The Loss That Decides the Quarter
Property damage is the loss you can photograph. Business interruption is the loss that shows up in the accounts, and for a data center it is usually the larger of the two. The reason is structural: the physical plant can be repaired or replaced for a knowable sum, but the revenue lost while the facility is dark accrues continuously, day after day, until service is restored. Underwriters and operators who focus on the building and treat the income loss as an afterthought have the ratio backwards.
Business interruption cover — sometimes called business income cover — responds to two things when an insured peril damages the facility: the income the operator would have earned, and the continuing expenses that do not stop just because the servers do. Staff, financing, and fixed commitments carry on. The cover is designed to put the operator, financially, roughly where it would have been had the event not occurred. Two mechanisms bound it, and both matter more in a data center than in an ordinary commercial building.
The first is the indemnity period. This is the maximum length of time the insurer will keep paying while the business recovers, and it is the single most important number in the policy. It is chosen at placement, not at claim, and if it is set too short the operator is exposed for the tail of the outage — the period after cover stops but before revenue returns. The ordinary commercial assumption is that the indemnity period should roughly match reconstruction time: how long to rebuild the damaged structure. For a data center this assumption is dangerous, because the constraint on recovery is rarely the building. It is the equipment inside it, and specifically the pieces with long replacement lead times. A fire may spare the shell entirely and destroy a switchgear lineup or a transformer whose replacement is quoted in months. Setting the indemnity period to reconstruction time when the real gate is equipment procurement is the most common way these policies fail the operator.
The second mechanism is the waiting period, or time deductible — the initial span of downtime the operator absorbs before cover engages, expressed in hours or days rather than money. It functions as a deductible denominated in time, filtering out short outages so the cover responds to genuine, sustained interruptions. For a facility whose value proposition is continuous uptime, even the waiting period represents a real, uninsured exposure, which sharpens the incentive to prevent short outages from becoming long ones.
Quantifying the loss is its own discipline. In a colocation or wholesale model, revenue is contracted and comparatively legible, but the operator's exposure is compounded by service-level agreements: contractual credits or penalties owed to tenants when uptime commitments are breached. Those obligations can turn a technical outage into a cascading financial one, and whether they fall inside or outside the BI cover is a question to settle before a loss, not after. In an owner-operator model — a hyperscaler running its own compute — the 'revenue' lost is harder to pin down, because the facility supports a business whose income is not booked at the rack. Underwriters and operators negotiate a basis for valuing that loss, and the ambiguity is precisely why sophisticated buyers document their business-interruption values as carefully as their property values.
The practitioner's discipline follows from all this. Model recovery from the equipment outward, not the building inward. Identify the longest-lead critical component and let its realistic replacement time — including procurement, shipping, installation, and commissioning, not just the vendor's optimistic quote — set the floor for the indemnity period. Treat the waiting period as a retained exposure and price mitigation against it. And reconcile the contractual penalties owed to tenants against what the policy actually covers. The building is the easy half of the recovery. The revenue is the half that decides the quarter.