AI data centers are changing the risk landscape. Claims must evolve with them.

Artificial intelligence is often discussed in terms of software, automation and productivity. But the rapid expansion of AI is also creating a very physical risk challenge: the growth of large, complex data centers that require enormous amounts of power, water, cooling infrastructure and operational continuity.

As AI adoption accelerates, data centers are becoming critical infrastructure for modern business. They support cloud computing, machine learning, advanced analytics and the digital services many organizations rely on every day. But as these facilities grow in size, scale and importance, they are also reshaping the risk environment for owners, insurers, adjusters and the communities around them.

Recent reporting has brought two related issues into focus: the resource demands of data centers and the increasing exposure of planned AI data centers to natural catastrophe risk. In Virginia, where much of the state is under a drought warning, Governor Abigail Spanberger has encouraged residents to conserve water, while data center water usage has drawn public attention. According to the WRIC ABC 8News report, the Sierra Club Virginia Chapter said a Virginia data center uses an average of four million gallons of water a day, while the average person uses around 90 gallons a day. 

While data centers may seem like hidden infrastructure, they support many of the digital services people rely on every day. From checking social media and streaming videos to making online purchases, storing photos, navigating with GPS and collaborating remotely, these facilities power much of modern life. They enable platforms such as Facebook, Instagram, Netflix, YouTube and Amazon, support online banking transactions through services like Venmo and Zelle, host cloud-based photo storage, facilitate virtual meetings and collaboration tools, and deliver critical communications such as email, weather alerts and emergency notifications. In many ways, data centers have become as essential to daily routines and business operations as roads, utilities and telecommunications networks. Their growing importance makes resilience, resource availability and operational continuity increasingly critical concerns.

As I discussed in a recent interview with WRIC ABC 8News, many companies have historically relied on fresh drinking water in their data center operations, while newer facilities are increasingly incorporating reclaimed water. Examples include water captured from nearby ponds, rainwater collected from roofs and property, and reclaimed water from sewage treatment plants. 

That shift matters. Newer facilities may be designed with sustainability and alternative water sources in mind, but older data centers often face meaningful cost and infrastructure challenges if they were not originally built to support reclaimed water systems. In my experience, transitioning older buildings away from fresh drinking water can come down to “dollars and cents,” because significant costs may be involved when a building was not constructed to use renewable or repurposed water sources. 

From my perspective as a loss adjuster, the issue is bigger than water usage alone. It points to a broader question: how resilient are AI data centers when essential resources, including water, power and cooling capacity, are constrained?

Resource availability is becoming a business continuity issue

Data centers are designed to run continuously. Their equipment produces significant heat, and cooling systems are essential to maintaining operations. When water availability becomes stressed, or when facilities are located in areas where drought, heat, storms or infrastructure strain are present, resource planning becomes part of the overall risk profile.

I recently emphasized this point in an interview with WRIC ABC 8News: “We can no longer just look at the sky for a storm. We really need to look at our resource availability. These data centers need to remain reliable neighbors to everyone in Virginia. Sustainability, disaster recovery must go hand in hand.” 

That perspective is especially important for loss adjusting. Traditionally, catastrophe and property loss planning has focused on visible physical perils, such as wind, flood, fire, hail or structural damage. Those risks remain central. But data centers introduce another layer of exposure: the potential for operational disruption when the resources required to keep the facility functioning are unavailable, constrained or compromised.

In this environment, a loss event may not be limited to direct physical damage. It may involve interruption to cooling, water supply, electrical infrastructure, backup systems, network connectivity, specialized equipment, tenant operations or downstream customer services. For adjusters, that means the scope of investigation, documentation and causation analysis may become more complex.

Natural catastrophe exposure is also growing

At the same time, many AI data centers are being planned or built in areas with substantial natural catastrophe exposure. According to Risk Management Magazine, Lloyd’s specialty insurer MS Amlin found that 56% of 670 U.S. data center projects under construction or in planning are located in states highly exposed to hurricanes, severe convective storms, earthquakes or winter storms. These hazards can disrupt power networks and create business interruption risks for the data center and its downstream customers. The at-risk projects represent nearly $800 billion in investment. 

This creates a significant challenge for insurers and risk professionals. Data centers are high-value assets with highly specialized infrastructure. When multiple projects are concentrated in hazard-exposed regions, the potential for aggregation risk increases. Risk Management Magazine also noted that data centers are typically insured through various business lines, which means facilities can unknowingly accumulate exposure across multiple policies. 

For loss adjusters, this has practical implications. A single severe weather event affecting a growing data center corridor could trigger complex property claims, equipment claims, business interruption losses, contingent business interruption issues and service interruption impacts. The loss may not be confined to one building or one insured. It may ripple through utility providers, tenants, technology vendors, customers and dependent businesses.

What this means for loss adjusting

AI data centers require a more integrated approach to claims handling and risk evaluation. The physical building is only one part of the exposure. The operational ecosystem around the facility may be equally important.

Adjusters may need to evaluate:

• Whether the facility’s cooling systems were affected by physical damage, resource constraints or infrastructure failure.
• How water availability, reclaimed water systems or fresh water dependence contributed to the operational impact.
• Whether backup systems performed as intended.
• How long replacement equipment, specialized labor or system validation may take.
• Whether business interruption calculations reflect the full dependency of the facility and its customers.
• Whether multiple policies, coverages or insured entities are implicated by the same event.
• How natural catastrophe exposure and infrastructure fragility may interact.

This is where technical expertise becomes increasingly important. Data center losses require property adjusting experience, but they may also require engineering, equipment, electrical, mechanical, environmental, business interruption and cyber-related expertise. As AI infrastructure expands, claims professionals will need to understand not only what failed, but how that failure affected interconnected systems.

A broader resilience conversation

The rise of AI data centers is not simply a technology story. It is a story about the critical infrastructure that powers modern life, as well as the property, sustainability, business continuity and risk challenges that come with it. These facilities are becoming essential to the digital economy, but their growth brings new questions about where they are built, how they are cooled, what resources they depend on and how losses will be measured when disruption occurs.

For owners and operators, the message is clear: resilience planning must account for more than physical assets. It should include water strategy, power access, utility dependencies, catastrophe exposure, backup systems and community impact.

For insurers, the challenge is to understand how exposures may accumulate across locations, policies and business lines.

For adjusters, the work ahead will require broader expertise, earlier involvement and a deeper understanding of how physical infrastructure supports digital operations.

AI may be transforming business through data, automation and intelligence. But the infrastructure behind it is tangible, resource-intensive and increasingly exposed to both environmental and operational pressures. Sustainability and disaster recovery must go hand in hand. For the claims industry, that means the future of data center loss adjusting will depend on seeing the full picture: not just the storm, not just the building, and not just the equipment, but the complete operating environment that keeps these facilities running. 

Sources

Insights in this article were informed by my recent interview with WRIC ABC 8News on data center water consumption and resilience planning, my Claims Journal article “When the Cloud Goes Dark: Data Center Claims And Specialized Adjusting Expertise,” as well as Risk Management Magazine’s coverage of natural catastrophe exposure associated with planned AI data center development in the United States. [wric.com], [reinsurancene.ws], [claimsjournal.com]