On the Sources podcast, Sam Altman confirmed that OpenAI intends to build a humanoid robot. Asked directly whether the company would, his answer was two words: "We will."
His reasoning for the human shape is practical rather than aesthetic. "The world is very much designed for people," he said. Doors, keyboards and tools all assume a particular body, so a robot meant to move through the same spaces should probably match it. "Matching that form factor seems good."
What he is not spending his attention on is the hardware. Those questions are "less important than really figuring out ... the brain that makes the robot work," he said. Arms and legs are not much use to a machine that does not know what to do with them.
The timeline is not soon
Altman told Fortune: "I think we'll have a cool demo in 2027 ... I don't think there will be robots walking the streets for a few years longer." OpenAI has not said what the demo will show, what the robot will look like, or when anyone outside the company will get near one.
Why it matters
- The "brain" framing is doing a lot of work. A useful humanoid has to see, interpret an instruction, plan, coordinate its body and correct itself when the real world interferes. Picking up a mug means identifying it, finding the handle, estimating the grip and applying the right force, or you get a spill and a mop.
- OpenAI has done this before, narrowly. Its Dactyl project taught a robotic hand to manipulate objects with training done entirely in simulation, and later solved a Rubik's Cube. That is the same wall every robotics company hits: a model can learn what should happen, but physics gets the final say.
- The hiring says the hardware is not an afterthought. Open roles cover motion control, state estimation, actuator control, whole-robot behaviour, custom electromechanical actuators with motors, transmissions, sensing and thermal systems, plus manufacturing and supply chain. There are 27 robotics openings on the careers page.
- The stated plan starts smaller than a robot in your kitchen. One posting describes a short-term goal of general-purpose robots that help skilled workers build infrastructure, with a personal robot as the long-term aim. No price, release date, subscription model or robot-specific privacy framework has been announced.
OpenAI is not first. Tesla has been working on Optimus for years, and XPENG's IRON humanoid recently walked off the production line on its own after assembly. Both are a reminder that good software still has to coax a complicated set of motors and joints into doing something useful.