assumed the best way to make progress on robotics was to build on its work on AI models such as Sora, a video generation model that, in order to create realistic videos, had to develop some implicit understanding of how the physical world works. Since then, the spokesperson said, the company's general-purpose frontier models, including GPT-6 Astra, have shown they can be directly applied to robotics. That discovery motivated OpenAI to disband the stand-alone robotics team and move its affiliated researchers onto teams working on more general AI models.

Cost Pressures and a Narrowing Focus

The reorganization also comes at a time when OpenAI is under pressure to rein in its costs and demonstrate that it is more focused on a few core products. Last year, Altman told the company it needed to abandon "side quests," and OpenAI has since jettisoned some efforts—including Sora—to reallocate computing resources to more core and potentially profitable endeavors. The company is under pressure to stanch its heavy operating losses, which reportedly totaled $10.9 billion in the most recent quarter, as it seeks to close a new funding round and prepares for a possible IPO in early 2027. Those milestones—the new funding round, the possible early-2027 IPO, and Altman's targeted 2027 demonstration of the company's humanoid robot—now double as the nearest points at which the results of this reorganization will become visible.

A Challenge to Robotics Startups—With Caveats

This line of thinking—that general AI models can be applied directly to controlling robots, without the need to create specialized models with "spatial intelligence"—would seem to challenge the assumptions underpinning many of today's robotics startups, which have raised billions of dollars on the premise that robots need brains of their own. Figure, for one, said in February 2025 that it would develop its own robot AI after ending its OpenAI partnership—effectively a bet on dedicated robot models.

Russ Tedrake, an MIT professor and cofounder and CEO of Walden Robotics, countered that robotics companies still hold an advantage the AI labs lack: their robots are already gaining work experience in real places, such as on factory floors. Such "deployment" experience—and the data these robots gather—helps robots adapt to specific workplaces and tasks. Better foundational models make that site-specific data more valuable, Tedrake said, because robots can learn from it faster.

Recent experiments have tested how much work a general-purpose model can do without specialized training on robotic data. Researchers gave Astra camera images, robot information, and a control interface that allowed the model to command the robot to take certain actions. In an Nvidia study, an Astra-powered agent achieved a 74.8% success rate on a simulated manipulation benchmark—in which the model commands a simulated robot in a digital simulation, as opposed to a real robot—compared with the 51.0% reported by the best specialized robotic models on the benchmark leaderboard.

Shi said advances in general models' spatial and reasoning capabilities already let them handle tasks that previously required dedicated training. They can cope when a bottle is in a different position or the lighting changes, he said, and can reason about the sequence of a task or identify which step went wrong.

Meanwhile, large language models remain powerful but slow, and they need to be hosted in the cloud, meaning the robot or device requires a constant network connection. Shi still sees a need for smaller, specialized models that can run on a robot's onboard computer without needing constant internet access.

"There may be different starting points for how to grow intelligence," Pete Florence, cofounder of Generalist, a startup training general-purpose AI models for robots, told Fortune. He said it was unclear that any existing model would be the one to dominate robotics in the long run.

For its part, OpenAI said it reassigned its robotics researchers to other teams because it now believes this is the best way to accelerate progress toward building general-purpose robots.

Back at booth #208 at CoRL, a few former members of the once-unified, stand-alone OpenAI robotics team milled about. They still drew a crowd, despite having no display or demo. After all, one of the conference's hottest questions was how much robotics expertise would remain valuable as general-purpose models grow more capable—and the experience of those former OpenAI robotics researchers, now dispatched to other teams, offered one answer.

This story was originally featured on Fortune.com.