01 June 2026 | Interaction | By Editor Robotics Business NEWS <editor@rbnpress.com>
As demand for AI infrastructure, renewable energy, and large-scale power projects accelerates, construction companies are under pressure to deliver more with limited labor resources. Xpanner has emerged as one of the fastest-growing Physical AI companies in the sector, helping contractors automate critical construction operations through a software-first approach. In this interview with Robotics Business News, Ryan Park, Co-founder, CFO and CSO of Xpanner, shares insights on the company's rapid U.S. growth, its transition to Automation-as-a-Service, the rise of Software-Defined Machinery, and why Physical AI is becoming a critical foundation for the future of infrastructure development.
Q1. Xpanner has achieved rare traction in construction Physical AI with profitability and zero churn. What were the biggest operational or cultural decisions that helped the company scale so quickly in the U.S. market?
The core principle that shapes most of what we do at Xpanner compresses into four words: Making the useful usable. The aim is to take genuinely advanced technology and turn it into products that work on real job sites without requiring customers to learn new systems, restructure their workflows, or absorb disruption. That philosophy has guided our product decisions and our hiring from the beginning.
The first major operational decision was Forward-Deployed Engineering. Our R&D does not operate from a lab at a distance from the customer. Engineers are placed directly on active sites, absorbing the variables that simulations rarely capture and feeding those observations back into the product roadmap. The gap between an issue surfacing in the field and a fix reaching the product tends to be measured in days.
The second decision was about the profile of people we built the team around. Many of our engineers come from global companies (Volvo, Hexagon Group, Boeing) with product development backgrounds shaped by high-reliability, high-precision industries. That background raised Xpanner's internal bar for reliability and usability meaningfully. Building the underlying technical infrastructure took time. But the quality of what we created, has generally held up well against what the market expected.
That reliability has been the most direct driver of field adoption. When a product works on day one without disruption and continues to improve on the same site over time, customers stay. In construction, where schedule pressure is constant and any friction compounds quickly, that matters more than almost anything else.
We think of ourselves as a product-driven company; one where the completeness, usability, and day-to-day impact of the product is treated as the highest priority. That focus, combined with the significant tailwind of AI infrastructure demand creating unprecedented construction activity, gave us conditions to grow more quickly than we had anticipated.
Q2. The company's Automation-as-a-Service (AaaS) model shifts construction automation from a capital expense to a subscription model. How are customers responding to this approach, and what advantages does it create for Xpanner long term?
Customer response has been encouraging. Some enterprise customers were, in fact, waiting for this shift before formally committing. The move to Automation-as-a-Service (AaaS) was not driven by a single rationale. There are two distinct reasons working together, and both shaped the decision.
The first is lowering the adoption hurdle. Upfront capital requirements are removed, multi-year bets on a single technology are no longer required, and the overhead of managing fixed assets is reduced. The field engineering and operational support that customers rely on stays in place. The goal was to reduce financial friction without changing what customers depend on most.
The second rationale is less frequently discussed but equally important. Xpanner operates in a space where the underlying hardware is also evolving - sensors improve, compute components advance, and kits will be upgraded in hardware, not just software. In a traditional purchase model, customers who bought an earlier version face the burden of repurchasing to access a meaningfully better product. That burden can slow both early adoption and continued investment. The subscription model addresses this more naturally. Customers access updated versions of the product as part of their subscription. Xpanner can evolve the hardware on a faster cycle without creating disruption or financial pressure for the customer. And customers have less reason to resist the change, which matters for a company trying to improve the technology continuously.
The longer-term structural benefit of moving from one-off hardware sales to an ARR-based model is the predictability and stability it creates. The more deeply customers integrate X1 Kit into their workflows, the higher the cost of switching becomes over time, not because we make switching difficult, but because the product becomes genuinely embedded in how they operate. That is a kind of durability that hardware sales alone tend not to create. Subscription is not a pricing adjustment. It is a different way of designing how the business works and how customers continue to benefit from it.
