"The winner in the Physical AI era will not be the company that builds the best robot. The companies and countries that easily deploy humanoids and systems to the field for specific purposes will win."
Michael Patrick Perry, Head of Commercial Strategy at Persona AI, made the remarks Thursday in a keynote address at Seoul Forum 2026, which opened at the Shilla Hotel in central Seoul. He pointed to "deployment capability" as the decisive factor that will determine dominance in the future industrial humanoid robot market. Perry is a strategy veteran in the global robotics industry, having served as North America head of drone giant DJI and later as a vice president at Boston Dynamics and Dexterity.
Perry illustrated the importance of field application using cases from the Physical AI companies he has worked at. "DJI was neither the first to make drones nor the company that made the cheapest drones," he said. "Even so, by reflecting the voices of actual filming-site personnel in technology development, its market share surpassed 70%." He continued, "Boston Dynamics also did not succeed because it built the most advanced quadruped robot. Building robotic systems that factory managers could use as soon as they were deployed to the field — such as SDKs (software development kits) and fleet management software — played the most important role in winning customer contracts."
The goal of Persona AI, a humanoid robot developer, is also to internalize field-application capabilities. "Persona AI's goal is to flexibly apply next-generation humanoid robots to heavy industry sites," Perry said. "Building a robot that does everything is impossible. Rather, robots that can perform specific tasks are easier to apply in heavy industries."
Perry warned that humanoid robot developers must be wary of falling into "tech supremacism." "Many companies are emphasizing their technical capabilities by showing robots doing taekwondo or moving boxes, but what matters is the final destination," he said. "The best new-technology strategy is to first define the end use and then build the robot for that purpose." No matter how flashy the movements, he said, they are meaningless without practicality — such as the ability to repeatedly weld hull joints for eight hours in an actual factory.
He singled out the "pilot trap" as the biggest hurdle. "Many robot projects succeed about 40% of the time at the technology demonstration stage involving one or two units, but only 15% are actually commercialized," Perry said. To minimize this side effect, he advised, technology development and deployment timelines must be merged. "What is really difficult is deploying robots at scale in the field," he stressed. "In traditional automation processes, deployment to customer sites comes two to three years after development. Merging that timeline is what gives customers confidence."
The Seoul Economic Daily is hosting Seoul Forum 2026 over two days through Friday under the theme "New Core, New Industry." The forum focuses on how AI is transforming industries including manufacturing, bio and robotics, and what strategies Korea should prepare amid the global AI competition.
Following Thursday's keynote, Su-In Lee, a professor at the University of Washington, will deliver a special lecture Friday on "explainable AI" and AI reliability issues in the medical and bio fields. In a second special lecture, Park Sung-hyun, CEO of Rebellions, will present a vision for sovereign AI infrastructure. Other sessions include the "Robotics Venture Forum," which will discuss investment attraction and global expansion of domestic robot startups, and a special session examining the ethics and social responsibilities arising from AI development.
