The Third Constraint

Data-center siting has been a story of successive binding constraints. First came power: the industry learned that the availability, cost, and speed of a large grid connection decided where facilities could go, and it began clustering wherever electricity was abundant, cheap, and quick to interconnect. Then came water: as cooling loads grew, water availability and the right to use it emerged as a second gate, redrawing the map away from water-stressed regions and forcing new cooling designs. The argument of this course is that a third constraint is now arriving on precisely the same logic, and its name is insurability. The insurer's willingness to stand behind a location — and the price of doing so — is becoming a gating input to site selection alongside power and water.

The mechanism is exposure, the third of the underwriter's levers, and its defining property is that it is fixed at siting. Recall the asymmetry from Module 1: occupancy is largely intrinsic, protection is improvable throughout a facility's life, but exposure is set the moment the location is chosen and cannot be meaningfully changed afterward. A finished building cannot be lifted off a flood plain, moved out of a wildfire corridor, or relocated away from a seismic fault or a hail-prone plain. Whatever catastrophe exposure the site carries, the facility inherits for its entire operating life. Because catastrophe exposure is the dominant driver of large property and business-interruption losses, and because it is locked in at siting, the insurability of a facility is effectively decided when the map is marked — which is exactly what makes it a siting constraint rather than a procurement decision.

Several forces are converging to make this constraint bind now rather than in theory. Natural-catastrophe losses have been rising across the insurance industry, and insurers price physical risk from current conditions rather than historical calm, so worsening flood, wildfire, and severe-convective-storm activity feeds through to data-center terms quickly. At the same time the facilities themselves have grown, concentrating ever more value in single locations and raising the maximum foreseeable loss that any one catastrophe can produce. Insurance capacity for these large, concentrated, catastrophe-exposed risks is finite, and when demand for cover rises faster than insurers' appetite to provide it, the market rations through price, through sublimits on the sharpest perils such as flood and wildfire, and ultimately through declining to write particular locations at all. The same clustering that solved the power problem compounds this, because siting many facilities in one favourable-power region concentrates insured value there and can exhaust the catastrophe capacity available for the whole cluster.

The practical form the constraint takes is worth being precise about, because it rarely appears as a flat refusal. More often it is a set of terms that quietly changes the economics of a site: a flood sublimit far below the facility's value, leaving a large uninsured layer the developer or its lenders must absorb; a business-interruption indemnity period the insurer will not extend to match real equipment lead times in a high-exposure zone; a premium load heavy enough to alter the project's returns; or a requirement for protective works as a condition of cover. Each of these is insurability expressing itself as cost and constraint rather than as a yes-or-no, and each can render an otherwise attractive site — good power, adequate water, willing jurisdiction — uneconomic. That is the sense in which insurability now sits beside power and water: not usually as an absolute prohibition, but as a real gate that can close a site that clears every other test.

There is a second-order effect that sharpens the point. Insurance terms are increasingly a condition of the capital that builds these facilities. Lenders and investors require adequate cover before they will fund a project, so an uninsurable or thinly insured location is not merely riskier to operate — it can be harder to finance at all. Faraday's read: this is what makes insurability behave like a hard constraint rather than a soft preference. When the insurer will not stand behind a site on acceptable terms, the money often will not stand behind it either, and the project stalls for the same reason it would if the power or the water were not there. The three constraints have begun to reinforce one another, and a location that fails on any one of them fails on the whole.