Q3. Xpanner's Software-Defined Machinery (SDM) philosophy focuses on upgrading existing equipment through software rather than replacing hardware. How disruptive do you believe this model could become for the broader construction industry?
SDM challenges a long-standing assumption in the industry: that the capability of a construction machine is largely fixed at the point of manufacture. We believe that assumption is increasingly open to question, and that the implications of revisiting it are significant; though how that plays out at industry scale is still being understood.
One concrete illustration of what SDM can mean in practice: equipment purchased in the 1990s or early 2000s, with the right kit and software applied, can perform comparably to more recent machines across many common operations. The machine itself does not need to change. What changes is the intelligence running on it. For customers managing aging fleets or renting equipment across multiple projects, that shifts the capital calculus of automation considerably.
The first-order impact shows up in asset utilization. A large portion of construction equipment is rented or hired for specific project types. Under a traditional model, a machine configured for solar piling is largely limited to that role. With SDM, the same excavator can move from piling to BESS trenching to grading on a data center site within a single season. Idle time drops, the range of deployable applications expands, and the economics of the fleet improve for the customer.
The second dimension involves the relationship between Xpanner and equipment manufacturers. Many leading manufacturers are already developing or actively exploring their own software-defined capabilities — this is increasingly where the industry is headed. What we see as an interesting opportunity is the possibility of working alongside those manufacturers to help complete and accelerate that journey together. For a manufacturer that integrates Xpanner's AaaS infrastructure (HW/SW) alongside their own product development, the collaboration opens a path to offering customers an entirely new kind of service relationship; one where the machine becomes a platform for continuous value delivery, not just a capital purchase. Revenue from that ongoing service can be structured as a shared model, giving manufacturers a meaningful new dimension to their existing business. Rather than replacing what manufacturers are building, the aim is to find where our capabilities genuinely complement theirs and where a joint model creates something neither could offer independently. We are already in active discussions with select manufacturers along these lines, and similar conversations are underway with major rental and fleet companies, who see a clear fit between our model and the utilization economics of their business.
Rental and fleet companies represent another natural fit. Their entire business model hinges on utilization, and SDM lets a single machine shift across solar piling, BESS trenching, and data center earthwork as project mixes change. Because our solution augments value across the broader equipment chain, we have already established meaningful engagement with rental companies in this space, and active conversations on deeper collaboration continue to develop.
As AI capabilities advance, the frontier of construction automation will likely move from controlling individual machine operations to coordinating multi-machine workflows and eventually completing full site sequences with reduced human direction. For AI to work at that level effectively, the equipment it directs needs to be fully software-addressable. The SDM infrastructure we are working toward is intended to reach a point where that kind of AI-to-machine interaction is natural and low-friction. We think that is where the field is heading, and we want to be a useful part of building that foundation.
Q4. Labor shortages continue to impact large-scale infrastructure projects globally. How is Xpanner positioning its Physical AI platform as a direct solution to the workforce challenges facing solar, BESS, and AI data center construction?
Xpanner started in solar, deploying X1 Kit first into pile driving and panel lifting. These are two operations with clear bottlenecks and significant labor dependency. By demonstrating measurable improvement across workforce relief, cost, and safety simultaneously, we built credibility with major EPC customers. That credibility is now enabling expansion into adjacent verticals.
This year, we are bringing a meaningful number of new task capabilities to market beyond solar. The X1 kit and software license model has been validated in the field, and we are using that foundation to extend into data centers, gas turbine power plants, BESS, and broader infrastructure construction. The aim by late 2027 (when our next-generation kit is planned), is a substantially wider portfolio of validated use cases across these verticals.
When automation reduces the workforce requirement on a specific operation, that does not translate to fewer people in construction overall. The most significant bottleneck to delivering AI-era infrastructure at the required pace is the available workforce. There are simply not enough people to execute the volume of work the market is creating. What we are working toward is making it possible for the same number of workers to accomplish considerably more - not by cutting headcount, but by extending the reach of each person on site. Demand for skilled construction professionals will grow. The question is how much more each of them can accomplish with better tools.
