Natural disasters resulted in $107 billion in insured losses in 2025, with almost half of total economic losses covered by insurance, according to
The insurance industry is facing the growing frequency and severity of natural disasters — driven by $40 billion in insured losses from the Los Angeles wildfires and $50 billion in insured losses from severe convective storms — as well as the increase of population movement to high-risk urban areas, the study said.
Climate experts, insurance leaders and tech innovators discussed the major risks, trends and technology shaping the insurance industry across the U.S. at ClimateTech Connect, held April 8-9, 2026, in Washington D.C.
Here were the key takeaways:
'It's not a silver bullet problem. It's a silver buckshot problem'
Patrick Keegan, chief enterprise underwriting officer, Travelers
"[Climate resilience] is such a complex, multivariate issue that requires a solution that's well beyond the insurance industry. It requires engagement across industries and from policymakers, on the local and state levels. The issues of affordability and availability that we're dealing with in insurance are actually a symptom of the underlying core issues that are creating stress in the system. You don't treat the symptom; that's a short-term issue. You have to really focus on what the drivers are.
"We do think it's a local issue, and the issues that you have to solve for in California are different than in Texas… The effort should be grassroots, community-based, multi-industry engagement, from the builders to the realtors to capital markets… They're the ones that actually have the power to drive change. It's the people in the community that represent industry. It's the local governments and politicians who care about the bigger issue that roll up their sleeves and collaborate."
Sarah Kapnick, global head of climate advisory, J.P. Morgan
"[Building resilience] is a systematic issue across different sectors and locations that have very local supply solutions, but also solutions that are required across government, corporate, individuals and consumers. It's not a silver bullet problem. It's a silver buckshot problem, where you need a lot of different pieces coming together in the planning. What I'm seeing increasingly in my discussions with corporates, state governments and federal governments around the world is trying to pull these different pieces together, and based on the policy or the spending power of corporations in different places, the solutions are different.
"For resilience, your specific set of problems and solutions rely on your location for the natural hazards, the physical risks and also your existing infrastructure: the age of a structure, what building code it was built on, what are your capital pools available to fund resilience projects. That's why it requires that nuance at the local scale and can't necessarily be done at a high level."
'Think about why you choose certain data sets, and how are you accounting for the biases'
Kapnick, on how organizations are thinking about embedding AI tools in predictive catastrophe models
"[Data] is foundational to the creation of these types of models and our understanding. It puts emphasis back on the data collection and observation that is critical… and the other place I'm starting to see it come in on model development is when you're adding disparate information, trying to create forecasts or predictions of what the physical climate looks like.
"On the impacts on society or finance or infrastructure, that is data that you otherwise would not put together. That is where there's interesting capabilities of forecasts that are so fast. You're not running a physical risk model, then running the social, then running the financial model. When you start blending it for speed, that's when you'll start having changes in behavior on forecasts that will be interesting."
Kelly Hereid, head of catastrophe research and development, Liberty Mutual
"I think we need to tread carefully. There are certainly places where there are interesting opportunities to automate...The fact that we can automate [satellite data] using AI-related processes now is super powerful and a really effective way of scaling up what you can do to ingest data sets that would be inaccessible from just an operational standpoint before.
"For vendors, think about why you choose certain data sets, and how are you accounting for the biases in those data as you are training your models. What are the relative strengths and limitations of those underlying data choices? For people who are producing AI-related analytics: if you want those to be embedded in organizations, it's not enough to say you have a model that just produces a number. You have to explain how you have scientifically justified the choices in what you build."
'It's not the big named disasters anymore that keep insurers up at night, it's something far more frequent and harder to predict.'
Denise Garth, chief strategy officer, Majesco
"Recent headlines note that severe convective storms dethroned hurricanes as consistent costly insured weather perils. It's not the big named disasters anymore that keep insurers up at night, it's something far more frequent and harder to predict. At the same time, we continue to see legal system abuse, and increased volatility and cost. The result is a growing protection gap. Insurance is no longer just about coverage, it's about resilience, proactive mitigation and closing that protection gap in real time."
Robert Pick, EVP and CIO of Tokio Marine North America, group deputy CITO of Tokio Marine
"There's the coverage gap, but there's also another gap — some people have adequate coverage, but they actually can't afford to pay their deductible…That deductible gap is where parametric programs can be extremely valuable.
"[Parametric] is basically based upon a parameter, so you will be paid a sum of money in the event of a natural catastrophe. For example, it could be for a Category 5 hurricane, you get $10,000 … It's a lump sum that is, in part, entitled to just give you that jumpstart, either pay your deductible or get back and rebuild your property."
Mike Gulla, CEO and Co-founder, Adaptive Insurance
"It's an 'if-then' statement. If this event happens, then I get paid for that event. I look at parametric as the first opportunity for the industry to actually get consumers to invest in resilience.
"As an example, if you're a restaurant owner in Austin, Texas, you may have a parametric product for a power interruption in your business. It's not about adjudicating the claim. It's about getting the funds into the hands of that business as fast as possible after a loss occurs. That business owner is now making a more informed decision about how they can keep their business in business over time, and it's the same for homeowners.
"The first dollar of coverage on every homeowner claim is to the consumer, not to the insurance company. If the consumer doesn't have that dollar to pay for that loss, the insurance company is now taking a greater risk on that piece of property because the customer may not do those repairs … Parametric is a great option for that. It's not the only but it's a new way for us to bring both new premium dollars into the market and new capital into the claim space that can be deployed very, very rapidly after one of these events occurs."
