Amazon Web Services, Nvidia and MassRobotics have named nine startups for the 2026 cohort of the Physical AI Fellowship, highlighting the growing push to develop artificial intelligence systems that operate in real-world environments.
The selected companies are building robotics and AI technologies designed to perform tasks in settings such as farms, factories, retail stores and industrial worksites, according to a Thursday (March 12) news release. The initiative reflects broader interest in what many in the industry describe as “physical AI,” systems that combine robotics, sensors and machine learning to interact directly with the physical world.
While AI has rapidly advanced in software applications such as data analysis and digital automation, applying the technology to machines that must operate in dynamic environments remains more complex. Systems must function reliably under changing conditions, integrate with existing workflows and deliver measurable operational benefits.
Several of the startups in the cohort are focused on agriculture and sustainability. Burro develops autonomous “follow-me” robots designed to assist agricultural crews by transporting tools and materials across farms and outdoor worksites. Terra Robotics is building AI-powered laser weeding systems that remove weeds without chemicals, targeting more sustainable farming practices.
In manufacturing, Deltia uses computer vision and artificial intelligence to analyze manual assembly lines and identify opportunities to improve productivity and product quality. Luminous Robotics focuses on renewable energy infrastructure, developing robotic systems that automate the handling and placement of solar panels during large-scale solar farm construction.
Other companies are focused on the underlying tools needed to build and operate robotics systems. Config develops infrastructure and data systems designed to help organizations deploy and manage robots in real-world environments. Roboto AI offers a data platform that helps robotics teams organize, search and analyze large volumes of multimodal sensor data generated by machines.
The remaining companies in the cohort include Haply Robotics, which develops haptic interfaces designed to improve how humans interact with robots; Telexistence, which builds retail robots capable of automating tasks such as shelf and refrigerator restocking; and WIRobotics, which creates wearable robotic devices designed to support mobility and reduce physical strain in industrial and healthcare environments.
Together, the startups reflect a growing focus on AI systems that do not just analyze data but perform physical work in operational settings.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.