Why Atlanta Is a Strategic Data Center Market for Cooling-Water Treatment
Georgia ranks fifth in the United States with 213 operational data centers as of May 20, 2026, behind Virginia (603), Texas (461), California (287), and Illinois (228); the U.S. accounts for 4,286 of roughly 11,400 facilities worldwide (UGA Cooperative Extension TP-121, 2026-05). Metro Atlanta carries the bulk of that footprint, with major clusters in Fulton, Douglas, Newton, Rockdale, and Gwinnett counties. The same publication notes that approximately 57% of U.S. data centers draw from potable supplies and that 80–90% of that withdrawal comes from watersheds shared with public water systems. Atlanta is not officially classified as a Western-style water-stressed basin, but the watershed overlap alone makes blowdown recovery a permitting and ESG question, not just an OPEX question.
The local draw is no longer abstract. The Meta data center in Newton County, opened in 2018, consumes roughly 500,000 gpd — about 10% of total county water use — and pending permit applications could add up to 6 MGD, more than doubling current county consumption (Commercial Water Lab, 2026-06). With Atlanta's hot, humid summers favoring open evaporative cooling, that scale of withdrawal is now landing in front of county boards, EPD reviewers, and pretreatment coordinators simultaneously.
Add a regulatory layer: Georgia Senate Bill 421 — the Data Center Transparency Act, introduced late January 2026 — would bar local governments and authorities from using nondisclosure agreements to keep data center water and electricity data secret. As of May 21, 2026, SB 421 is pending in the State and Local Governmental Operations Committee (LegiScan GA SB421/2025). For an Atlanta EHS manager, this means the gallons, the discharge quality, and the recovery rate you design in 2026 will likely be a public record within a planning cycle. That is the frame the rest of this article works inside.
What Goes Into Atlanta Data Center Cooling Tower Blowdown
Cooling-tower blowdown is the controlled purge that prevents dissolved solids from concentrating past operating limits as pure water evaporates from the tower. At 4 cycles of concentration (CoC), every kilogram of makeup water leaves roughly 0.25–0.30 kg of mineral and chemical residue in the recirculating stream; at 6 CoC the residue density rises proportionally. Atlanta-area municipal feed water typically carries 40–80 mg/L total hardness and 30–60 mg/L silica, so blowdown commonly reaches 250–500 mg/L hardness, 150–300 mg/L silica, 500–1,500 mg/L chloride, and TDS values that can exceed 2,000 mg/L before any chemical additive is counted (Zhongsheng field data, 2026).
EPA guidance and UGA Extension (TP-121, 2026) both flag the additive package that rides with blowdown: anti-scaling and corrosion inhibitors (phosphonates, polymers, zinc), oxidizing biocides (chlorine, bromine, chlorine dioxide), non-oxidizing biocides (isothiazolones, DBNPA), and glycol traces from coil freeze-protection. Cycles of concentration amplify all of them. Direct-liquid and immersion-cooling sites add a second risk layer: PFAS-bearing dielectric fluids and F-gases from refrigerant leaks, neither of which is removed by softening or media filtration and which require adsorbent or membrane polish downstream.
Thermal pollution is the third leg. Any site returning once-through cooling water to a receiving stream (Chattahoochee, Flint, or a reservoir) must evaluate dissolved-oxygen depression and thermal-plume limits under Clean Water Act §316, where state criteria and designated uses can push delta-T requirements into the 1–3 °C range. The chemistry, the additives, and the heat are why a process train — not a single unit op — defines a defensible design.
| Parameter | Typical Atlanta feed water | Blowdown at 4 CoC | Blowdown at 6 CoC | Driver for treatment |
|---|---|---|---|---|
| Total hardness (as CaCO₃) | 40–80 mg/L | 160–320 mg/L | 240–480 mg/L | Scale on tower fill and condenser tubes |
| Silica (SiO₂) | 30–60 mg/L | 120–240 mg/L | 180–360 mg/L | Hard scale; RO scaling at high recovery |
| Chloride | 10–30 mg/L | 40–120 mg/L | 60–180 mg/L | Stainless pitting; discharge toxicity |
| TDS | 80–150 mg/L | 320–600 mg/L | 480–900 mg/L (can exceed 1,500 with additives) | Local POTW limit; NPDES permit limit |
| Phosphonates / polymers | 0 mg/L | 3–15 mg/L | 5–25 mg/L | RO biofouling; P discharge limits |
| Free + combined chlorine | 0.5–1.5 mg/L (tap) | 2–6 mg/L | 3–9 mg/L | Aquatic toxicity; dechlorination required |
| Zinc (from galvanized coils) | <0.05 mg/L | 0.2–1.0 mg/L | 0.3–1.5 mg/L | Local categorical pretreatment limit |
| PFAS (immersion/DLC sites) | Not detected | Site-specific; can exceed 100 ng/L PFOA + PFOS | Higher under CoC | Future Georgia EPD PFAS triggers |
Source: UGA Extension TP-121 (2026-05); EPA Water-Related Permits for Data Centers (2026); Zhongsheng field data (2026). The point of the table is not the absolute numbers — every site is different — but the multiplier effect of cycles of concentration, and the fact that a single blowdown stream carries both conventional pollutants (TSS, metals, pH) and emerging contaminants (PFAS, biocides) that a single unit op cannot address alone. Pretreatment chemistry typically starts with a PLC-controlled chemical dosing system for pH adjustment, scale inhibitor, and biocide feed before the stream hits any physical separator.
