Is Shipyard Welding the Right First Job for Humanoid Robots?

We have a problem in the robotics industry: everyone is building toys while waiting for a miracle. If you follow the space, you’ve seen the slick YouTube demos of bipedal robots dancing, backflipping, or struggling to fold a T-shirt. It’s impressive engineering, but as a founder, I have to ask: where is the business? Most humanoid startups are chasing "general purpose" utility, which is a polite way of saying they don't know what their product is for yet.

Persona AI is taking a different path. Instead of trying to build a robotic butler that might be ready in 2035, they are focusing on one of the oldest, grittiest industrial tasks there is: welding. Specifically, shipyard welding. It’s not glamorous, it’s not "viral" content, but it might be the smartest move I’ve seen in the humanoid space in years.

The Trap of Unskilled Labor

Most of the big players in robotics are obsessed with replacing unskilled labor. They want robots in warehouses moving boxes or in homes picking up toys. The logic is that these tasks are simple, so they should be easy to automate. But that logic is flawed because the economics don't work. When you compete with unskilled labor, you are competing on price in a race to the bottom. If your robot costs $100,000 to build, but it's replacing a job that pays $15 an hour, the ROI takes years to materialize. Most businesses won't touch that.

Nicolaus Radford and Jerry Pratt, the founders of Persona AI, realized this early on. Radford, who led NASA’s Valkyrie program, and Pratt, a veteran of Figure and IHMC, initially looked at the same broad markets as everyone else. But they pivoted toward skilled trades. By focusing on a high-value skill like welding, they aren't competing with cheap labor; they are solving a labor shortage in a high-margin industry.

Why Shipyards Matter for Builders

Shipyards are a nightmare for traditional automation. You can't just put a robot on wheels and expect it to work. These sites are hundreds of meters long, filled with obstacles, vertical spars, and tight portholes. This is where the humanoid form factor actually makes sense. We don't need legs so the robot looks like us; we need legs because the environment was built for us. If you want to move through a ship under construction, you have to be able to step over things.

For founders, there is a massive lesson here in "last-mover advantage." Persona watched the industry struggle with the "general purpose" myth and decided to go deep on a specific tool-based niche. They aren't trying to invent a new way to weld; they are building a robot that uses the same handheld tools humans use. This lowers the barrier to entry for the customer and makes the robot a drop-in replacement for a missing worker.

The Physics of Profit

Welding is a uniquely "robot-friendly" skilled trade for three reasons:

• Speed Limits: You can only weld as fast as metal melts. This means the robot doesn't need to move at superhuman speeds. A steady centimeter per second is enough.
• Precision: High-quality welding requires millimeter-scale consistency over long periods. Humans get tired, their hands shake, and they get bored. A robot doesn't.
• Volume: Modern shipbuilding requires hundreds of kilometers of linear welds. It is repetitive, grueling work that is perfectly suited for a machine that never needs a break.

Persona is currently working with HD Hyundai and POSCO. These aren't small pilot programs; these are massive industrial players who are labor-constrained. When a company has a backlog of ships they can't build because they don't have enough welders, they aren't looking to cut costs—they are looking to increase their top-line revenue. If a robot allows them to finish a ship faster, the cost of the robot becomes almost irrelevant.

A Reality Check on General Purpose AI

We need to be honest about the state of AI in robotics. The industry talks about "foundation models for movement" as if we are months away from a robot that can learn any task just by watching a video. The reality is much slower. Persona is spending thousands of hours collecting expert demonstration data just to get welding right. They are building high-fidelity simulations to test specific motions.

This is the work that actually moves the needle. As a founder or an investor, you should be skeptical of any company claiming they can do everything at once. Complexity is the enemy of reliability. By mastering one tool—the welder—Persona is building a "beachhead." Once they own the welding niche, they can expand to grinding, painting, and other fabrication tasks. But they are starting with a winnable fight.

The Founder Perspective: Why Focus Wins

Radford mentioned that people have told him he isn't "thinking big enough." In the world of VC-backed tech, that's often code for "you aren't promising me a trillion-dollar fantasy." But thinking big without a starting point is just dreaming. Persona’s ambition is to become the largest repository of industrial skills. That is a massive vision, but it starts with a single, functional application.

The takeaway for builders in the AI and robotics space is clear: find the environment where your form factor is a necessity, not a novelty. If a wheeled robot can do the job, use wheels. If a stationary arm can do it, use an arm. If the environment is so chaotic and human-centric that only legs will work, then you have a use case for a humanoid.

The value added by a robot in a skilled trade is high enough that you don't have to cut corners on quality just to compete on price.

We are entering a phase where the "cool factor" of AI is wearing off, and the "utility factor" is being scrutinized. The companies that survive won't be the ones with the best dance videos; they will be the ones whose robots are covered in soot, sparks, and grease, actually building the infrastructure of the world.

The Bottom Line

Persona AI still has to prove they can scale. They have the partnerships and the right technical pedigree, but moving from a controlled test to a functional shipyard is a brutal transition. However, their strategy is sound. They aren't waiting for the world to change; they are building a tool for the world as it currently exists. For any founder looking at the intersection of AI and hardware, that is the only blueprint that matters.

Read the original at IEEE Robotics →