The industrial data center market is one of the fastest-growing segments within commercial real estate, driven primarily by surging demand for computing power due to artificial intelligence (AI). No longer a niche asset class, data centers are quickly becoming core infrastructure, with global investment projected to reach approximately $6.7 trillion by 2030.
While this demand is reshaping the broader real estate market, pushing land values higher in locations with access to reliable power and accelerating the adaptive reuse of older industrial and office buildings, it is also changing the risk profile in ways that traditional underwriting models do not always address. Data centers concentrate high-value equipment and critical infrastructure in a single location, where even a short disruption can carry significant operational and financial consequences.
This means insurers are underwriting data centers more carefully with increased scrutiny on operational risks and loss drivers. Rising loss severity and tighter liability terms may shift more risk to owners. Insurability is no longer defined solely by the asset itself, but by how well a facility can operate under stress.
For data centers, that means more scrutiny of the systems that keep facilities running: power reliability, cooling performance, and fire protection design. It also means owners may need to show how well core systems can withstand disruption.
Retrofitting data centers: what insurers look for
Retrofitting existing buildings for data center use has become an increasingly popular path for owners looking to move quickly and reduce upfront development costs. But retrofits can introduce risks that aren’t always obvious once the facility is finished, and insurers are closely evaluating how well legacy structures have been adapted to meet the demands that modern data center operations require.
Two common retrofit approaches:
- Adaptive reuse. Repurposing warehouses and older office properties can unlock value in underutilized assets, but these buildings were typically not designed for the power density, cooling loads, or redundancy standards data centers require—and bringing those systems up to spec can add substantial complexity and cost.
- Updating existing data centers. AI workloads demand significantly more power per rack than traditional computing. Facilities built even five years ago may need major electrical and cooling overhauls to meet current demands.
With any type of retrofit, the underwriting concern is the same: how well critical systems have been built to perform under stress, and how well owners can demonstrate it.