The Macro: The Permit Stack Is the Real Bottleneck in Building Anything
I keep hearing about the AI infrastructure buildout. Billions flowing into data centers. Hundreds of new factories planned under reshoring initiatives. Renewable energy projects stacking up in every state. The money is there. The demand is there. The engineering talent, arguably, is there. What is not there is the ability to actually get permission to build.
Infrastructure permitting in the United States is broken in a way that most people outside the industry do not appreciate. A single data center project can require coordination with 15 to 30 different regulatory agencies. Environmental reviews, zoning approvals, water permits, air quality assessments, endangered species consultations, traffic impact studies, utility interconnection agreements. Each one has its own timeline, its own submission format, its own review process. Miss one filing deadline and the entire project slips by months.
The numbers are staggering. The average large infrastructure project takes 4.5 years from announcement to groundbreaking. Permitting and regulatory compliance account for roughly 40 percent of that timeline. The consulting industry built around navigating this process generates tens of billions in annual revenue. Firms like AECOM, Stantec, and Tetra Tech charge hundreds of dollars per hour to manage permit applications that are fundamentally data entry, document assembly, and deadline tracking problems.
This is the kind of workflow that should have been automated a decade ago. The reason it was not is that every jurisdiction is different, every project type has different requirements, and the regulatory landscape changes constantly. The complexity kept it in the domain of expensive human consultants. Until now, the AI models capable of parsing regulatory code and generating compliant submissions simply did not exist.
The Micro: An Aerospace Engineer Who Got Tired of Watching Rockets Wait for Paperwork
Giuseppe Rapisarda founded Nerviom after spending five years at Rocket Factory Augsburg designing and manufacturing composite components for UAVs, rockets, and satellites. That is a background where you learn, viscerally, what it feels like to have a fully engineered product sitting on a shelf because someone has not filed the right paperwork with the right agency in the right format by the right deadline. Rockets do not wait well. Neither do data centers burning through capital before they generate a single watt of compute.
Nerviom came through Y Combinator’s Fall 2025 batch. The product is an AI autopilot for the entire permitting lifecycle. You feed it your project parameters, location, facility type, capacity, environmental footprint, and it maps the full regulatory landscape. Every permit you need, every agency you need to file with, every deadline you need to hit, every document you need to prepare. Then it starts generating the actual submissions.
The platform targets the three hottest segments in infrastructure development right now: data centers, factories, and energy projects. These are not small markets. Data center construction alone is projected to exceed $100 billion globally by 2028. The Inflation Reduction Act and CHIPS Act have triggered a manufacturing construction boom not seen in decades. And renewable energy projects, solar, wind, battery storage, are constrained more by permitting than by technology or financing.
What makes Nerviom interesting to me is the scope of ambition. This is not a tool that helps you fill out one form faster. It is an attempt to model the entire regulatory graph for infrastructure development and automate navigation through it. That means understanding federal, state, and local requirements simultaneously. It means knowing which permits can be filed in parallel and which are sequential. It means tracking regulatory changes in real time and adjusting project timelines accordingly.
The competitive landscape is thin. Enstoa and nPlan do project management for construction but not permitting automation specifically. Permitio and PermitFlow target residential and commercial building permits, which is a fundamentally different problem from infrastructure-scale regulatory compliance. Palantir has contracts with government agencies but works the other side of the desk. Nerviom is going after the gap between generic project management tools and the armies of consultants who currently manage this process manually.
The Verdict
I think Nerviom is attacking one of the genuinely under-automated categories in the economy. Permitting is a multi-billion dollar pain point with no dominant software solution, and the current process is so slow that it is actively constraining industries with nearly unlimited demand. The timing is excellent. Every major infrastructure trend right now, AI compute, reshoring, clean energy, runs directly into the permitting wall.
The risk is execution complexity. Regulatory compliance is not a domain where “mostly right” works. A submission that is 95 percent correct gets rejected the same as one that is 50 percent correct. The AI needs to be precise across thousands of jurisdictions with different requirements, and the cost of errors is measured in months of project delays and millions of dollars in carrying costs. Giuseppe’s aerospace engineering background suggests an appreciation for precision, but this is a hard problem even for the most rigorous team.
At 30 days, I want to see a case study. One real project where Nerviom cut permitting time by a measurable amount. At 60 days, the question is whether infrastructure developers trust AI-generated regulatory submissions enough to file them. At 90 days, I want to know if the platform can handle multi-jurisdiction complexity, because the biggest projects, the ones with the biggest budgets, span multiple states and federal agencies simultaneously. If Nerviom can prove reliability at that scale, this becomes a category-defining company. The infrastructure boom is not waiting. The permits are.