Iron Grid Is Betting That Hardware Insurance Is Broken, and Physics Can Fix It

The Macro: Nobody Knows How to Insure a Robot

The insurance industry is old. Unbelievably old. And the way it prices risk has not fundamentally changed in over a century. You take a population of similar things. You count how many of them break. You build a table. You price policies against the table. This works beautifully when the thing you are insuring has been around long enough to generate reliable failure data. Cars, buildings, cargo ships. The actuarial math is solid because the sample sizes are enormous and the failure modes are well understood.

Now try to insure a grid-scale battery. Or a hydrogen production system. Or an autonomous warehouse robot. These are frontier hardware categories. They have been deployed in meaningful numbers for maybe five years. In some cases, less. The failure data is thin. The degradation curves vary wildly by manufacturer, use case, and environment. Traditional insurers are either pricing these policies with huge uncertainty margins, which makes coverage prohibitively expensive, or they are declining to write them at all.

This is not a small problem. The energy storage market alone is projected to hit $400 billion by 2030. Every major battery deployment needs insurance. Every industrial robotics installation needs coverage for hardware failure. The gap between what frontier hardware operators need and what traditional insurance will offer is widening every year.

The incumbents have noticed, but their response has been slow. Hartford, Zurich, and AIG have all started writing specialty policies for energy storage, but they are still pricing off limited historical data. Specialty MGA players like Kingsbridge and Energetic Insurance are closer to the problem but still rely on actuarial approaches that were designed for asset classes with decades of claims history. Nobody has built insurance pricing from physics up.

The Micro: Battery Scientists Who Decided to Sell Insurance

Iron Grid was founded by Fern Morrison and Gabriele Pozzato, and their backgrounds explain the entire thesis. Fern was on Apple’s battery team and then employee number nine at Mitra Chem, a cathode materials startup. Gabriele was the modeling lead at Form Energy, the iron-air battery company, and a Stanford postdoc researcher. Both have Stanford engineering credentials. These are not insurance people who learned about batteries. These are battery scientists who realized the insurance industry was doing the math wrong.

Their approach is to replace actuarial tables with physics-informed machine learning models. Instead of counting failures across a population, Iron Grid builds degradation models for specific hardware using live telemetry data. A battery installation sends operational data to Iron Grid’s platform. The models predict how that specific system will degrade under its specific operating conditions. The insurance pricing follows the physics, not the historical averages.

They came through Y Combinator’s Summer 2025 batch as a two-person team in San Francisco. Since then, they have raised $4 million in seed funding and expanded to hold P&C and Surplus Lines licenses across multiple states. They are underwriting directly, which means they are cutting out the traditional MGA-to-carrier-to-reinsurer chain that typically eats 40% of premiums in intermediary overhead.

The product scope is expanding beyond batteries. Their current pipeline includes hydrogen production and transport, smart inverters, industrial robotics, HVAC systems, autonomous systems, and power electronics. Basically anything where the hardware is new enough that the insurance industry does not have reliable failure tables and complex enough that physics-based modeling would outperform population statistics.

They are actively hiring underwriters for manufacturing and automation at salaries between $115K and $185K. That hiring profile tells you something about their ambitions. They are not just building software. They are building a full-stack insurance operation.

The Verdict

I think Iron Grid has one of the cleanest founder-market fits I have seen this year. Two people who spent their careers modeling battery degradation at the highest level decided the insurance industry was mispricing their domain of expertise. That is the kind of insight that produces real companies.

The direct underwriting model is aggressive but smart. Insurance distribution chains are bloated. If your risk models are genuinely better than what traditional carriers use, there is no reason to give 40% of every dollar to intermediaries who are adding uncertainty, not removing it.

In 30 days I want to see loss ratio projections. Physics-informed models sound great in a pitch, but insurance is a business where being wrong about risk by 5% can destroy your margins. The question is whether their models are materially better than actuarial baselines on real deployments.

In 60 days the question is distribution. Selling insurance to energy developers and robotics operators requires specialized sales relationships. The product can be great and still stall if the go-to-market is too slow for the capital intensity of an insurance business.

In 90 days I want to understand reinsurance. Direct underwriting means Iron Grid is taking balance sheet risk. At some scale, they will need reinsurance partners, and the terms they get from reinsurers will determine the economics of the whole business. If reinsurers trust the physics models, Iron Grid wins. If reinsurers insist on traditional actuarial overlays, the cost advantage shrinks.

The founding team, the technical approach, and the market timing all line up. Frontier hardware is being deployed faster than the insurance industry can figure out how to cover it. Iron Grid is building the pricing engine the market needs.