Occupancy, Protection, Exposure: The Three Levers of Price
Every property premium is a story about three things, and a data center tells all three loudly. Occupancy, protection, and exposure are the underwriter's decomposition of risk, and understanding how each moves the price is the difference between accepting a quote and shaping it.
Occupancy is the intrinsic hazard of what the building contains and does. A data center's occupancy is defined by dense electrical load, high-voltage distribution, and the energy stored to keep it running through an outage. The ignition sources are electrical: switchgear, transformers, cabling, power distribution units. The aggravating factors are the backup systems themselves — diesel generators with fuel stored on site, and increasingly large banks of lithium-ion batteries in uninterruptible power supplies. Lithium-ion is the occupancy story of the moment, because a battery in thermal runaway produces its own oxygen, resists conventional suppression, and can propagate cell to cell. Underwriters treat battery rooms as a distinct hazard requiring their own detection, their own suppression, and physical separation from the halls they protect. Occupancy is largely fixed — you cannot make a data center stop being full of energised electronics — but design choices about where hazards sit and how they are isolated shift how much that occupancy costs to insure.
Protection is the defence, and it is where the operator holds the most leverage over price. It runs in layers. Detection comes first: very-early-warning smoke detection that samples the air continuously can catch overheating insulation before there is a flame, buying the minutes that separate a nuisance from a catastrophe. Suppression comes next, and here the water problem drives design toward clean-agent systems — inert gases or chemical agents that suppress fire without soaking the equipment — often backed by pre-action sprinklers that keep water out of the pipes until two independent signals agree there is a real fire. Compartmentation limits how far any single event travels: fire-rated walls between data halls, separation of electrical and battery rooms, containment that stops one hall's problem from becoming the whole building's. Redundancy — the N+1 and 2N configurations of power and cooling — is a reliability concept that doubles as loss prevention, because a facility that can lose a component without losing the load is a facility less likely to suffer a cascading failure. And underpinning all of it is human protection: testing regimes, maintenance discipline, hot-work permits, housekeeping. Insurers reward the discipline because most large losses trace back not to absent equipment but to defeated equipment — the alarm silenced, the valve left closed, the maintenance deferred.
Exposure is what the outside world can do to the building, and it is largely set at siting. Natural catastrophe dominates: flood is the sharpest, because water and electronics are a total-loss combination and because flood risk is knowable from maps and elevation before a shovel breaks ground. Wind and hail threaten roofs and the rooftop cooling plant. Wildfire is a rising concern as facilities push into cheaper land at the urban fringe. Seismic exposure governs whole regions. Beyond nature sits infrastructure exposure: the reliability of the grid feeding the site and the water feeding its cooling, both of which shape not just operations but the frequency of the events insurers pay for. Exposure is the least improvable of the three levers after construction, which is precisely why it belongs in the siting decision — a point the course returns to as its central thesis.
The three levers are not independent, and the interactions are where practitioners earn their keep. Strong protection can partly offset difficult occupancy — excellent detection and clean-agent suppression make a hall of dense electronics far more insurable than the raw occupancy would suggest. But no amount of protection fully offsets bad exposure: you cannot suppress a flood, and an insurer facing a facility on a flood plain will either price the exposure heavily, cap the flood sublimit, or decline it. This asymmetry — protection is improvable, exposure mostly is not — is the structural reason siting and insurability have fused. The lever you cannot move after construction is the one you must get right before it.
Read a data-center placement through these three levers and the premium stops being a mystery. It is the underwriter's estimate of intrinsic hazard, discounted for the quality of the defences, loaded for what the location invites. Move the levers you can — protection above all — and document them, because in engineering-led underwriting the facility that can prove its defences is the facility that pays less.