Experts argue that physical AI, involving robots and autonomous machines interacting with the real world, may provide a direct path to artificial general intelligence. Elon Musk's comments on Tesla's Optimus robots highlight this potential, amid growing investments in related technologies. The year 2026 is seen as a key inflection point for the field.
Elon Musk recently stated that Tesla’s Optimus robots could one day achieve artificial general intelligence, sparking discussion on the role of physical AI. Physical AI encompasses systems that go beyond content generation to operate in real environments, including robots, autonomous machines, and the foundational models that guide their behavior.
The momentum in physical AI has built over years, with 2026 marking a significant inflection point. At CES in January, Nvidia CEO Jensen Huang declared that the ChatGPT moment for robotics had arrived, pointing to the potential for AI models combined with computing infrastructure to enable large-scale commercial adoption.
Key investments focus on models that help machines interpret and respond to their surroundings. Nvidia has introduced Cosmos and GR00T, open models designed for robot learning and reasoning, along with the Blackwell-powered Jetson T4000 module for industrial edge applications. These tools form a robotics operating system, with partners such as Boston Dynamics, Caterpillar, and LG Electronics already integrating them.
Google integrated its robotics software unit Intrinsic fully in-house from Alphabet earlier this year, creating a vertically integrated stack from foundation models to deployment software and cloud infrastructure.
Deployment data underscores the competition. In 2025, global humanoid robot installations reached approximately 16,000 units, with China accounting for more than 80% in logistics, manufacturing, and automotive sectors. Morgan Stanley data indicates China filed 7,705 humanoid patents over five years, five times the U.S. total, and held 54% of global industrial robot installations. Chinese firms like Unitree shipped roughly 36 times more units last year than U.S. rivals Figure and Tesla combined, benefiting from domestic production of components like motors, sensors, and harmonic reducers.
This scale generates valuable data for improving models, as robots in real environments provide training signals that enhance future iterations. Tesla plans to deploy Optimus robots in its factories to perform tasks and collect data for software refinement.
A Deloitte survey of more than 3,200 global business leaders found that 58% are already using physical AI in some form, with expectations of 80% adoption within two years. This infrastructure buildout supports the thesis that physical AI could underpin broader intelligence advancements.