Looking ahead, we expect unmanned processes to expand meaningfully alongside supervised automation. The broader direction we are moving toward is what we think of as an infrastructure transformation, not just building what people need, but building, at greater speed and lower cost, the infrastructure that AI itself requires to operate: power generation, data centers, transmission networks.
Q5. Since more than 90% of Xpanner's revenue now comes from the U.S., what lessons have you learned from entering and scaling in one of the world's most competitive construction markets?
The most important lesson was understanding the people who work in construction.
Before we entered this market, construction was often described as slow-moving, resistant to technology, old-fashioned. That framing has some truth to it, but it also misses something important. What we found was that the workers on these sites (ie. project managers, operators, site engineers) are not, in any meaningful sense, behind. Most are technically fluent, comfortable with digital tools in their daily lives. They use the same apps, the same ride-sharing, the same consumer technology as anyone else. They are not a different category of person. They are doing a different kind of job.
The reason construction has appeared slow to adopt new technology is less about the people in it and more about the tools they have been offered and the conditions under which they work. The average U.S. construction professional is managing a substantially heavier workload today than counterparts did a decade or two ago, with tools that have improved more slowly than the demands placed on them. There is limited time to evaluate a new product, run a careful pilot, absorb failures, and return a considered verdict.
This insight changed how we approach product development. A solution that requires weeks of adjustment before it delivers value is unlikely to be adopted under real site conditions; not because the workers are unwilling, but because the cost of that transition period is genuinely too high given everything else they are managing.
What we are trying to build is less a productivity multiplier and more a restoration of capacity. Site managers of an earlier era were effective in part because they had time, information, and organizational support, to make well-considered decisions. This has been squeezed considerably out of the modern construction role. The aspiration for Xpanner is to give some of that back: enabling people with fewer resources and a smaller required skillset to perform at a level that their most experienced predecessors achieved.
The lesson the U.S. market has reinforced, above all others, is that what earns trust here is performance on a live site - not a demo, not a controlled pilot, but real results against a real EPC schedule. That is the standard we have tried to hold ourselves to from the beginning.
Q6. Xpanner is already working with major EPCs such as Mortenson and QCells. What has feedback from these enterprise customers revealed about the future demand for autonomous construction solutions?
[Note for editorial: At this time, please do not reference Black & Veatch by name in published responses. Their formal approval has not yet been obtained for public mention.]
The clearest signal from enterprise customers is a shift in the nature of the conversation. A year ago, the typical question was whether to explore automation. Today, the question is more often how quickly we can deploy on a specific site. That shift says a great deal about where the market has moved.
Across Mortenson, QCells, and the broader set of top-tier EPCs we work with, these organizations are already in the middle of strategic transformation - reviewing business models, reallocating resources, building new internal capabilities, and rethinking investment priorities. The question inside these companies is no longer whether automation matters. It is how to reorganize around it effectively. The consistent bottleneck in those conversations is workforce: the right skills, in sufficient numbers, with the cohesion to execute at the scale these projects demand.
The demographic picture adds a layer of urgency that many of these companies are already planning around. A significant portion of the most experienced U.S. construction workforce, which is the senior cohort carrying decades of institutional knowledge and judgment, is on a retirement timeline that concentrates after 2030. Organizations mapping their five-year plans are treating that transition as a meaningful operational risk, and many are already building toward automation as part of how they manage it.
The volume of demand being generated by the AI era reinforces this. Data centers, power generation, transmission infrastructure, substations - capital flows in the current economy are converging into construction projects that need to be delivered faster than the existing labor model can support. The companies we work with have largely moved past debating whether autonomous construction is necessary. The more active discussion is about which capabilities are available now, and how to begin integrating them into ongoing projects.
Q7. The new funding round comes at a time of explosive growth in AI infrastructure and renewable energy construction. Beyond solar piling and material handling, which new applications or verticals do you see as the next major growth opportunities for Xpanner?