Regulatory Pathway: Clean Water Act, Georgia EPD, and Atlanta POTW Pretreatment
Atlanta data center discharges sit at the intersection of three regulatory layers, and getting the layering wrong up front will push the schedule back by months. Direct discharge to a surface water — Chattahoochee, Flint, a reservoir, or any tributary — requires an NPDES permit issued by Georgia EPD under federal Clean Water Act delegation; the same permit covers any thermal discharge under §316 if a once-through or hybrid cooling loop is in play. Discharge to a sanitary sewer falls under the local POTW's industrial pretreatment program, with Georgia EPD retaining approval authority over the program itself. EPA's 2026 guidance on water-related permits for data centers (epa.gov/watersense/water-related-permits-data-centers) explicitly identifies cooling-tower blowdown as a wastewater stream that "may require permits before discharge," and the general pretreatment standards at 40 CFR Part 403 frame the categorical and local-limit conversation regardless of which path the engineer picks.
For metro Atlanta, the receiving POTWs most likely to be in the path are City of Atlanta (R.M. Clayton and Utoy Creek WRFs), Fulton County, DeKalb County (Snapfinger and Pole Branch), Gwinnett County (F. Wayne Hill WRF), and Newton County. Each maintains its own local limits on top of federal categorical standards; the parameters that bite cooling-tower chemistry hardest are TDS (often capped as low as 1,500 mg/L), TSS, total metals (Zn, Cu, Pb from coil corrosion), pH (typically 5.0–10.0), oil & grease, and — increasingly — temperature and PFAS monitoring triggers for sites using direct-liquid or immersion cooling. Early coordination with the receiving utility's pretreatment coordinator, not the building department, is the single highest-value step in concept design.
Add the disclosure layer: Georgia SB 421 (the Data Center Transparency Act, 2026 session) is currently pending in the State and Local Governmental Operations Committee. If enacted, facility-level water and electricity data — including volumes withdrawn, discharged, and recovered — would be public record, regardless of existing nondisclosure agreements. For a procurement lead, that is now a compliance risk input alongside chemistry, and it is one reason treatment trains with quantified recovery rates and named discharge points are the more defensible basis of design in 2026.
Process Train Design: From Side-Stream Filtration to Full ZLD
Four process options cover essentially every discharge scenario an Atlanta data center will face, and they scale linearly with discharge liability and water-stress value. The engineering work is matching the train to the site's local limits, available sewer capacity, and ESG posture, not picking the most expensive option by default.
Option 1 — Pretreatment only. Rotary screening, mechanical bar screening for gross solids, a DAF or lamella clarifier with coagulant and flocculant dosing for TSS, FOG, and metals precipitation (Zn, Cu), and pH neutralization. Discharge to a generous local sewer. Recovery benefit is indirect — it lets the tower operate at 4–5 CoC without fouling. CAPEX is in the low six figures for a mid-sized facility.
Option 2 — Side-stream filtration + softening. A side-stream multi-media filter takes particulates and a fraction of the metals; weak-acid cation or lime-soda softening drops hardness and silica so cycles can climb from 4 to 6, cutting blowdown volume by ~30% (Zhongsheng field data, 2026). Genesis Water Technologies places the CAPEX band for side-stream filtration at $50,000–$200,000 for typical data center capacity (genesiswatertech.com, 2026). This is the lowest-cost path to meaningful freshwater savings and is the right answer when the receiving POTW has a workable TDS limit.
Option 3 — Brackish RO on blowdown. DAF for TSS and metals precipitation in blowdown upstream, cartridge polishing, then a brackish RO skid for blowdown recovery. RO recovery on cooling-tower blowdown is limited by scaling potential, with realistic operating bands of 50–85%; permeate at 10–50 mg/L TDS returns as cooling-tower makeup, and concentrate goes to sewer or to a downstream concentrator. This option captures most of the water-recovery value at a fraction of full ZLD cost.
