NextFin News - Amazon Web Services, Nvidia, and MassRobotics have selected nine startups for the 2026 Physical AI Fellowship, a move that signals a decisive shift from digital-only generative models toward "embodied AI" that can manipulate the physical world. The cohort, announced on March 12, 2026, includes Burro, Config, Deltia, Haply Robotics, Luminous Robotics, Roboto AI, Telexistence, Terra Robotics, and WIRobotics. These firms will receive eight weeks of intensive technical support, cloud credits, and direct access to Nvidia’s hardware ecosystem, marking a strategic attempt by the world’s two most valuable compute providers to dominate the next frontier of automation.
The selection of these specific startups reveals a clear industrial thesis: the "brain" of AI is now ready for a "body." While 2024 and 2025 were defined by Large Language Models (LLMs) processing text and images, the 2026 fellowship focuses on startups that translate those digital insights into physical movement. Burro, for instance, focuses on autonomous collaborative robots for agriculture, while Telexistence is scaling remote-controlled and autonomous robotics for retail and logistics. By backing these ventures, U.S. President Trump’s administration sees a pathway toward revitalizing domestic manufacturing and addressing chronic labor shortages in sectors like farming and warehousing through high-tech intervention.
Nvidia’s involvement is particularly telling. The company is no longer content simply selling H100 or B200 chips to data centers; it is positioning its Isaac platform and Jetson modules as the standard operating system for anything that moves. For Nvidia, the Physical AI Fellowship is a funnel for future enterprise customers. If a startup like WIRobotics or Haply Robotics builds its foundation on Nvidia’s simulation tools today, they are locked into that ecosystem as they scale to thousands of units. This is a land grab for the "edge" of the network, where latency matters more than raw throughput.
AWS is playing a complementary but equally aggressive game. By providing the cloud infrastructure and generative AI tools through its GenAIIC (Generative AI Innovation Center), Amazon is ensuring that the massive datasets generated by these robots—petabytes of video and sensor data—stay within the AWS environment. The fellowship offers a glimpse into a future where the distinction between a "software company" and a "robotics company" disappears. Every startup in this cohort is essentially a data company that happens to have wheels, arms, or sensors.
The inclusion of MassRobotics, the Boston-based hub, provides the necessary bridge to the physical world. Robotics is notoriously capital-intensive and prone to "hardware hell," a reality that has claimed many promising startups in the past decade. By providing these nine companies with a direct line to 500+ robotics firms and 30+ corporate sponsors, the fellowship mitigates the traditional risks of scaling hardware. It creates a protected environment where the "Physical AI" can be refined in simulation before being deployed in the messy, unpredictable real world.
This cohort arrives at a moment of intense global competition. As other nations race to automate their industrial bases, the collaboration between the largest U.S. cloud provider and the dominant AI chipmaker represents a formidable "Team USA" approach to industrial policy. The success of these nine startups will likely determine whether the next generation of industrial giants is built on American silicon and software. The focus is no longer on whether AI can think, but on how effectively it can work.
Explore more exclusive insights at nextfin.ai.