The verticals we are focused on share a common characteristic: demand driven by the AI era is growing faster than construction capacity can currently absorb. Physical AI is one of the more realistic paths toward closing that gap, and that is where we are directing our energy.
The area where we are moving most actively right now is gas turbine power plants. When large-scale energy complexes or data center campuses are being designed, solar and renewables alone often cannot meet the consistency and reliability requirements for mission-critical infrastructure. Gas generation has been filling that role, and it is seeing renewed investment. Our first project in this vertical is planned for 2026, and we are developing the specific automation capabilities that earthwork and foundation operations at these sites require.
Data centers represent the largest concentration of growth opportunity for our next phase. The capital committed to AI infrastructure has moved faster than most projections anticipated, and the construction demand that follows is stretching the industry's existing capacity. For Xpanner, what matters is that the operations most critical to data center construction timelines (like earthwork, grading, foundation work), are where our technology has already been tested and validated. We are also beginning to look carefully at automation inside data center facilities, not only on the surrounding site work.
BESS shares many of the same physical construction characteristics as solar: large sites, repetitive earthwork, demanding schedules, and the same workforce constraints. It is the vertical where our existing capabilities transfer most directly. Beyond that, we are looking at power distribution infrastructure and, at a broader level, at road and harbor construction as areas where targeted testing can help us understand how our existing technology applies.
We identify the most pressing operation within a new vertical, build and validate a solution for that specific operation in the field, and establish credibility there before expanding further.
Q8. Investors highlighted Xpanner's ability to 'de-risk the hardest part' by proving profitability and enterprise-scale adoption early. As you look ahead, what milestones will define the company's next phase of growth and global expansion?
The milestones defining our next phase fall along four axes, and the first is the one that shapes what is possible across the others.
The most foundational work underway is restructuring how the business scales operationally. Deploying Physical AI at meaningful scale is capital-intensive. Field operations, kit inventory, service infrastructure, and logistics all require ongoing investment. To grow without being constrained by that capital weight, we are developing a model that brings in dedicated capital partners who participate in the economics of deployment. Under this structure, Xpanner does not carry the full capital risk of scaling alone. Partners provide deployment capital in exchange for a share of the operating economics, while Xpanner retains the technology, the data, and the direct customer relationship. If this model comes together as we intend, the result is a business that can scale with significantly less capital drag. That is the milestone we believe unlocks most of what follows.
The second is building a deeper and more durable data foundation. Every site we operate on generates Physical AI training data - machine behavior, soil conditions, operational variables. As deployment grows, so does the quality of data informing our models. We are working to accelerate this in two ways: by extending lightweight sensor and data capabilities to partner-operated equipment beyond our own direct deployments, and by developing a proprietary computing device that consolidates the processing components we currently source externally. When that device is ready, the quality and volume of data captured per kit should improve meaningfully, and the downstream benefit to model performance should compound over time.
The third milestone is completing the transition to subscription-based revenue. We are targeting 100% AaaS within this year. Moving to a recurring revenue structure creates the predictability, margin stability, and unit economic clarity needed to support the next stage of growth and ensures that every deployed kit is continuously contributing both revenue and data.
The fourth is geographic and vertical expansion. Within the U.S., the near-term focus is extending from solar into BESS, data centers, gas turbine power plants, and broader infrastructure. Geographically, the markets we are considering next, after the U.S. and Korea, are Australia, Europe, and Japan. The pace of that expansion will be governed by our ability to maintain the execution standard we have built in the U.S. Coverage for its own sake is not the goal. The goal is to build a company that earns a position as a trusted part of how Physical AI gets deployed in global construction infrastructure - one validated market and one validated vertical at a time.
The recent Series B bridge round was structured precisely to make this next phase executable. The early signals since closing have already validated that the playbook scales, and faster than we modeled. A more substantial Series C is now in active preparation. For this next round, we are also actively exploring partnerships with strategic investors who bring operational depth in the industry, alongside financial capital.
