Global data center spending is projected to hit $7 trillion by 2030, and the United States leads the world with 4,184 facilities. AI workloads are pushing these builds to new levels of density, power consumption, and cost - redefining what insurers must cover. Jeff Bellmont, SVP at Intact Insurance, said the complexity demands coordinated expertise across property, cyber, energy, and business interruption coverage, a combination few carriers fully deliver.
Construction costs surge as AI workloads reshape facility design
Data center construction costs rose sharply in 2025. Average project costs jumped 70 percent that year, hitting $633 million per project and $1,033 per square foot, according to ConstructConnect. Construction starts reached $77.7 billion, a 190 percent year-over-year increase, with the fourth quarter alone accounting for $44.4 billion. Even mid-market builds now cost between $500 million and $2 billion, per McKinsey.
"It's all changing quickly. And while it's easy to focus on the big $20 billion data centers, there's also a lot of change in the more 'average-sized' data-center builds, as well as existing data centers, that the industry needs to be aware of," Bellmont said.
Why AI-focused data centers challenge traditional underwriting
Facilities built for AI workloads operate at a scale and density that outstrip conventional data centers. They demand more power, more cooling, and more connectivity, all of which compound underwriting complexity. "Those built for AI workloads are meaningfully more dense than the data centers we've seen over the last decade. And with that, they require more power, more cooling, and more connectivity. From an insurance perspective, this all adds to the complexity, both in terms of the items that you're insuring inside and also the risk that comes with possible business interruption," Bellmont said.
He added that the way insurers assess these risks is not changing as fast as the risks themselves are growing. "It's about pulling in experts from [varying] backgrounds too, because the complexity of what they're building is new to a lot of underwriters." Intact has underwritten this segment within its Technology group for years, drawing on risk engineers, technology experts, and power and renewable-energy professionals.
Closing coverage gaps across construction and operational phases
One of the most significant exposure points in the market is the gap between a data center's construction and operational phases. The risk profile shifts once internal contents come into play, and policies written across multiple jurisdictions - London, the US, and Canada - make coordination harder. "You could have holes in the coverage as you're going from one phase to the next; you could have holes in the coverage when you're insuring one part but not the other," Bellmont warned.
His firm's approach is to understand the full life cycle without necessarily insuring every component. "Our view is that we don't necessarily have to insure the full life cycle, but we do have to understand all parts of that cycle in order to help brokers and customers," he said.
Business interruption, aggregation risk, and what brokers need to know
Business interruption is one of the most consequential exposures in data center insurance. Occupancy rates are projected to peak above 95 percent in late 2026, up from around 85 percent in 2023, according to Goldman Sachs Research. Even a brief outage carries severe financial consequences. "With data centers, just a few minutes of downtime is a massive problem," Bellmont said. He drew a parallel to a dairy manufacturer needing a backup plant if something goes wrong - the same urgency applies to data center operations.
As large projects transition from construction to operational status, aggregation risk becomes a defining constraint on market capacity. Six percent of US electricity is already consumed by data centers, and that share continues to grow. "While insurers have dug deep to find the capacity thus far, there remains an open question as to whether that will continue to be the case as these $20 billion projects transition to operational assets, and aggregation risks become more real," Bellmont said. He advises carriers to manage it with discipline from the start. "Consistency is a key theme here - you have to be very thoughtful upfront so that you're consistent over the next few years - don't yo-yo on price or on appetite."
For brokers, speed is critical. "While the demand is going up, there's also a need for speed - so our broker partners are trying to get placements done very quickly," Bellmont noted. He added that the right insurer combines deep underwriting knowledge, engineering capability, and claims responsiveness. "At Intact, we have our own internal experts who understand the coverages, understand the risks - they won't just write a check; they dive in to understand it and ask the right questions." Size alone is not the differentiator. "We're not the biggest insurer, but we happen to have expertise in all the places where data-center clients need it." Brokers who want to deepen their technical knowledge of AI-driven risks can explore AI for Insurance Courses.
Why this matters for insurance professionals
Data center insurance is no longer a niche corner of commercial property. With $7 trillion in spending on the horizon and AI redrawing the risk map, brokers and carriers need to build cross-disciplinary expertise, lock in consistent underwriting approaches, and prepare for aggregation challenges that will test market capacity. The window to get ahead of these shifts is open now.
Your membership also unlocks:
