Persona AI is a robotics company developing humanoid robots powered by embodied AI to perform industrial tasks, particularly in labor-constrained sectors like shipbuilding and manufacturing.
Could you tell us about the original idea behind Persona AI and the relationship to NASA and robotics?
Almost 25 to 30 years of my life have been dedicated to humanoid robotics. I was fortunate to start at NASA, where I spent 15 years and eventually became a principal investigator leading several flagship programs, especially around humanoid systems. That work culminated in Robonaut, which became the first humanoid robot in space and was sent to the International Space Station to interact with the crew.
That experience was hugely formative. Seeing a robotic astronaut developed at NASA and then accepted by the astronaut corps shaped how I think about markets and adoption. With humanoids, there is always a technological, economic, and social component, and real transformation only happens when those three align. After 15 years, I decided to leave NASA and pursue entrepreneurship. Persona is now my fifth company, and the past 12 or 13 years have been an utter roller coaster.
Many people assumed AI would affect white-collar work first, while more variable physical work would not be replaced anytime soon. Is that narrative changing?
It is important to think about the adoption curve. The beginning of an S-curve always feels slow, even though progress is happening underneath the surface. Then you reach a tipping point and suddenly everything changes. That is the stage we are in now.
Technologies often look unimpressive for years before they become unavoidable.
Email and self-driving systems are both examples of that. The capability develops steadily, but widespread adoption takes much longer than people initially expect.
I think the same is true here. This will be one of the largest transformations in labor in human history, but it does not need to happen tomorrow for that to be true. We try not to oversell the pace. What matters is that there are industrial sectors under enormous pressure, with severe labor shortages and long-term structural need. Everybody wants to be an influencer or a YouTuber, not a welder. So the necessity for embodied AI in those environments is real, even if the adoption curve is still in its early phase.
Why are you skeptical about domestic use cases, and why focus instead on industrial markets?
I have been very critical of the idea that home use is the first real market. I do not think that is anytime soon. The domestic market does not yet have the purchasing power, the willingness to adopt, the safety framework, or even the business model fully figured out. One day we may well have robots doing dishes, folding clothes, and making beds, but I cannot responsibly say when that will happen.
Industrial markets are different. They are under extreme pressure to produce, and some are facing what is close to an existential labor crisis. In those sectors, a humanoid is not a luxury. A robot trained as a welder is critical. That is why we are focused on areas where the need is immediate, the demand is serious, and the value proposition is clear.
Given that adoption is still early and the work is capital intensive, what is your strategy for staying afloat in the first years?
We are targeting customers that think on very long time horizons. In sectors like shipbuilding, roadmaps stretch across five or 10 years, and companies are managing multibillion-dollar backlogs. What has been encouraging is how mature their thinking is around introducing these technologies responsibly, first to supplement open positions rather than replace an existing workforce. In shipbuilding alone, there are tens of thousands of unfilled welding roles, and those roles are critical.
Our go-to-market strategy is deliberately narrow. We are not starting with a general-purpose humanoid and hoping that consumer demand magically appears. We want to prove reliability, viability, and economics in one focused market first. That is a major difference between Persona and some of the more generalist humanoid companies. We believe you go in, dominate the first market, prove the labor-supplement thesis, and then expand into adjacent ones.
What can we expect from your first commercial launch expected at the end of this year?
We signed an array of agreements and recently entered the next phase of our work with HD Hyundai, the world’s largest shipbuilder, and its shipbuilding arm KSOE. We are on an aggressive 24-month plan to produce three generations of machines, culminating in delivering units to them at the end of this year for commercialization activities in 2027. We are on plan, on target, and things are progressing extremely well.
Even at this pace, it is still nearly a three-year journey from where we started in January 2025. We are not pretending we will suddenly be producing 100,000 machines a year. There are a lot of ambitious plans in the market, but we are taking a more measured and realistic approach.
What will those first humanoids with Hyundai actually do?
I cannot say much more than what we have already put out publicly, because HD Hyundai is protective of certain details. What I can say is that the platform is advancing through multiple generations, Gen 1, Gen 2, and Gen 3, and we will be delivering Gen 3 units to Hyundai at the end of this year. Those machines will then go through the necessary qualification steps before entering operation in the yard.
The intended application is welding. That is the concrete use case. We are also working not only with in-country partners, but with the regulatory bodies that govern this kind of work, because this is not like sending a robot into a home to make a sandwich. Welding is regulated, inspected, and subject to formal standards, so deployment has to be approached accordingly.
If the launch is successful, what comes next? Are you targeting other industries too?
Yes. There is already public information about our relationship with POSCO, the third largest steel producer in the world. Our roadmap focuses on what we call the 4Ds (dull, dirt, dangerous and declining), jobs in difficult environments with extreme ergonomic strain, repetitive tasks, heat, cold, and exposure to the elements. These are exactly the kinds of jobs where the technology can now make a meaningful difference.
I see the humanoid robot, even with its AI brain, as a tool. It is a very capable form of robotic automation that can make real-time decisions in uncertain conditions, but it is still a tool. The question is how humans choose to deploy it and how it works in partnership with the workforce. In that sense, it is an extension of the broader history of automation.
Do you see alignment between Persona AI and the current push for reshoring?
Yes, I do. I like the idea of a more industrialized and self-sufficient America, even while recognizing that we live in a global economy with deeply interconnected suppliers, customers, and technology providers. I think it is good for a country to have strong manufacturing capability, and I absolutely support reshoring certain industries back to the United States.
Shipbuilding is a strong example. It is not just a commercial issue, it is a strategic one. If we want to strengthen those industries again, it is going to happen through robots and automation. I cannot think of a better sector to see grow in the U.S. than shipbuilding.
Where do you picture Persona AI by the end of the decade?
My hope is that Persona AI becomes a beacon for how to build and deploy this kind of high-tech company the right way, with a responsible business plan, responsible uptake, and strong partners. I want the company to be known not just for the technology, but for the discipline behind how it is commercialized.
By the end of the decade, I think this market may start to resemble the early automotive industry, where many players narrowed into a smaller group of leaders. My dream is that Persona will be part of that conversation, one of the companies people see as helping define the category.
