PROPERTY RISK IN THE AGE OF CLIMATE CHANGE: HOW UNDERWRITERS ARE FIGHTING BACK WITH GEOSPATIAL AI
Author: Jerry Papadatos, Pranav Despande
- June 04, 2026
- 5 Mins read
Share us on:
In January 2025, Munich Re summarized the year that had just passed in four words: “Climate change is showing its claws.”
The numbers behind that statement are stark. Global insured catastrophe losses reached $137 billion in 2024, the fifth consecutive year above $100 billion, with losses growing at 5 to 7 percent annually in real terms over three decades. More than double the rate of GDP growth. Hurricanes Helene and Milton alone generated $44 billion in combined insured losses in a single season.
For commercial P&C underwriters, the challenge is not primarily risk volume. It is risk visibility. Carriers are pricing exposures they cannot fully see, on properties they cannot accurately value, using data that reflects a climate and economic reality that no longer exists.
That is a solvable engineering problem. And geospatial AI is how the industry is beginning to solve it.
THE VALUATION CRISIS NO ONE TALKS ABOUT Geospatial AI for Property Underwriting
Commercial property replacement costs have risen nearly 40 percent above pre-2020 levels. Construction labor shortages, materials inflation, and supply chain disruptions drove costs up sharply through 2021 and 2022, and those elevated costs have become the new normal.
Meanwhile, most commercial property valuations are refreshed on 3 to 5 year cycles. The result: 87 percent of commercial buildings are currently undervalued, with insurance-to-value gaps commonly exceeding 30 percent.
A policy written on a 2021 valuation in a 2026 cost environment isn’t priced for the risk it covers. It is priced for a building that is, in economic terms, no longer the same building. When a total loss claim arrives, the gap between insured value and replacement cost falls on the carrier, the policyholder, or both — with no good outcome in either direction.
This isn’t an edge case or a tail risk. It is a systemic condition across the commercial property book, and it is getting worse as the gap between current costs and assessment dates widens.
WHY CURRENT DATA FAILS UNDERWRITERS
The data that would support more accurate commercial property underwriting exists, in fragments, scattered across systems that were never designed to work together.
Building permit records capture renovation activity and structural changes. Physical inspection reports contain condition assessments. Satellite imagery provides visual evidence of roof condition, structural characteristics, and surrounding hazard context. Renovation histories reveal investment patterns and aging trajectories. Geospatial AI hazard zone maps quantify flood, wind, and wildfire exposure.
Each of these sources contains signals that experienced underwriters would act on. But in practice, 42 percent of underwriting executives cite lack of information or analytics access as their primary barrier to better decisions (Accenture, 2025). The data exists. The infrastructure to connect it, normalize it, and deliver it to an underwriter at the moment of decision does not — at most carriers and analytics providers.
Physical inspections address the visibility problem for individual properties, but they cannot scale to a commercial portfolio. They are expensive, slow, and constrained by geographic and scheduling logistics. Satellite imagery can cover millions of properties in a single update cycle — but only if the organization has built the infrastructure to process petabyte-scale geospatial ai datasets, extract the right structural signals, and integrate them into the underwriting workflow in a form underwriters can actually use.
Legacy architectures were not built for that workload. The result is underwriters spending time on data aggregation rather than risk assessment.
WHAT GEOSPATIAL AI CHANGES
Computer vision models trained on satellite and aerial imagery can do at scale what physical inspections do for individual properties — and in a fraction of the time.
Roof condition. Structural characteristics and material type. Proximity to flood zones, active wildfire perimeters, and wind exposure corridors. Evidence of recent renovation or visible deterioration. These signals are visible from above, at the level of individual properties, across an entire commercial portfolio, updated with each new satellite pass.
When geospatial ai data flows through automated ingestion pipelines into a unified property data model — standardizing roof age, electrical status, structural details, and hazard proximity into a consistent, queryable schema — and surfaces through real-time APIs, underwriters gain a 360-degree view of each property at the point of quote. No manual aggregation. No waiting for physical inspection scheduling. No 2021 valuations applied to 2026 replacement costs.
THE LAKEHOUSE FOUNDATION
Geospatial AI does not operate in isolation. Its value compounds when satellite-derived building intelligence combines with structured risk records, historical claims data, portfolio concentration information, and real-time climate hazard signals in a unified analytics environment.
The Databricks lakehouse architecture provides the infrastructure layer that makes this possible. It handles petabyte-scale geospatial ai datasets while delivering the query performance and data quality guarantees that production underwriting workflows require. It unifies structured and unstructured data sources like satellite imagery alongside policy records, geospatial ai signals alongside claims histories into a single analytical layer that data science teams can build on without spending the majority of their time cleaning and preparing data.
The downstream AI applications are high-value and increasingly achievable: climate risk overlay models that score individual properties against real-time flood, wind, and wildfire exposure data; portfolio concentration analytics that surface hidden accumulation risk before it becomes a catastrophic-loss event; GenAI-assisted underwriting tools that translate complex property intelligence into plain-language decision summaries at the point of quote.
Accenture projects that AI and GenAI’s share of underwriting tasks will grow from 14 percent to 70 percent within three years. Commercial property is where many of the most transformative early applications are already demonstrating measurable results.
THE CARRIERS WHO WILL WIN
The commercial P&C carriers who grow profitably through the next decade of climate volatility will not be the ones who raised rates the fastest or exited the most exposed geographies. They will be the ones who can see what competitors are guessing at the property level, at the point of underwriting, with data that reflects today’s replacement costs and today’s hazard environment.
That requires investing in the geospatial ai data infrastructure now — before the next major loss event, not in response to it.
At Nallas, we build the data pipelines, lakehouse architectures, and AI-ready property intelligence foundations that enable carriers and analytics providers to underwrite commercial property at the precision the current risk environment demands. Connect with the Nallas Insurance Practice to discuss your property data infrastructure and where geospatial AI fits in your underwriting strategy.
Authors
Jerry Papadatos
Director - Sales
Pranav Despande
Lead Strategy
