OpenAI CEO Sam Altman confirms plans to build a humanoid robot, noting that creating the machine intelligence poses a much greater hurdle than hardware.
According to Engadget, OpenAI is expanding its development efforts beyond AI models and consumer hardware to create a humanoid robot. Speaking on the Sources podcast, CEO Sam Altman confirmed that the company is actively pursuing the project because the human world is specifically designed around human dimensions. Doors, tools, and keyboards require a matching form factor for effective navigation.
When asked directly if OpenAI would build one, Altman offered a succinct affirmation. However, he emphasized that designing the physical shell is less critical than developing the machine intelligence required to operate it. Giving a mechanical body arms and legs serves little purpose if the software lacks the capability to direct them properly. A functional humanoid must process visual sensor data, interpret instructions, execute actions, coordinate physical movement, and adjust dynamically when encountering real world obstacles.
Engadget reports that OpenAI is currently recruiting specialists across motion control, whole-robot behavior, state estimation, actuator control, and machine-learning infrastructure to address these physical AI challenges. The company previously explored physical manipulation through its Dactyl research project, which trained a robotic hand to handle objects using simulation before tackling a Rubik Cube.
Despite prioritizing the software brain, OpenAI is not neglecting the physical components. Job listings reveal that the company is developing custom electromechanical actuators that incorporate motors, sensing, thermal systems, and transmissions, alongside supply chain and manufacturing roles for scalable production. OpenAI joins a competitive landscape that includes Tesla with its Optimus project and XPENG, which recently showcased its IRON humanoid walking off a production line.
OpenAI has outlined distinct timeline expectations, with Altman telling Fortune that a demonstration could arrive in 2027, though commercial robots walking down public streets remain several years away. The organization has not yet disclosed consumer pricing, a final release date, privacy frameworks, or precise design specifications. Short-term objectives focus on deploying general-purpose robots to assist skilled workers with infrastructure tasks, while long-term ambitions point toward household personal assistance.
Background
OpenAI maintains a history of robotics research through projects like Dactyl and currently lists numerous job openings dedicated to advancing physical artificial intelligence systems. While the enterprise evaluates hardware manufacturing and custom electromechanical actuators, leadership maintains that the primary engineering obstacle involves developing sophisticated intelligence capable of navigating physical environments and executing complex physical tasks safely.
Why It Matters
The push by OpenAI into physical robotics signals a convergence between advanced digital intelligence and physical automation. By framing the robot brain as the principal engineering obstacle, the company highlights a broader industry shift where software capability outpaces the physical reasoning required for real-world deployment. If OpenAI manages to deliver a functional demonstration by 2027, it could accelerate the commercial timeline for autonomous mechanical helpers in both industrial workplaces and domestic environments. However, substantial questions remain regarding safety, consumer pricing, and data privacy for devices equipped with cameras and microphones operating inside private homes.
Illustration: AI-generated image, not a photograph of the event.