Option 4 — Full ZLD. RO or NF pre-concentrates to 70–80% recovery, then a brine concentrator or mechanical vapor recompression (MVR) brings total dissolved solids above saturation, and a crystallizer produces a solid cake for disposal. ZLD achieves 95–99% overall water recovery, with solid waste under 1% of original blowdown volume; CAPEX lands in the $3–8 million band for typical data center capacity (Genesis Water Technologies, 2026).
Worked example, 60,000 gpd blowdown at 65% RO recovery: permeate ≈ 39,000 gpd, concentrate ≈ 21,000 gpd. Blending the permeate back into the tower at 6 CoC displaces roughly 1.02 MGD of freshwater per month (per the S5 worked example), a ~25% reduction in freshwater demand for a 10 MGD makeup facility. That is the number an Atlanta EHS manager takes into a budget meeting.
| Process train | Recovery | Permeate quality | CAPEX band | Best-fit discharge path |
|---|---|---|---|---|
| Pretreatment only (screen + DAF + chemical dosing) | 0% direct; supports 4–5 CoC | TSS <30 mg/L, metals at local limit | Low six figures | Generous POTW sewer |
| Side-stream filtration + softening | Indirect; ~30% blowdown reduction | Hardness <50 mg/L, silica <30 mg/L in tower | $50,000–$200,000 | Standard POTW sewer with TDS ≤1,500 mg/L |
| Brackish RO on blowdown | 50–85% | 10–50 mg/L TDS permeate (reusable) | Mid six to low seven figures | Sewer or surface water with tight limits |
| Full ZLD (RO/NF + brine concentrator + crystallizer) | 95–99% | Distillate <10 mg/L TDS; solid cake <1% volume | $3–8 million | Zero surface discharge; solids to landfill |
Source: Genesis Water Technologies (2026); EPA 40 CFR Part 403; Zhongsheng field data (2026). The table is the procurement artifact: same columns, same units, same units of cost, so a CFO can read it without translation. The PLC-controlled chemical dosing system referenced earlier is the integration point across all four options.
Atlanta-Specific Design Considerations: Heat, Humidity, and Water Stress
Atlanta's hot, humid summers are exactly the operating envelope that open evaporative cooling is designed for — and exactly the envelope that maximizes evaporative loss and blowdown volume. A medium-sized Atlanta data center can consume up to 300,000 gpd and a hyperscale facility up to 5 MGD for cooling (UGA Extension TP-121, 2026-05). On a design day, blowdown is a fixed percentage of makeup; raising cycles of concentration from 4 to 6 with the right pretreatment is the single most cost-effective lever in the engineer's hand.
Nationally, roughly two-thirds of data centers built since 2022 sit in water-stressed regions, per Bloomberg analysis cited by the Lincoln Institute. Atlanta is not on the published Western water-stressed list, but the same rule of thumb applies: when the watershed serves both the data center and the public water system, recovery is no longer optional. New construction in metro Atlanta should default to closed-loop or hybrid designs that UGA Extension estimates at up to 70% water-use reduction versus open evaporative cooling. Where open towers remain, RO permeate returned to the cooling loop or to adiabatic/free-cooling augmentation during peak humidity is a high-value reuse path, and it requires a residual-disinfection step — typically on-site on-site ClO₂ for cooling-loop microbiological control — to keep Legionella under control at the lower turnovers that come with reuse. For readers facing similar humidity-driven design loads in other metros, the tropical-humidity blowdown guidance from our Kuala Lumpur guide tracks the same logic.
Cost, Compliance, and Selection Framework for 2026
The decision between pretreatment-only, side-stream + softening, brackish RO, and full ZLD reduces to three questions an Atlanta EHS or consulting engineer can answer in a meeting. First, where does the discharge go — POTW sewer, surface water, or zero? Second, what is the binding local limit — TDS as low as 1,500 mg/L for some metro Atlanta POTWs, categorical metals, a future PFAS trigger? Third, is the site under public or supply-chain scrutiny — Newton County, an SEC-reporting hyperscaler, a colocation operator with a customer water-stewardship covenant? Each combination points at a different train.
