Reclaimed Water for Data Centers

Typical chemistry envelope across the two make-up categories. Site-specific values depend on source water and treatment train design.

Data center water consumption is one of the fastest-growing industrial water demands in the world. A single hyperscale facility can withdraw up to 5 million gallons of water per day — equivalent to the consumption of a town of 30,000–50,000 people. U.S. data center water draw reached 17.4 billion gallons in 2023 and is projected to exceed 72 billion gallons by 2028.

Operators are responding. AWS uses purified wastewater at more than 20 sites in Virginia and California. Google has committed to replenishing 120% of the freshwater it consumes by 2030. Microsoft is piloting zero-water cooling and bridging the transition with reclaimed make-up. The Veolia and Amazon partnership in Mississippi alone is expected to save more than 83 million gallons of potable water per site, per year.

The strategic question is no longer whether reclaimed water belongs in data center cooling. It is how to engineer it without compromising uptime.

There is no single category. In data center cooling, reclaimed water typically falls into one of four sources:

Each source brings different ionic composition, organic load, microbiological profile, and treatment requirements. Reclaimed water is not a single chemistry — it is a category of chemistries. A treatment train designed for tertiary municipal effluent will not survive on industrial reuse water without recalibration.

Reclaimed water reaching the data center fence usually meets non-potable reuse specifications: low TSS, low turbidity, low BOD. That is not the whole story. The dissolved and microbiological load is fundamentally different from potable make-up — and it is the dissolved and microbiological load that drives cooling chemistry.

Running a cooling tower on reclaimed water without redesigning the treatment train is not sustainability. It is asset risk in slow motion.

Across the reclaimed-water cooling installations we support, three patterns recur on the make-up side. Operators who design for them stabilize early. Operators who do not, do not.

Pattern 1 — Feed variability is the design constraint, not the average.
Reclaimed sources are described by their average specification, but the cooling chemistry program lives or dies at the extremes. The treatment train must be sized for the worst-quality day, not the typical one.

Pattern 2 — Pretreatment maturity decides adoption speed.
Sites with mature make-up conditioning and RO pretreatment trains adopt reclaimed water years earlier than sites that have to build them. Reclaimed-water readiness is, first, treatment-train readiness.

Pattern 3 — Permit pathways are accelerating faster than supply pathways.
In many basins, the regulatory framework now actively prefers reclaimed water — but the municipal reuse infrastructure to deliver it consistently lags. Operators are increasingly negotiating reuse supply agreements years before commissioning.

Milton Roy supports the plant-scale dosing layer required to make reclaimed water viable as cooling make-up:

• Pretreatment chemical dosing — coagulation, pH, biological stabilization, dechlorination
• RO antiscalant and acid injection sized for variable reclaimed feed
• Plant-scale dosing integration with LMI controllers for cooling tower service
• Blowdown neutralization and dechlorination for permit compliance
• Reuse loop dosing for closed-loop blowdown recovery
• Pre-engineered DosaSkid systems for rapid deployment during data center construction