- ep 19
- 6 min read
- June 21, 2026
Insuring AI Data Centers: Why Business Interruption Insurance Now Drives the Deal
Hosted by Ashwin Agarwal and Katie Dowson
Data centers are the hottest asset class in commercial real estate right now — capital is pouring in, valuations are enormous, and the buildings are some of the most complex structures going up anywhere. But the insurance behind them is quietly becoming one of the hardest parts of getting a deal done. On a recent episode of The Advocate Insurance Desk, host Katie Dowson and Advocate co-founder and CEO Ashwin Agarwal sat down with Rachel Nixon of IMA — who has placed data center coverage for more than 20 years and recently helped structure a $4 billion placement. Their conversation kept circling one idea: in this asset class, business interruption insurance, not the building itself, is what really drives the deal.
Key takeaways
- A data center can lose staggering sums for every minute it is offline, so business interruption — tied to uptime and service-level agreements — is the center of gravity in the coverage stack.
- A program is really five lines working together: property, business interruption, cyber, liability, and construction (wrap-up) coverage.
- Capacity is growing, but carriers favor best-in-class risk, so how well a facility is risk-engineered increasingly sets the rate.
- The biggest unsolved gaps sit in the gray area between property and cyber, and in sourcing permanent coverage once a center comes online.
- Rachel Nixon's prediction: data center insurance becomes its own asset class — "a class of its own" — within a few years.
What makes data center insurance so different?
A single facility can cost hundreds of millions to build, and its risk profile looks nothing like the warehouses or office towers most commercial property programs are built around.
What has changed most, Nixon says, is the scale and the speed. The work is no longer about finding real estate and insuring a building — it is about uptime, power dependency, and the backup plan for the backup plan.
"The scale and the speed have really changed. It's not just about finding the real estate anymore — it's what's the uptime, how fast can we get there, and how fast can these facilities come online."
Why is business interruption insurance the core of a data center program?
Revenue in a data center is tied directly to uptime, which makes business interruption the line everything else revolves around. Many operators are structured as both landlord and tenant, so the exposure has to be solved from both directions. The trigger is usually a service-level agreement (SLA) on uptime — which is why Nixon urges clients to negotiate those SLAs with their lawyers on the front end, before anything is built.
The sharpest illustration is the deductible. In most property lines, the business interruption waiting period is measured in days. In this asset class, days of downtime would be catastrophic.
"Business interruption used to be measured in days — here's your waiting period until coverage kicks in. In this space, days of downtime would be catastrophic, millions of dollars. It isn't days anymore."
What's actually in a data center insurance coverage stack?
A finished program threads several lines together:
Property covers the building and equipment — the core asset. Coverage is data-driven: the more detail on the machines and their value, the more accurately they can be insured at replacement cost without over-insuring. Fast-depreciating gear (think the latest GPUs) is often revalued quarterly so the program tracks real equipment value.
Business interruption sits on top, tied to uptime and SLAs as above.
Cyber has touched every industry for years; here it covers data breach, network interruption, and related exposures
Liability handles third-party exposure.
Construction coverage — builder's risk and wrap-up programs (owner- or contractor-controlled, OCIP/CCIP) — protects the project during the build, then transitions to permanent property coverage once the facility comes online. That handoff is a notoriously awkward moment, and avoiding duplicative or lapsed coverage is part of the broker's job.
At large scale, structuring the property "tower" is its own discipline — knowing which carriers want to sit at the bottom of the risk versus the top, the same way a directors and officers program gets layered.
Where are the biggest coverage gaps in data center insurance?
Three areas stand out. The first is catastrophe-exposed locations: most centers are deliberately sited away from cat-prone areas, but a coastal or otherwise exposed site gets expensive fast and needs specialized coverage. The second — and the one Nixon flags as the real frontier — is the gray area between property and cyber.
"If there's a power outage without a physical loss, you sneak into the middle. There wasn't physical damage, and there wasn't a cyber event. The market needs to evolve and respond to that area in between."
The third gap is timing: permanent coverage after a data center comes online is harder to source than it should be. Nixon ties that back to capacity and to a market that still needs educating on why these are best-in-class risks worth underwriting.
How does the AI boom change how data centers get underwritten?
The funding wave into data centers is unlike anything commercial real estate has seen. For a broker, that has shifted the role from transactional to strategic — clients now bring their advisor in early, on contracts, indemnification language, and risk engineering, long before the building goes up. The payoff is a facility designed from day one to be something property carriers want to write.
Capacity, importantly, is growing rather than drying up. But carriers still hold the line on best-in-class risk, which makes risk engineering the lever that moves price.
"Capacity is continuing to grow, but the people who have the best risk are going to get the best rate. The better the risk is engineered, the lower the rate for our client."
At the very top of the market, a placement of this scale is less a purchase than an assembly job. Nixon's team "writes the submission" before it ever reaches carriers — laying out why the risk is favorable and where each partner should sit in the tower.
What does real market transparency look like here?
This is where data starts to matter as much as coverage. On the episode, the Advocate team pulled up the Terminal app and walked through the US National Property and Liability indices live — property softening off its peak over the past year, liability bottoming last fall and climbing steadily since. The story is the divergence: two core lines heading in opposite directions at the same time.
Data center data isn't on the platform yet — the market is still early — but Nixon's wish list points straight at where price benchmarking and insurance data analytics earn their keep. She wants a way to score facilities on how they are risk-engineered: fire suppression, water mitigation, power redundancy, and that all-important backup for the backup. That is exactly the kind of like-for-like benchmarking that turns a strong risk into a better rate, and it is how coverage gaps get spotted before they become claims.
What's next for data center insurance?
Several products are still being invented. SLA insurance protects against failing a contractual uptime commitment and is becoming a competitive battleground as operators bid against each other on guarantees. Parametric coverage — trigger-based rather than loss-based — looks almost purpose-built for SLAs and grid reliability. And the industry will eventually need a hybrid cyber-property policy to close the gray area so carriers stop pointing fingers across it.
Nixon's bigger prediction reframes the whole category.
"I think this becomes its own asset class — even separate from real estate. There will be specific insurance built around this infrastructure. It'll be a class of its own."
FAQ
Didn't find your answer?
If you couldn't find the answers you need, feel free to reach out to the host.
