Make reclaimed or non-potable water the default permitted supply for cooling

The proposal

Where evaporative cooling is retained, condition the permit on the source rather than only the volume: treated municipal effluent, brackish or non-potable groundwater, industrial process water, or captured stormwater, with a potable connection permitted only by explicit exception and with a stated justification. The utility and the operator jointly fund the reclaimed water main where one does not exist.

Why it would work

Cooling does not need drinking-water quality. It needs water with controlled scaling, corrosion and biological characteristics, all of which treated effluent can meet with conventional side-stream treatment. Separating the quality requirement from the source removes the direct competition with drinking supply that drives most local opposition, and it does so without the energy penalty that closed-loop mechanical cooling carries.

Reported practice: Microsoft's San Antonio site reports 79% of consumption from recycled or non-potable sources (DCD). Google reports 87% of freshwater withdrawal from low or medium risk sources in 2025 (Google). Both are company-reported. The industry-wide share of non-potable use is not published anywhere, so anyone citing one is estimating.

Implementation path

The instrument is the water utility's service agreement plus the land use permit. Three components:

1. A source condition in the permit, naming the acceptable supplies and requiring the operator to demonstrate availability before construction.
2. Cost sharing for the reclaimed main, since the absence of purple-pipe infrastructure is usually the binding constraint rather than the water itself.
3. A drought clause specifying the curtailment order, so the facility's allocation is reduced before residential supply during a shortage.

Chandler, Arizona's per-square-foot cap and Marana's ban on municipal potable supply to data centres are working examples of the permit-side instrument.

Trade-offs and limitations

• Infrastructure is the constraint. Reclaimed distribution exists in a minority of jurisdictions, and building it for one customer is expensive. This is a strong argument for requiring the commitment before rezoning, when the operator's leverage is lowest.
• It does not reduce consumption, only reallocates the source. In a closed basin where effluent currently returns to a river or recharges an aquifer, diverting it to evaporation is still a consumptive loss. The benefit is real where effluent is discharged to sea or to a losing reach, and much weaker where it is already reused downstream.
• It leaves the indirect footprint untouched. The water embedded in the electricity, roughly ten times the direct cooling volume in US averages, is unaffected by the on-site source.
• Replenishment credits are not a substitute. Watershed replenishment projects in one basin do not relieve scarcity in another, and Amazon's water-positive accounting rests on projects "expected" to return volumes, with no requirement that the benefiting communities are the ones hosting the facilities. Source substitution is a physical change in the affected basin; replenishment accounting is not.