OpenAI CEO Sam Altman has confirmed the company is developing humanoid robots, with a potential demonstration in 2027 and a strong focus on artificial intelligence.
OpenAI, known for developing advanced artificial intelligence models, is reportedly expanding its work into humanoid robotics. CEO Sam Altman has confirmed that the company is working on humanoid robots, raising expectations about its plans to bring AI into the physical world.
During an appearance on the Sources podcast, Altman explained that human environments are designed around the human body. From doorways and stairs to everyday tools and household objects, most surroundings are built for people rather than machines.
This makes a human-shaped robot a potentially practical design choice. A humanoid machine could navigate familiar environments and interact with objects without requiring extensive changes to existing infrastructure.
However, OpenAI has not yet revealed the appearance, technical specifications or intended applications of its proposed robot.
Altman has indicated that the public could get an early look at OpenAI's humanoid robotics work in 2027.
Speaking to Fortune, the OpenAI CEO suggested that the company could present a demonstration during that year. Such a showcase could provide an initial indication of how its AI technology might operate inside a physical machine.
However, Altman also cautioned against expecting humanoid robots to become a common sight in everyday public spaces immediately. He suggested that several more years could pass before robots routinely walk among people on streets.
This indicates that OpenAI's near-term objective may be to demonstrate its technological progress rather than introduce a mass-market humanoid product.
The company has not announced a confirmed launch date, pricing details or a commercial rollout schedule.
Altman has emphasised that the intelligence powering a robot could be more important than its physical appearance.
A humanoid robot requires coordinated arms, legs, hands, fingers and other mechanical components. Yet the hardware alone cannot make a machine useful in unpredictable real-world situations.
Robots need AI systems capable of processing information from cameras and sensors, understanding their surroundings, planning actions and controlling movements.
For example, a robot performing household tasks would need to recognise different objects, determine how to handle them and adjust its movements when circumstances change.
Developing reliable systems for these activities remains a major technical challenge. OpenAI's focus on the intelligence behind the machine suggests that its research could centre on connecting advanced AI models with physical actions.
Reports indicate that OpenAI established a humanoid robotics laboratory in 2025 to develop AI technologies for robots.
The company's recruitment activity also points towards efforts to expand its robotics expertise. According to reports cited in the supplied article, OpenAI has listed 27 robotics-related vacancies covering areas such as motion control, whole-robot behaviour, manufacturing and supply-chain operations.
These roles suggest that the company's ambitions may extend beyond software development to include the practical challenges of building and producing robotic hardware.
Motion control helps robots coordinate their physical movements, while whole-robot behaviour involves managing multiple systems to complete tasks. Manufacturing and supply-chain capabilities would become increasingly important if OpenAI eventually decides to produce robots at scale.
Nevertheless, the company has yet to disclose the full scope of its robotics programme or the hardware it intends to develop.
OpenAI has explored the relationship between AI and robotics before announcing its current humanoid ambitions.
One notable example was Dactyl, a research project that trained a robotic hand to manipulate physical objects using techniques involving simulation and reinforcement learning.
The project demonstrated how an AI system could learn complex hand movements and apply its training to physical tasks. OpenAI also used related techniques to teach a robotic hand to solve a Rubik's Cube.
This earlier work provides useful context for the company's current direction. Rather than treating robots simply as mechanical devices, such research focuses on enabling machines to learn, reason about actions and interact with the physical environment.
The extent to which these earlier methods will contribute to a future OpenAI humanoid remains unclear.
OpenAI is entering a competitive field in which established technology companies and robotics startups are developing increasingly capable humanoid machines.
Tesla, led by Elon Musk, has been developing its Optimus humanoid robot, with ambitions to use robotics in industrial and other practical settings.
China's Unitree has attracted attention for its advances in humanoid robot mobility, including demonstrations of fast running and dynamic movement.
Chinese electric vehicle maker XPeng is also advancing humanoid robotics, including demonstrations related to autonomous movement and automated production.
These developments illustrate how companies are working to improve robot mobility, coordination, manufacturing and practical usability. OpenAI will need to translate its AI expertise into dependable physical capabilities to compete effectively in this market.
Humanoid robots are part of a wider trend in which AI companies are exploring products that bring their technology beyond computers and smartphones.
Reports have also suggested that OpenAI is developing consumer hardware, including a possible AI-powered smart speaker. Such devices could offer new ways for users to interact with AI through voice and other interfaces.
However, the company's plans for these products remain subject to further announcements, and details about specifications, availability and pricing have not been confirmed in the supplied report.
Conclusion
OpenAI's humanoid robotics initiative could represent an important step towards integrating advanced AI with physical machines. Sam Altman's indication of a possible 2027 demonstration offers an early timeline, although widespread adoption may take longer. The project's success will depend on developing reliable robotic intelligence, effective hardware and practical real-world applications.