OpenAI Plans Humanoid Robot Prototype Demo in 2027

OpenAI expects to unveil a humanoid robot prototype in 2027, CEO Sam Altman said on the Sources podcast.

Altman presented that timeline alongside a clear qualification: the project’s core difficulty is not the robot’s mechanical form but the artificial intelligence required to control and guide a humanlike body. When asked whether OpenAI will build such a robot, he replied, "We will do it."

Altman said the primary challenge is developing the AI to control the robot, not its physical body.

Technical challenge: the robot's brain

Altman argued that a human-form chassis is logical because our environments are designed for human bodies: doors, keyboards and tools assume human proportions. "Adapting to that body form seems reasonable," he said on Sources.

Building the robot’s brain, he explained, means developing systems that can understand data from cameras and sensors, interpret commands, decide how to act, coordinate bodily movements and adjust performance when facing unexpected situations. Altman noted that having two hands and two feet is not enough; the machine also has to know how to use those limbs to perform tasks.

Timing and unknowns

In an interview with Fortune, Altman said, "I think we'll have an interesting demo in 2027. I don't think the robots will be walking in the streets for several years after that." OpenAI has not disclosed what capabilities the prototype will demonstrate, what the robot will look like or when people outside the company might be able to see it in person.

Altman gave a concrete example of the problem space: a robot that intends to pick up a mug must first identify the mug, locate the handle, determine how to grasp it and apply the appropriate force to lift it. If any of those steps are handled incorrectly, the robot could spill the mug's contents instead of picking it up.