Making the Invisible Visible

For the last thirty years, venture capital has largely been shaped by a software worldview. The model was simple enough: write a check, help a company scale fast, chase the hockey stick, and win through speed. That framework made sense in a world where software was supposed to eat everything. But that is not the world we are entering now. What's emerging instead is a much more complex and consequential arena: the infusion of intelligence into the physical world. Manufacturing. Healthcare. Energy. Robotics. Supply chains. Industrial systems. Real infrastructure. This is where some of the biggest opportunities now sit. And this is also where the traditional venture model starts to show its limits. Physical innovation does not move at software speed. It takes longer. It costs more. It requires coordination across far more variables. You are not just moving ones and zeros around. You are working through factories, deployment, regulation, hardware constraints, research institutions, supply chains, and real-world adoption. That journey is harder, lonelier, and far less forgiving than the old venture playbook ever accounted for. That is why what Eclipse is doing deserves attention. The headline is easy to misunderstand. Another billion dollars raised for physical AI is interesting, but that is not the real signal. The real signal is that a massive chunk of that capital is being allocated not just to investing, but to building. It suggests that even major institutions now recognize that writing checks is not enough. If you want to transform the physical world, you need a model that is more operational, more collaborative, and more deeply involved in execution. Eclipse is not alone in making this bet. SOSV, through its HAX program, has built one of the most active hardware-focused investment operations in the world, running founders through a hands-on development process inside a fully equipped facility with machine tools, engineering labs, and on-staff technical experts. Monozukuri Ventures, whose name draws from the Japanese concept of "the art of making things," pairs early investment with prototyping know-how and manufacturing expertise rather than treating those as the founder's problem to solve alone. These are not incubators in the traditional sense. They are something closer to operating partners with capital attached. The category is still being named. Venture. Studio. Incubator. Accelerator. Fund. Holding company. Operator network. None of those labels fully captures what is emerging. The point is not the label. The point is the mechanism. The mechanism reflects a simple reality: physical innovation requires a different kind of institutional support than software ever did. In software, a small team can sometimes brute-force its way from idea to product. In physical tech, that is much harder. The path demands what might be called a fellowship model: the right operators, the right institutions, the right supply-chain experience, the right technical and commercial guidance, brought together at the right time. It takes collaboration, not just capital. That first stretch, between a validated idea and a product actually deployed in the world, is where most of the real work happens. Everyone loves the breakthrough moment. The actual journey from "I have a vision" to "this is running in the real world and changing an industry" is rarely a solo act, and it is almost never funded well by a check and a board seat. This is beginning to reshape how the investment landscape is organized. At the top, you have mega-institutional ambitions, the JP Morgan and Bezos-scale bets on reindustrialization and physical infrastructure. Then you have firms like Eclipse, moving beyond traditional venture toward a more operator-infused model aimed at large industrial problems. There are earlier-stage players as well, working from the ground up: more 0-to-1 in orientation, more translational, more directly focused on helping researchers and founders navigate that brutal first stretch. At Conduit, that is where we have focused our work. The thesis across all of these efforts is not that founders need less ambition. They need more. They need the willingness to pursue big, important, disruptive problems. But that ambition must be paired with operational discipline and a support structure that understands what physical-world execution actually demands. The people building at the intersection of machine intelligence and physical reality need more than encouragement and a term sheet. They need access to operators who have done it before, to factories and labs and institutional relationships, to manufacturing insight and strategic guidance that make hard things possible. There are now firms organized specifically to provide that, catering to specific verticals: robotics, energy, defense, medtech, industrial automation. The model is maturing. The old venture model was built for software. The next innovation stack will be built for the physical world. And not by capital alone. About the Author: Amish Patel leads Conduit Venture Labs, a venture studio dedicated to building and backing companies that merge software intelligence with the physical world. By leveraging expertise in operational execution, product design, and venture capital, the studio transforms innovative technologies into impactful products and businesses. Amish's background includes leadership roles in product and design within consumer and medical technology, with a passion for blending design, behavior science, and engineering to build trusted, widely adopted solutions.