Inside the cavernous double-hulls of a commercial ship under construction, welding is among the most punishing jobs in heavy industry: cramped, hot, poorly ventilated, and increasingly hard to staff. That is the workplace Persona AI, a Houston-based humanoid robotics startup founded in 2024, has chosen as its first deployment environment. The company has signed a partnership with HD Korea Shipbuilding & Offshore Engineering (KSOE) and HD Hyundai Robotics to develop and deploy AI-powered humanoid welding robots for shipyards [The Robot Report]. For a pre-seed company barely a year old, landing one of the world's largest shipbuilders as a development partner is the kind of opening chapter most hardware founders spend years trying to write.
Persona AI is building bipedal humanoid robots aimed squarely at heavy industry rather than the warehouse logistics or consumer-home use cases that dominate sector headlines. The company describes its near-term targets as hull welding and repair work in shipbuilding, plus pipe welding, inspection, and maintenance in the energy sector, with confined-space operations cited as a core design constraint [Persona AI]. That focus is a deliberate wedge. Shipyards and energy facilities pay well for skilled welders, face acute labor shortages, and operate in environments where the safety case for sending a machine in first is straightforward. It is a different bet from Figure or Tesla's Optimus, both of which have publicly emphasized general-purpose form factors. Persona is choosing a narrower problem with a clearer buyer.
The macro tailwind is real. Independent forecasts project the humanoid robot market to reach roughly USD 29.57 billion by 2032 at a 39.2 percent CAGR, with industrial labor shortages cited as the primary demand driver [industry market report]. Heavy industry in particular has seen welding workforces age out faster than apprenticeships can replace them, and shipbuilding capacity has become a strategic concern for both Korea and the United States. A humanoid form factor matters here because shipyards and refineries were designed around human bodies climbing through human-sized openings. Retrofitting facilities for fixed automation is often impractical; sending in a bipedal robot that can use existing ladders, scaffolds, and tool interfaces is the more tractable path. That is the case Persona is making to industrial buyers, and the HD Hyundai relationship suggests at least one major customer finds it credible.
The company raised approximately $27 million in an oversubscribed pre-seed round announced in May 2025, with Capital Factory among the backers and Unity Growth Fund LLC also participating [The Robot Report]. Earlier reporting referenced a $10 million figure, which appears to have been a partial disclosure ahead of the final close [InnovationMap]. Pre-seed rounds at that scale are unusual outside of deeptech, but humanoid hardware development carries capital requirements closer to an automotive program than a software seed. The round size signals investor conviction in the team more than in any shipped product.
Pre-seed round (May 2025) | 27 | $M
Earlier disclosed tranche | 10 | $M
That team is the clearest reason to take Persona seriously. Co-founder and CEO Nic Radford previously led humanoid and robotics work at NASA, founded Nauticus Robotics, and worked at Jacobi Motors [IEEE Spectrum]. Co-founder and CTO Jerry Pratt spent decades at the Florida Institute for Human and Machine Cognition, where his team's bipedal work placed second in the DARPA Robotics Challenge finals, one of the most rigorous public tests humanoid systems have faced [IEEE Spectrum]. The early hires reinforce the bench: Michael Patrick Perry, formerly of Boston Dynamics and DJI, has joined as Head of Commercial Strategy, Matt Carney is Chief Engineer, and Jide Akinyode serves as Chief Operating Officer [Persona AI]. For a humanoid program, that combination of NASA systems engineering, competition-tested bipedal control, and Boston Dynamics commercial experience is a meaningful concentration of relevant scar tissue.
The honest counterfactual is that humanoid robotics is among the most capital-intensive and technically unforgiving categories in deeptech, and Persona will compete for talent, supply chain, and customer attention against companies with vastly larger war chests. Figure has raised at multibillion-dollar valuations, Agility Robotics is already deploying its Digit robot in commercial logistics settings, and Tesla continues to pour internal resources into Optimus. Bears will argue that a $27 million pre-seed, while large by venture norms, is a thin runway for a humanoid program that needs to fund actuator development, control software, safety certification, and field trials simultaneously. The bull answer is precisely the focus: by targeting shipyard welding rather than general-purpose manipulation, Persona can constrain the perception, planning, and end-effector problems to a domain where success criteria are measurable (weld quality, cycle time, defect rate) and where a single anchor customer like HD Hyundai can fund iteration through a paid development partnership rather than purely through equity dilution.
There is no peer-reviewed performance data on Persona's robots yet, and the company has not publicly disclosed a timeline for first revenue-generating deployments. What to watch over the next twelve months: a first hardware reveal with specifications, any expansion of the HD KSOE relationship into a commercial purchase order rather than a development agreement, and a likely seed or Series A round to fund the move from prototype to pilot fleet. A second named industrial customer, particularly in energy or defense (both sectors Persona has cited as targets), would be a strong signal that the shipyard wedge is generalizing. The company is also actively hiring, with a robotics software engineering internship currently posted [Persona AI careers].
The standard of care today for confined-space industrial welding is, simply, a human welder in protective gear, often working in shifts limited by heat exposure and air quality, supported by fixed gantry robots only where geometry allows. Quality depends heavily on individual skill, and the pipeline of new welders entering shipbuilding has not kept pace with attrition in either Korea or the United States. Any system that can credibly take on even a fraction of the confined-space hours, with comparable weld quality and an acceptable safety record, would find willing buyers. Whether Persona's humanoids are that system is a question its first field trials will answer. The patient population here is the shipyard workforce, and the outcome that matters is whether fewer humans have to spend their shifts inside a hull. That is the lede worth tracking.
Pulse Raman, Health and Bio Correspondent, Startuply