On June 1, Sam Altman, CEO of OpenAI, officially announced the company's strategic entry into the physical robotics SECtor via a recruitment post on social media. altman revealed that the company is establishing a new team, "OpenAI robotics," and is ACTively recruiting exceptional full-stack hardware, Operations, systems, and machine learning engineers. The team's mission is to "program and manufacture robots that are truly useful for society."
According to Altman, openai's Robotics strategy is divided into short-term and long-term objectives. In the short term, the company will focus on developing assistive robots designed to help Skilled workers build future infrastructure. Looking further ahead, OpenAI envisions a future where everyone can own a personal robot capable of handling a wide variety of daily needs.
Altman noted that this strategic move is driven by the rapid advancements of an internal research project called "Worldsim" (World Simulation). Over the past year, this project has evolved into OpenAI Robotics and is now led by Aditya Ramesh, OpenAI's Vice President of Research and the core developer behind the DALL·E Image Generation and Sora Video Generation models. The foundation of this initiative lies in the deep integration and co-design of robotic hardware and machine learning research.
A Strategic Return: The Legacy of Dactyl
This move marks a significant "return" for OpenAI. In its early years, robotics was a crucial direction for its exploration of artificial general intelligence (AGI). Between 2016 and 2019, OpenAI launched the reinforcement learning benchmark environment OpenAI Gym and the open-source robot simulation platform Roboschool. The company also successfully developed "Dactyl," a dexterous robotic hand.
In 2019, using Reinforcement Learning and "Automatic Domain Randomization" (ADR) Technology, OpenAI trained an AI system that enabled a humanoid robotic hand to successfully solve a Rubik's Cube. This breakthrough dEMOnstrated the feASIbility of training capabilities in a simulated environment and transferring them to real-world robots. However, due to the scarcity of robotics training data and slow iteration speeds at the time—contrasted with the massive availability of internet text and image data—OpenAI made a strategic decision around 2020 to disband its robotics team. Resources were redirected toward the development of large language models (LLMs), a decision that ultimately led to the creation of ChatGPT.
IPO Ambitions and Financial Pressures
In the years following, OpenAI ignited a global LLM boom with its chatgpt series, becoming the world's most valuable AI unicorn. Reports indicate that OpenAI secretly submitted a draft IPO prospectus on May 22, planning for a potential listing as early as September 2026. Following its latest funding round in March, the company's Valuation reached $852 billion. Institutions like Deutsche Bank predict its IPO valuation could exceed $1 trillion, with a Fundraising scale of up to $60 billion, potentially making it the largest tech IPO in U.S. public market history.
However, OpenAI faces immense financial pressure. The company forecasts a loss of APProximately $14 billion for the full year of 2026, with cash burn expected to widen. positive cash flow is not anticipated until 2030 at the earliest, with gross margins sitting at only around 33% due to the high inference costs of AI models.
From Investment to In-House Development
During the years its internal robotics team was disbanded, OpenAI did not completely abandon the sector. Instead, it adopted a "spray and pray" Investment Strategy through its venture fund, backing several robotics startups such as Norway's 1X Technologies, U.S. humanoid robot star Figure AI, and Physical intelligence.
The most notable collaboration was with Figure AI in February 2024. OpenAI not only participated in Figure AI's $675 million Series B Funding but also announced the development of an exclusive multimodal AI model for Figure's robots. Just 13 days after the partnership was announced, the Figure 01 humanoid robot, powered by OpenAI's model, demonstrated Fluent natural language interaction, object reCognition, and autonomous manipulation capabilities.
However, this Partnership lasted less than a year. In February 2025, Figure AI founder Brett Adcock officially announced the termination of the collaboration to focus on developing its own end-to-end Robotics AI model. The split was primarily due to分歧s in technical direction; Figure believed that General LLMs could not adequately adapt to robotic hardware needs and that a vertically integrated end-to-end model was essential. This breakdown Prompted OpenAI to "resurrect" its robotics team after six years, elevating robotics from an external investment to a core internal strategic business.
A New Growth Curve for the Capital Market
This move also serves to paint a new growth curve for the capital market ahead of OpenAI's IPO. It showcases the company's grand vision of expanding from pure software to a combination of software and hardware, and from the virtual world to the physical world. By telling the story of "embodied AI," OpenAI aims to hedge against market concerns regarding the sustainability of its business model and its massive financial losses.
OpenAI's advantage in entering the robotics sector lies in its world-leading AI large model capabilities, particularly its "World Model" for understanding and simulating the physical world. Its technical path may differ from many companies that start with hardware bodies. Instead, OpenAI is likely to follow a "brain first, body later" logic—using powerful world models to let AI underStand physical lAWS before instilling these capabilities into physical robots. If successful, this approach of DeFining hardware through software and algorithms could reshape the R&D model of the entire robotics industry.
Comments & Questions (0)
No comments yet
Be the first to comment!