This former PG&E engineer is building a ‘Google Maps for the underground’

The Invisible Infrastructure Problem

Building anything in the physical world is a nightmare of red tape and ancient blueprints. If you are a founder in the software space, you deal with bugs and latency. If you are a founder in construction or utilities, you deal with hitting a gas line because a map from 1974 was off by three feet. It is a massive, expensive bottleneck that slows down everything from broadband rollout to clean energy projects.

A former engineer from PG&E, someone who has actually been in the trenches—literally—is trying to fix this. His startup just pulled in $26 million in Series A funding to build what they are calling a Google Maps for the underground. While the marketing term sounds like typical Silicon Valley fluff, the underlying tech is a pragmatic use of computer vision and AI that builders should actually pay attention to.

The Paper Map Trap

Most people assume that because we have GPS and high-resolution satellite imagery, we know where everything is. We don't. Once you go six inches below the surface, our data becomes incredibly unreliable. Utility companies often rely on paper records, hand-drawn sketches, or "as-built" drawings that were never actually updated after the pipes were laid.

For a founder, this represents a massive data integrity gap. When a construction crew prepares to dig, they have to call in locators who spray-paint lines on the ground based on these shaky records. It is a manual, error-prone process that costs the industry billions in damages and delays every year. This startup isn't just digitizing these maps; they are using AI to ingest various data sources—including ground-penetrating radar and historical records—to create a high-fidelity 3D model of what lies beneath our feet.

Why This Matters for AI Builders

We spend a lot of time talking about LLMs that can write poetry or generate images of cats. But the real value in AI right now is in structured data extraction from messy, legacy environments. This is a prime example of "Vertical AI"—taking a specific, high-stakes industry problem and applying machine learning where it actually moves the needle.

The $26 million investment is a signal that venture capital is shifting back toward hard problems. Investors are looking for teams that understand the friction of the physical world. For builders, the lesson here is simple: stop looking for problems in the digital cloud and start looking for problems in the dirt. The most valuable data sets are often the ones that are currently trapped in a filing cabinet in a municipal basement.

The Technical Challenge: Data Fusion

Mapping the underground isn't just about taking a photo. It requires data fusion. You have to combine electromagnetic sensors, acoustic data, and historical records into a single source of truth. The AI's job is to reconcile conflicting data points. If a paper record says a water main is at ten feet, but a sensor detects something at eight feet, the system has to weigh the probability of which one is correct.

This is the kind of engineering that matters. It’s not just about the algorithm; it’s about the hardware-software loop. The startup is effectively building a digital twin of the subsurface, which is a prerequisite for the autonomous construction equipment we keep hearing about. You can't have a robot excavator if the robot doesn't know where the power lines are.

Cutting the Red Tape

Beyond the tech, there is a massive regulatory play here. The goal of this platform is to reduce the administrative friction that kills projects. If a utility company can prove their map is 99% accurate, they can bypass weeks of manual locating and permitting. This is where the business model becomes defensible. Once you become the standard for safety and compliance, you aren't just a tool; you're the infrastructure itself.

For founders, this is the ultimate goal. You want to build a product that becomes so embedded in the regulatory workflow that it’s impossible to remove. By focusing on the "unsexy" world of utility mapping, this team is positioning themselves at the center of the trillion-dollar infrastructure boom.

The Skeptic's View

As always, we should be skeptical of the timeline. Mapping the entire world's underground is a monumental task that requires cooperation from thousands of fragmented utility providers who are notoriously slow to adopt new tech. Raising $26 million is the easy part; getting a legacy utility company to trust an AI over a guy with a metal detector is the real hurdle.

However, the founder's background at PG&E gives this project more credibility than a typical tech transplant. He knows exactly how broken the system is because he worked inside it. That "insider-turned-disruptor" profile is usually a better bet than a generalist looking for a market to flip.

Founder Takeaway

The takeaway here is that the next generation of massive AI companies won't be building chatbots. They will be building the bridges between legacy physical systems and modern digital twins. If you can find a way to make the invisible visible—whether that's underground pipes, supply chain routes, or energy grids—you are building something with actual staying power.

• Focus on high-stakes data: Small errors in utility mapping lead to explosions and floods. The higher the cost of failure, the higher the value of the solution.
• Vertical AI is the play: Don't build a general tool. Build the definitive tool for a specific, painful industry workflow.
• Integrate with the physical: The most defensible startups are those that solve problems in the real world, not just on a screen.

We will see if this "Google Maps for the underground" can actually scale, but for now, it’s a refreshing break from the endless sea of generative AI wrappers. It’s real tech solving a real, dirty, expensive problem.

Read the original at TechCrunch Startups →