OpenAI is preparing to take artificial intelligence beyond the computer screen and into the physical world. CEO Sam Altman has confirmed that the company is developing its own humanoid robots, with plans to demonstrate an early prototype as soon as 2027.
The first machines are expected to focus on industrial work, including tasks in factories and data centers. The longer-term ambition is considerably broader: robots capable of helping people with everyday chores at home.
Altman confirmed the company's robotics plans during a conversation with journalist Alex Heath on the Sources podcast. In a separate interview with Fortune, he also outlined the timetable.
"I think we'll have a cool demo in 2027," Altman said. "I don't think there will be robots walking the streets for a few years longer, but we'll have something impressive."
The announcement marks another step in OpenAI's effort to expand beyond the AI models and digital services that have made it one of the world's most prominent technology companies. This time, however, the challenge extends far beyond software. Building humanoid robots also requires expertise in hardware, mechanical engineering and manufacturing.
Not just talking, but doing
One of the biggest obstacles in developing robots is creating the intelligence that controls them. Unlike a chatbot, which receives a question and generates a response, a robot operating in the physical world must navigate an environment that can change at any moment.
It needs to recognize objects, determine where they are, make decisions in real time and translate those decisions into precise movements, all while avoiding harm to the people around it.
The decision to build robots that resemble humans is also deliberate. Most of the environments in which people live and work were designed around the human body, from staircases and doorways to corridors and everyday tools. In theory, a robot with a similar physical structure could operate in these spaces without requiring extensive modifications.
According to reports, OpenAI initially intends to deploy robots in work related to building facilities and operating data centers. Eventually, the company hopes to develop more versatile machines capable of performing a wider range of tasks beyond industrial settings.
OpenAI has not yet revealed what its robots will look like, what their precise capabilities will be or when commercial production might begin. The 2027 target refers to a prototype demonstration, not the launch of a product consumers will be able to buy.
OpenAI has tried building robots before
This is not OpenAI's first venture into robotics. The company previously operated a robotics research team before disbanding it. In recent years, it has resumed work in the field, apparently with more ambitious goals.
In January 2025, TechCrunch reported that OpenAI was planning to develop its own robots, including specialized sensors and hardware components. Job postings published by the company also pointed to plans for machines capable of adapting to changing tasks and environments rather than simply repeating a single predefined action.
A robot developed in collaboration with Nvidia, displayed at an Nvidia event
OpenAI's careers website now lists positions in mechanical engineering, control systems, electronics and safety. Among the specialists the company is seeking are engineers who can develop motors and actuators, the components responsible for converting computer commands into physical movement.
Those recruitment efforts suggest OpenAI is not content with merely developing the AI models that power robots. It is also investing in the physical systems needed to build them.
In theory, that approach would allow the company to develop hardware and software simultaneously, tailoring each to the other instead of relying entirely on robots manufactured by outside companies.
OpenAI also has a notable history in this field. In 2018, it introduced Dactyl, an AI system trained to control a robotic hand and perform delicate movements, including grasping and rotating objects.
A year later, the company demonstrated an even more impressive achievement: a robotic hand capable of solving a Rubik's Cube.
The breakthrough lay largely in the training method. Rather than teaching the robot through countless physical trials, researchers trained the AI system in computer simulations before transferring those skills to a real robotic hand.
The experiment demonstrated how artificial intelligence could acquire complex motor skills without physically experiencing every possible scenario. However, the leap from a robotic hand performing a specific task to an autonomous humanoid operating in unpredictable surroundings remains enormous.
Taking on Tesla and the other robotics giants
OpenAI is entering a market that is already attracting enormous investment and intense competition. In recent years, humanoid robots have become one of the technology industry's most closely watched developments, as companies race to transform advances in AI into machines capable of physical labor.
Among the best-known competitors is Elon Musk's Tesla, which has spent years developing its Optimus humanoid robot. Tesla envisions the machines working in factories before eventually expanding into household applications.
Tesla's Optimus humanoid robot
(Photo: Tesla)
Another prominent player is the American company Figure, which raised more than $1 billion in September 2025 at a valuation of $39 billion to support the continued development of humanoid robots.
Interestingly, OpenAI has already participated in the robotics industry through investments and partnerships with companies including Figure and 1X. By developing its own systems, it could eventually find itself competing against some of the companies with which it previously worked.
The biggest challenge, however, is not simply building a robot that can walk, pick up objects or perform an impressive demonstration. To become a useful commercial product, a humanoid robot must carry out tasks independently, operate reliably for hours, respond to unexpected changes and malfunctions, and above all work safely alongside people.
For now, OpenAI has not publicly demonstrated a functioning humanoid robot or committed to a commercial launch date. If Altman meets his stated timetable, 2027 could offer the first substantial glimpse of how the company intends to bring its artificial intelligence out of the digital world and into everyday physical life.