WILDFIRE INSURTECHDELOS WRITES THE HOMES THE MARKET WON'T65% OF DECLINED RISKS = ACCEPTEDSATELLITES + GEOSPATIAL AIFOUR FIRE SEASONS, ZERO BURNED INSUREDS REPORTEDPROGRAM INSURTECH OF THE YEAR WILDFIRE INSURTECHDELOS WRITES THE HOMES THE MARKET WON'T65% OF DECLINED RISKS = ACCEPTEDSATELLITES + GEOSPATIAL AIFOUR FIRE SEASONS, ZERO BURNED INSUREDS REPORTEDPROGRAM INSURTECH OF THE YEAR
Person / Founder / Wildfire

Kevin Stein

He sold satellites for a living. Then he pointed them at the rooftops California insurers had given up on.

Kevin Stein, CEO and cofounder of Delos Insurance Solutions

Kevin Stein - the aerospace engineer who reads a hillside the way underwriters read an actuarial table.

2017
Delos founded
65%
Declined risks accepted
6 wks
To model new fire behavior
~$22.5M
Total funding raised
The Premise

A house can be in a fire zone and still be the wrong house to refuse.

Start with a number that should not exist. Of roughly six million American homes that struggle to find insurance across thirteen states, Kevin Stein says only about three million are actually exposed to serious wildfire risk. The other half are being turned away for living near a problem rather than being the problem. Delos Insurance Solutions, the San Francisco MGA he leads, was built on the gap between those two numbers.

Stein is the CEO and cofounder. The job, as he describes it, is less about underwriting in the old sense and more about measurement. Where a generalist carrier buys a third-party catastrophe model and applies a blunt rule - this ZIP code is out - Delos pulls satellite imagery, climate data, and a geospatial AI algorithm down to the individual property. The result is a company that writes coverage in the burn-prone canyons and ridgelines that mainstream carriers have spent the last decade fleeing.

"In this era of climate change, the traditional approach of a generalist carrier using a third-party data model for evaluating catastrophe exposure is no longer a viable solution."

Kevin Stein, CEO of Delos

The claim Delos likes to repeat is a stark one: across four fire seasons, it reports it has not insured a single home that later burned, while running 90 to 95 percent accuracy on its exposure analysis. For a market that treats wildfire as a coin flip, that is the entire pitch in one sentence.

The Pivot

From spacecraft to shingles.

Stein is a Bay Area native who studied engineering at UC Berkeley and went on to a master's at Stanford. His first career was orbital, not actuarial. At SSL - the satellite maker later folded into Maxar - he worked as a mechanical systems engineer, designing spacecraft and, by Delos's account, managing more than $50 million in satellite sales. The instinct he carried into insurance is the same one that gets a satellite into the right orbit: trust the data, distrust the assumption.

In 2017 he and cofounder Shanna McIntyre, also from the aerospace world, started Delos on a contrarian read of the California market. The big carriers were retreating from wildfire country. Stein and McIntyre saw the same imagery the rest of the industry had access to and concluded the retreat was indiscriminate - that satellites and modeling could separate the genuinely dangerous homes from the merely nearby ones.

To do it, Delos partners with Spatial Informatics Group, an environmental think tank of more than a hundred academics who build wildfire models for civil governments. Stein's framing of that talent is blunt: these are people who, in his words, get paid serious money to do nothing but wildfire research season after season. Delos folds that science directly into the MGA rather than renting it from a vendor.

"It takes Delos only six weeks after a new type of wildfire behavior evolves to understand what datasets would enable intelligent prediction."

Kevin Stein
The Model

Wildfire is the new cyber.

The analogy Stein reaches for is cyber insurance: a peril that got so specialized, so fast, that generalist carriers simply couldn't price it anymore and ceded the ground to specialists. He argues wildfire has crossed that same line. The loss driver most people underrate, he says, is not the fuel on the ground but the wind and weather pushing the flames - the single biggest driver of wildfire loss across the Western U.S., and the thing Delos's models weigh most heavily.

That conviction shows up in what Delos is willing to write. The company accepts roughly 65 percent of the business the primary market declines for perceived wildfire exposure - a very significant amount, in Stein's words, of homes other insurers won't touch. When the January 2025 fires tore through Southern California, Delos pointed to its models as having flagged the risk areas in advance, and Stein used the moment to make the public case for science-led underwriting over blanket withdrawal.

50%

Falsely "uninsurable"

Half of the homes that can't find coverage aren't actually exposed, by Stein's count.

100+

Academics on tap

Delos co-develops modeling with the wildfire researchers of Spatial Informatics Group.

90-95%

Exposure accuracy

The reported hit rate of Delos's exposure analysis over four fire seasons.

$9M

Series A, Oct 2024

The latest raise feeding the model, on roughly $22.5M of total funding.

The Person

An underwriter who plays bass.

For all the talk of algorithms, the detail Delos chooses to share about its founder is musical: in his free time, Stein moonlights as a bassist and songwriter. It fits a certain kind of engineer - the sort who likes systems with structure but room to improvise, and who is comfortable being the one holding down a line nobody else wants to play.

The throughline of his story is that he keeps showing up where the rest of the field is walking out. Carriers leaving California read like a verdict; Stein read it as a mispriced market. The same imagery that priced his satellites now prices your roof. And the bet underneath all of it is that climate change rewards specialists who measure carefully and punishes generalists who guess - a wager Delos is making one wildfire-zone policy at a time.

In His Words

"We leveraged our wildfire science and aerospace engineering expertise to develop a suite of wildfire models that includes a novel geospatial AI algorithm."

"Out of approximately six million homes across 13 states that struggle to find insurance, only three million or 50% are actually exposed."

"We're accepting a lot of business that other people are declining - a very significant amount."

"We know which homes are actually low-risk."

Pass it on
LinkedIn X / Twitter Facebook Instagram