This research is put together by a non-partisan group of Columbus residents. Page last updated 14 July, 2026.
Electricity bills for communities near data centers increase on average by 267%. (Bloomberg, 2025)
Newton County, GA experienced an over 50% increase in water-sewer bills after a Meta data center was built. (University of Washington, 2026)
63% of the rate hikes across Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, New Jersey, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, West Virginia, and the District of Columbia in 2025/2026 were due to hyper-scale data centers — amounting to $9.3 billion in extra costs paid in higher electric rates. (Institute for Energy Economics and Financial Analysis)
Georgia Power and the Georgia Public Service Commission's rate freeze applies to the base rate charge line item on your bill ONLY, for three years ONLY. Fuel costs — which data centers increase by 5–11% on average (PSC Public Interest Advocacy Staff, as cited in GPB, 2026) — plus financing for new major projects and power plants for data centers, with an estimated cost of $50–60 billion (SELC, 2025), and other costs will not be frozen.
The water data centers do return to the system comes back as industrial wastewater, or ‘blowdown.’ Recent research from George Mason University (2026) describes its potential impact:
“Cooling water blowdown contaminants such as sodium, magnesium, sulfate, and potassium can contaminate groundwater supplies, posing serious health risks when their concentrations increase (Soliman et al., 2021). For example, high sulfate levels in groundwater can cause skin irritation and diarrhea, especially in infants (Miao et al., 2011). Exposure to contaminated drinking water during pregnancy has also been associated with adverse fetal outcomes, including low birth weight and preeclampsia (Currie et al., 2013; Thompson et al., 2022).”Data centers are sources of PFAS forever chemicals. What are PFAS?
“PFAS, also known as forever chemicals, are very durable and accumulate – in both the environment and the body – over time, posing real concerns for long-term human and ecological health” (Environmental and Energy Study Institute, 2026).
Our city overlay prohibits PFAS from being used in the data center’s cooling liquid. There are a variety of other sources of PFAS, however:
“Data centers host tens of thousands of servers that run 24/7 in order to keep virtual networks, cloud storage, and computing in operation. Such servers require semiconductors, cooling systems, and fire suppressants—all sources of PFAS forever chemicals” (EESI, 2026).
Data centers emit a constant, industrial ‘hum’ that causes real harm to nearby communities and can generally be heard up to two miles from the facility — although on flat terrain it may travel further. (Sierra Club, 2025)
Researchers at George Mason University describe the serious impact this exposure can have on human health:
“Continuous operation of cooling systems and generators produces persistent noise exposure near data centers, creating a direct environmental stressor. Short-term effects include sleep disturbance and annoyance, while long-term exposure is associated with cardiovascular, cognitive, and mental health impacts (Babisch et al., 2013; Thompson et al., 2022).”
It continues:
“The health impacts of noise pollution include long-term risks such as heart disease, permanent hearing loss, and tinnitus; chronic effects such as hypertension and endocrine disruption; and acute effects like decreased sleep quality and temporary hearing changes (Babisch et al., 2013; Hammer et al., 2014). Noise pollution has also been linked with an increase in mental health concerns, such as stress-related disorders and increased annoyance, particularly in sensitive individuals (Meerlo et al., 2008; Sandrock et al., 2008).”Children living near such projects would be especially impacted:
“...studies have found that children in noisy environments, defined by the U.S. EPA as chronic exposure above 55 dBA (day–night average), exhibit poor school performance, increased stress, and behavioral issues, with noise pollution also being linked to decreased learning, lower reading comprehension, and concentration deficits (Lercher et al., 2003; Stansfeld and Clark, 2015). For children residing near data centers, the constant noise from 24/7 operations could further exacerbate these developmental and cognitive risks, potentially impacting their long-term health outcomes” (George Mason University, 2026).
The current Columbus overlay would allow for noise production at 65–70 dBA during the day (7am–10pm) and 55–65 dBA at all remaining hours of the night — all levels above the EPA’s designated chronic exposure threshold.
Yellow line = EPA chronic exposure threshold for children: 55 dBA (day–night average)
Incredibly Important
While the noise pollution may today be limited to those within two miles of Project Ruby, the overlay passed by our City Council actually allows hyper-scale data centers to be built not just in industrial areas, but in residential (RE-10, RE-5, RE-1) and commercial zones as well. Here’s a map of all those areas of Muscogee County in which a hyper-scale data center could be built if land were to become available:
Meanwhile, those communities, families, neighborhoods, firefighters, and more who would be most directly affected by the hyper-scale data center hum described above given Project Ruby’s current bounds are mapped below.
Georeferenced to nearby landmarks; treat as approximate (±500 ft) until a survey plat is filed. The 2-mile line is a screening distance, not a noise study – audibility varies with equipment, terrain, tree cover, weather, and time of day. Distances shown run from each landmark to the nearest point of the site boundary.
Hyper-scale data centers consume "up to 5 million gallons of water every day." (Lincoln Institute of Land Policy, 2025)
Columbus is a drought city. We have spent 60% of the last 26 years in an official drought state, and at least 16% in "extreme" or "exceptional" drought — the two worst categories that exist. (Climate Check; U.S. Drought Monitor)
No. The overlay approved by our City Council does not legally bind Project Ruby to any limit on water consumption. The language recommends closed-loop technology, which is a less water-intensive form of cooling for data centers, but it still allows any system "without limitation" if the company claims to 'minimize' water usage. Our current laws also don't include any reporting requirements.
