Physical Intelligence Secures $400 Million Investment from Jeff Bezos and OpenAI to Revolutionize Robotic AI

Physical Intelligence, an AI startup focused on creating adaptable software for many robots, recently announced a significant funding milestone. The startup raised $400 million in early-stage financing, with prominent backers including Amazon’s Jeff Bezos, OpenAI, Thrive Capital, and Lux Capital. The investment round, valuing Physical Intelligence at approximately $2 billion, marks a substantial leap from the $70 million raised during its seed stage earlier this year.

The primary aim of Physical Intelligence is to develop versatile, foundational software that can operate on any robot, shifting away from the conventional approach of building specific software for individual robots and tasks. This “generalist” software, as co-founder and CEO Karol Hausman describes it, acts as a universal brain capable of controlling multiple robotic types. “It’s not just a brain for any particular robot,” Hausman noted. “It’s a single generalist brain that can control any robot.”

Creating such a comprehensive AI model is no small feat. It requires extensive real-world data, much of which must be made. Consequently, Physical Intelligence has been building its own data sets, leveraging advancements in AI models that interpret visual data. In a recent publication, the company demonstrated its software—π0 or pi-zero—enabling robots to perform diverse tasks such as folding laundry, clearing tables, flattening boxes, bagging groceries, and even retrieving toast from a toaster. This wide-ranging functionality has led the company to describe its software as a “true generalist.”

Physical Intelligence’s founding team is led by Karol Hausman, a former robotics scientist at Google; Sergey Levine, a professor at UC Berkeley; and Lachy Groom, a former Stripe executive. According to Groom, the evolution of the company’s technology might mirror the trajectory of early chatbot models. He compared pi-zero to OpenAI’s GPT-1, suggesting that while it is still in its early stages, future breakthroughs could dramatically advance the field.

With the robotics AI sector becoming increasingly competitive, Physical Intelligence faces formidable peers. Companies like Skild and Figure AI—backed by Bezos and OpenAI—are developing general-purpose robotics software, while Covariant and other companies focus on specific applications for industrial robots. The recent surge in investment in AI and cloud technology is notable across the U.S., Europe, and Israel, with Accel projecting a total investment of $79.2 billion by the end of 2024.

Amazon’s involvement in robotics has intensified as it integrates more automation into its operations to cut costs and expedite deliveries. Meanwhile, Tesla’s CEO Elon Musk recently revealed ambitious projections, forecasting at least 10 billion humanoid robots in production by 2040, each priced between $20,000 and $25,000. At Tesla’s Robotaxi unveiling, the latest version of Tesla’s Optimus humanoid robot was showcased, underscoring the push toward scalable robotic solutions.

The future holds promise and challenges for Physical Intelligence as it strives to pioneer an AI system that functions universally across different robotic platforms. While pi-zero is in its early developmental phase, the rapid advancements in AI may bring breakthroughs sooner than expected, potentially redefining the capabilities of robots worldwide.