Sam Altman has a lot on his plate, and he just added a humanoid robot to it. Asked on the Sources podcast whether OpenAI will build one, he gave a two-word answer: “We will.” That’s about as committed as a CEO gets without a product to show.
According to Engadget, Altman argued that the human form makes sense because “the world is very much designed for people.” Doors, keyboards and tools all assume a certain body. He called the robot’s appearance “less important than really figuring out … the brain that makes the robot work.” He also told Fortune that OpenAI should have “a cool demo in 2027” and that robots walking the streets are “a few years longer” away. The company hasn’t said what the demo will show or what the machine will look like.
The brain framing is fair, and it’s also the hard part. A useful humanoid has to read camera and sensor data, interpret an instruction, plan the task, coordinate its limbs and fix its own mistakes when the world doesn’t cooperate. Picking up a mug means finding the handle, estimating the grip and applying the right force. Get it wrong and you have a coffee spill. OpenAI’s listed robotics roles cover motion control, state estimation, actuator control, whole-robot behavior and the ML infrastructure to train physical AI.
This isn’t new ground for the company. Its Dactyl project trained a robotic hand in simulation and later solved a Rubik’s Cube with it. It was an early demonstration of the gap that still plagues robotics: a model can learn what should happen in software, but physics has the final say.
Here’s where Altman’s pitch gets less tidy. The job postings point to custom electromechanical actuators, including motors, transmissions, sensing and thermal systems, plus manufacturing and supply chain roles for building hardware at scale. That’s a lot of effort for the supposedly less important half. OpenAI isn’t just writing models and hoping someone else builds the body. It wants to own the whole stack, much like Tesla with Optimus.
And Tesla is the obvious comparison. Optimus has been in development for years, and XPENG recently had its IRON humanoid walk off a production line on its own after assembly. Figure, Agility and Boston Dynamics are also in the race. OpenAI’s edge is its model research and its ability to attract talent and capital. Its disadvantage is time, since rivals already have working hardware and manufacturing experience.
The stated plan is modest. One posting says the near-term goal is general-purpose robots that help skilled workers build infrastructure, with a personal robot for a much wider range of tasks as the long-term target. That sequence is sensible. Industrial settings are more forgiving than a living room, and customers there will pay for partial capability.
So should you care? Not as a buyer. There’s no announced robot, price, subscription model, release date or privacy framework. That last one matters: a machine with cameras and microphones roaming a home raises questions OpenAI hasn’t addressed. But as a signal, it counts. The company that pushed language models into the mainstream is now betting that the same approach can reach the physical world, and it’s hiring accordingly, with 27 robotics openings listed. Watch the 2027 demo. Until then, treat it as a serious hiring push and a promise, not a product.