The company will never have to tell us how much water it’s using — not even in a drought. In Dalles, Oregon, for example, Google funded a lawsuit against a local newspaper to block the release of its data center water use as a “trade secret.”
Important
Developers often severely underestimate the amount of water their facility will use. Here are some examples:
A) Coweta County, GA: The developer claims on their website “a potential peak water use of 300,000 gallons of water per day.” However, in their actual zoning application (a legal document) they admitted the real figure was NINE MILLION gallons!
B) Tucson, Arizona: Before the Tucson City Council, the developers claimed that their data center would use no water. They then secretly applied directly to the State of Arizona – instead of Tucson – for a permit for 117 million liters/year.
Even if Project Ruby adopts the less water-intensive technology (closed-loop) recommended by our Council, an honest estimate of its daily water consumption would be 2.9 and 4.3 million gallons/day, conservatively. Why? A closed-loop system trades water for electricity, but that significant increase in electricity demand has its own massive water footprint. See the calculations
Research 1 · Closed-loop system — the water cost of electricity
source .... National Renewable Energy Laboratory estimate .. 0.2 gallons of water per kWh of electricity generated (conservative)600 MW × 24 hrs/day = 14,400,000 kWh/day 14,400,000 kWh × 0.2 gal/kWh = ≈ 2.9 million gallons/day
If built out to 900 MW: 900 MW × 24 hrs = 21,600,000 kWh/day × 0.2 = ≈ 4.3 million gallons/day
If the company chooses to use a different form of cooling technology, like closed loop with evaporative assistance, the withdrawal estimate increases to between 4.3 and 6.5 million gallons of water a day, conservatively. See those calculations
Research 2 · Closed loop with evaporative assistance
source .... UC Riverside, CalTech & RIT estimate .. 0.72 – 1 million gallons/day per 100 MW (0.72 = conservative)600 MW ÷ 100 MW = 6 × 0.72M gal = ≈ 4.3 million gallons/day (at least)
If built out to 900 MW: 9 × 0.72M gal = 6.48 million gallons/day
It would mean a permanent drain on our water supply in a region already threatened by drought. Why? Households return most of their water to the system; data centers do not.
When you take a shower, that water goes down the drain, through Columbus Water Works, and back into the Chattahoochee. Public water systems in the Southeast return about 90% of what they withdraw (U.S. Geological Survey, 2025). Data centers eat about that same amount — between 70–90% is never returned to local water sources (Han et al., 2026).
In addition, droughts – for ourselves and for our kids and grandkids here in Columbus – could become much more painful. This is because our state drought law was not made for hyper-scale data centers, and it does not regulate their water usage at all, no matter how bad a drought were to become (Wabe, 2025).
Hyper-scale data centers do not create any significant local employment. Each typically employs between a dozen and a few dozen persons — and often those people are not from the local community, as existing specialists are almost always brought in from other areas. Here are the striking results from a comprehensive review of the effect of all hyper-scale data centers in the state of Texas by a local economist:
"...the causal estimate of net job flows attributable to data centers is zero, while even the gross job flow estimates are very modest – less than 10% of the gross employment effect of a Walmart Supercenter."Coming soon: information on the job landscape at other GA data centers
The numbers Project Ruby is projecting for property tax revenue are absolute overprojections — at 1,614.5% more than what the highest-paying hyper-scale data center in Georgia so far has paid.
| Data Center | Actual Revenue/MW (2024/25) | Applied to Project Ruby |
|---|---|---|
| Switch Atlanta The Keep Campus | $6,174 | $3.7 million |
| Douglas County Google Data Center | $16,154 | $9.7 million |
| QTS Suwanee | $19,510 | $11.7 million |
| QTS Suwanee | $10,722 | $6.4 million |
| Meta Stanton Springs | $8,929 | $5.3 million |
| QTS Fayetteville | $14,689 | $8.8 million |
| QTS Atlanta | $5,956 | $3.6 million |
| QTS Atlanta | $18,765 | $11.3 million |
| Peach | $4,286 | $2.5 million |
| Average for all projects | $12,259 | $7.4 million |
The average hyper-scaler in Georgia pays $12,259 in taxes per megawatt. The exact amount differs based on the precise equipment used in the facility, values, costs, and variations in property tax governance. The absolute highest any facility similar to Project Ruby pays in Georgia is $11.7 million in annual taxes. This amount only depreciates over time, as the value of the equipment itself depreciates.
Question: how does this amount compare to what the city already makes?The average tax contribution scaled to a 600MW-sized data center is $7.4 million/year. Here’s a striking comparison: Columbus makes three times that amount in interest on its own cash balances alone (see FY 2024) – without any of the dangers outlined here.
Instead, hyper-scalers are well known for over-exaggerating to have their projects accepted by local councils — and they are in no way legally bound to their projections. Against any realistic expectation, this is what Project Ruby is projecting:
| Data Center | Estimated Revenue/MW (2029) | Annual Revenue Projection (2030) |
|---|---|---|
| Project Ruby — Muscogee County | $343,500 | $68.7 million |