Cost bands anchor the conversation. Pretreatment-only is the cheapest CAPEX but pushes the most mass to the sewer. Side-stream filtration + softening at $50,000–$200,000 is the lowest-cost path to ~30% blowdown reduction and supports 6 CoC operation. Brackish RO at moderate CAPEX captures 50–85% of blowdown as reusable permeate — the option that pairs best with water-stewardship disclosures. Full ZLD at $3–8 million CAPEX is reserved for sites where the local limit is essentially "no liquid discharge," where the ESG posture demands it, or where freshwater cost and scarcity make 95–99% recovery economic. The freshwater savings on a 10 MGD facility — roughly 25% reduction when cycles move from 4 to 6 with 65% RO recovery — support a 3–5 year payback under typical Atlanta water and sewer rates (Zhongsheng field data, 2026).
Two operational items that derail otherwise sound designs: engage Georgia EPD and the receiving POTW pretreatment coordinator at concept stage, not post-design, and budget for sludge handling from softening and DAF. A filter press for softening and DAF residuals, paired with a high-efficiency sedimentation tank upstream, prevents the engineer from solving a wastewater problem by creating a solid-waste problem. Treat the SB 421 status as a moving target: recheck before final design and again before commissioning. The high-temperature, high-TDS design parallels in our Dammam data center brief show how the same selection logic shifts when discharge is fully restricted.
| Selection driver | Pretreatment only | Side-stream + softening | Brackish RO | Full ZLD |
|---|---|---|---|---|
| Local TDS limit > 1,500 mg/L, no PFAS trigger | Fit | Over-spec | Over-spec | Over-spec |
| Local TDS limit ≤ 1,500 mg/L, moderate metals | Marginal | Best fit | Fit if reuse desired | Over-spec |
| Tight metals + PFAS monitoring trigger, ESG pressure | Insufficient | Partial | Best fit | Fit if discharge to surface |
| Zero liquid discharge required or strongly preferred | Insufficient | Insufficient | Partial | Best fit |
| CAPEX band | Low six figures | $50,000–$200,000 | Mid six to low seven figures | $3–8 million |
| OPEX profile | High discharge cost | Mid | Low discharge, mid energy | Lowest discharge, highest energy |
Source: Genesis Water Technologies (2026); Zhongsheng field data (2026). Use this as the procurement matrix; the four rows of the body are mutually exclusive at concept stage, and the cost bands translate directly to budgetary line items.
Frequently Asked Questions
What permits does a data center in Atlanta need to discharge cooling-tower blowdown?
Discharge to a surface water (Chattahoochee, Flint, or reservoir) requires an NPDES permit from Georgia EPD under federal Clean Water Act delegation, including §316 thermal-plume review if a once-through or hybrid loop is in use. Discharge to a sanitary sewer requires the receiving POTW's industrial pretreatment program approval, with Georgia EPD oversight per 40 CFR Part 403, including categorical metals, pH 5.0–10.0, and local TDS limits as low as 1,500 mg/L.
What is the typical recovery rate and CAPEX for a data-center blowdown RO system in 2026?
Brackish RO on cooling-tower blowdown realistically operates at 50–85% recovery with permeate at 10–50 mg/L TDS (Genesis Water Technologies, 2026). Full ZLD with brine concentrator and crystallizer achieves 95–99% recovery but at $3–8 million CAPEX for typical data center capacity, versus $50,000–$200,000 for a side-stream filtration and softening package that supports ~30% blowdown reduction at 6 cycles of concentration.
Does Atlanta drinking-water supply affect the design basis for a new data center?
Yes. Approximately 57% of U.S. data centers draw from potable supplies and 80–90% of that withdrawal comes from shared watersheds (UGA Extension TP-121, 2026). The Meta Newton County facility uses 500,000 gpd — 10% of county supply — and pending permits could add up to 6 MGD. That scale is why side-stream softening and RO reuse are increasingly a permitting input, not just an OPEX optimization, and why the process-train section above is the relevant design artifact.
What contaminants in cooling-tower blowdown do conventional softening and media filtration not remove?
Conventional softening and multi-media filtration do not remove PFAS from direct-liquid or immersion-cooling coolants, F-gases from refrigerant leaks, or low-molecular-weight biocides and phosphonates that pass through media. A membrane or adsorbent polish step is required where these are present, which is why the brackish RO and ZLD trains in the process-train section pair with upstream DAF and chemical dosing for credible removal.
How does Georgia SB 421 change the basis of design for a 2026 data center project in Atlanta?
Georgia SB 421 (Data Center Transparency Act, 2026 session) is pending in the State and Local Governmental Operations Committee as of May 21, 2026. If enacted, facility-level water and electricity data would become public record regardless of existing nondisclosure agreements. The practical impact is that water-withdrawal volumes, discharge quality, and recovery rates should be designed to a defensible public-disclosure standard from day one — which is why a documented process train with named discharge points and a quantified recovery rate is the more defensible procurement artifact in 2026.
