Persona AI is betting on ship welding as the first practical commercial use for its humanoid-like robots, seeking to move away from the wave of demonstrations that have yet to prove broad economic viability. The company believes the task does not require a robot capable of doing everything, but rather a machine that uses human tools and performs a specific type of repetitive welding in an environment suffering from a shortage of skilled labor.
Persona has two announced partnerships in this area: the first with HD Hyundai, which the source describes as the largest shipbuilding company in the world, and the second with POSCO, the third-largest steel producer globally. Both partnerships are in Korea, while the company says generally that it is targeting customers capable of supporting the deployment of hundreds of robots at a single site, without disclosing additional details.
From Unskilled Labor to Specialized Skills
When IEEE Spectrum first spoke with Persona in 2024, the company was considering areas such as warehousing, automobiles, and perhaps home use—the same areas targeted by most humanoid-robot companies. But the founders, Nick Radford and Jerry Pratt, concluded that focusing on basic or unskilled human labor did not, in their view, offer a clear economic path.
Instead, the company adopted a thesis focused on specialized professions and tool use. The idea is not that a robot’s ability to perform a task is sufficient on its own; its performance must also be efficient, reliable, and safe, and it must create financial value for both the robot and the customer. Persona therefore chose a task associated with high-cost labor, parts of which robots can perform repeatedly and consistently.
Why Is Welding Suitable for Robots?
The robot will not begin by performing the most difficult types of welding. Instead, it will focus on relatively simple, long linear welds. Radford notes that a single ship requires hundreds of kilometers of these welds, giving the task substantial scale even if its scope is limited.
Pratt explains that tool use is difficult, especially when robots are expected to use the same tools as humans. However, the selected type of welding reduces some of the complexity: the process speed is limited by the rate at which the metal melts, reaching about one centimeter per second, while high-quality welding requires repeated movements at the millimeter level. These precise, continuous movements suit robots and are not affected by fatigue or boredom in the way a human worker might be during long work periods.
Pratt expects the robot to reach a level of performance comparable to humans on these uncomplicated welds, and possibly surpass them later. However, the source presents this as the company’s expectation, not as a commercially proven result so far.
What Changes Practically in the Business Model?
Persona does not primarily present its robot as a means of replacing workers or reducing the wage bill. According to Radford, its current partnerships operate amid substantial accumulated demand and a labor shortage, so the proposed value focuses on increasing customers’ ability to build more ships. Even if the robot is more expensive than a human worker, it may still be useful if it enables a company to produce additional ships and generate revenue it could not otherwise access because of the shortage of skilled labor.
This logic differs from the model of selling a home robot or a broadly capable general-purpose robot, where price competition is greater and it is difficult to justify a high cost before reaching a massive scale of deployment. In shipyards, Persona believes that the potential added value allows it to focus on quality and capabilities rather than sacrificing them to lower the price.
Why Does the Robot Need Two Legs?
Choosing a bipedal robot is not merely an imitation of the human form. Open shipyards may extend hundreds of meters and contain horizontal and vertical beams that must be navigated constantly. Work must also be performed on the ground, above head level, and through openings in ships. For this reason, Persona determined that movement on two feet would be less disruptive to the existing workflow in the yard than a wheeled base, even though wheels are more stable in other conditions.
The nature of the site may also reduce some of the risks of bipedal robots; shipyard workers are trained to work near industrial equipment that may be dangerous. This does not eliminate safety challenges, but it makes the use environment different from a home or places where workers lack experience with comparable industrial equipment.
From Welding to Broader Industrial Skills
Persona views shipbuilding as a starting point, not the end of the product. The adjacent skills mentioned by the company include grinding, painting, and other fabrication work. Its stated ambition is to create the largest repository of industrial skills, but reaching that goal will require tens of thousands of hours of demonstration data provided by experts, high-fidelity simulation, and real-world testing.
The central challenge remains: Persona must prove that choosing a narrow, well-defined task can actually develop into an economically viable robotics company. Focusing on welding may help narrow the scope of the problem, but it does not eliminate challenges shared by home and industrial robots, such as control, locomotion, tool use, and safe operation in unpredictable environments.