DIGITAL INFRASTRUCTURE
FACT CHECK

""Data centers will double everyone's electric bills.""

The impact on individual household rates depends on multiple factors including utility size, existing rate base, project scale, and cost allocation methodology. LBNL's 2024 U.S. Data Center Energy Usage Report found that states with the largest data center demand growth generally saw electricity price decreases between 2019 and 2025, though some states saw increases. The primary driver behind electricity price increases during this period was utility investment in aging grid infrastructure, not data centers.

Rate impacts vary dramatically by state regulatory framework, utility financial health, project size relative to customer base, and whether developers fund infrastructure directly or costs are socialized across all customers. States with robust special tariff structures (e.g., Maryland's RELIEF Act, Oklahoma's HB 2992) show minimal rate impact; states without them face greater risk.

Understanding actual rate impact mechanisms helps communities negotiate better terms and policymakers design protective regulations. Alarmist claims (both "rates will skyrocket" and "there will be no impact") oversimplify a complex reality. The right question is not whether data centers affect rates, but whether the regulatory framework ensures they pay their fair share.

• 2024 United States Data Center Energy Usage Report — Lawrence Berkeley National Laboratory
• Data Centers and Their Energy Consumption — Congressional Research Service (R48646)

""Special rate structures for data centers shift costs to regular customers.""

Well-designed special tariffs ensure large-load customers pay the full marginal cost of serving their demand. Poorly designed ones may allow cost shifting. A growing number of states (Maryland, New Jersey, Oklahoma, Virginia) have enacted or are reviewing tariff structures requiring data centers to fund their own infrastructure. The key question is whether the tariff accurately reflects incremental infrastructure costs.

Tariff quality varies by utility and state regulatory oversight. Some jurisdictions have robust cost-of-service analysis requirements; others rely on negotiated agreements with less transparency. AEP Ohio introduced a rate schedule requiring data centers to pay for at least 85% of subscribed energy regardless of actual use.

The debate over special rates is really about transparency and accuracy in cost allocation, not about whether large customers should pay their fair share — they should. The growing number of state laws requiring data centers to fund their own infrastructure shows this is achievable through good policy design.

• Maryland Utility RELIEF Act — Maryland General Assembly (2026)
• Oklahoma Data Center Consumer Ratepayer Protection Act (HB 2992) — Oklahoma Legislature (2026)
• New Jersey Data Center Fair Share Act — New Jersey Legislature (2026)

""Data centers building their own gas plants will raise everyone's energy bills.""

A 2026 Energy Innovation analysis argues that on-site natural gas generation will increase gas prices for all consumers because natural gas is a market-traded commodity — data centers competing for gas supply raise prices. Since gas sets electricity prices in most hours, higher gas prices raise electricity costs. However, this analysis represents one perspective; others note that behind-the-meter generation reduces grid interconnection demand, potentially lowering infrastructure costs for other ratepayers.

The net impact depends on the scale of gas demand, regional gas market dynamics, whether the generation displaces or supplements grid power, and whether the data center funds its own infrastructure. Bloomberg NEF identifies 100 GW of on-site gas capacity planned nationally. The most efficient combined-cycle turbines are back-ordered 5-7 years, leading some developers to use less efficient equipment.

This is a genuine trade-off, not a simple either/or. On-site generation can reduce grid strain and protect ratepayers from infrastructure cost allocation, but it raises questions about gas market competition and air quality. Good policy addresses both sides of the equation.

• Behind-the-Meter Data Center Gas Plants Will Raise US Energy Bills — Energy Innovation Policy & Technology LLC
• Bypassing the Grid: How Data Center Developers Are Building Their Own Power Plants — Cleanview (2026)
• Bloomberg NEF On-Site Gas Analysis — Bloomberg New Energy Finance (2026)

""Data centers use as much water as small cities.""

Water consumption varies enormously by facility. A traditional evaporative-cooled data center can use 1-5 million gallons per day per 100 MW. An air-cooled facility of the same capacity might use less than 50,000 gallons per day. The IEA estimates a 100 MW U.S. data center may consume roughly 530,000 gallons per day on average. In The Dalles, Oregon, Google's data center water use reached 355 million gallons in 2021 — about 29% of the city's total water demand.

Cooling technology choice is the primary variable, followed by climate zone, facility efficiency (PUE/WUE), workload intensity, and water source type (potable vs. reclaimed vs. groundwater). Since 2022, nearly two-thirds of new U.S. data centers have been built in water-stressed areas, according to a Bloomberg News analysis.

Communities should ask specific questions about cooling technology and projected water use rather than accepting generic claims. The claim is true for some facilities and false for others. What matters is the specific facility's design and the local watershed's capacity.

• Data Centers Water Footprint: The Need for More Transparency — AGU (Earth's Future, 2026)
• Data Drain: The Land and Water Impacts of the AI Boom — Lincoln Institute of Land Policy (2025)
• Data Centers and Their Energy Consumption — Congressional Research Service (R48646)

""Modern data centers don't use any water.""

This is generally false. Even air-cooled facilities require some water for humidification, fire suppression, and sanitation. However, modern closed-loop designs can substantially reduce operational water use. Microsoft is rolling out closed-loop water recycling systems starting in 2026 for facilities in Arizona and Wisconsin, each saving an estimated 125 million liters per year. AWS plans to operate 120 data centers with reclaimed water by 2030, potentially saving 530+ million gallons of drinking water annually.

Actual water use ranges from near-zero (for facilities specifically designed for water scarcity using closed-loop systems) to very high (for facilities using traditional evaporative cooling in hot climates). Most new facilities fall somewhere between. The one-time fill requirement for a closed-loop system can still use a meaningful volume of water during construction.

Both extremes ("data centers use no water" and "data centers will drain our aquifers") mislead communities. Accurate technology-specific information enables better planning decisions. The trend toward closed-loop and reclaimed water systems is real and significant, but not universal.

• U.S. Water-Related Expenditures for Data Centers to Exceed $4.1 Billion Through 2030 — Bluefield Research (2025)
• Data Centers Water Footprint — AGU (2026)
• Data Centers Explained: Myths, Facts, and Answers — CoreWeave (2026)

""Data centers use less than 0.05% of total U.S. freshwater withdrawals.""

This figure, cited by the Goldwater Institute, is directionally accurate at a national level — data centers consume a small fraction of total U.S. water use compared to agriculture and thermoelectric power generation. However, national averages obscure local impacts. In Council Bluffs, Iowa, Google's data center consumed roughly 980 million gallons in 2023 — about 21% of the city's total water use. In Newton County, Georgia, a Meta data center uses 500,000 gallons per day — 10% of county consumption.

National-level statistics are misleading because water is inherently local. A data center that represents 0.05% of national water use may represent 20-30% of a single community's water supply. The relevant question is watershed-level impact, not national averages.

Comparing data center water use to national totals or to golf course irrigation deflects from the real question: does this specific facility strain this specific local water supply? Communities need facility-level and watershed-level data, not national averages.

• The Data Center Debate: Fact vs. Fiction — Goldwater Institute (2026)
• Data Centers Water Footprint — AGU (2026)
• Data Drain: The Land and Water Impacts of the AI Boom — Lincoln Institute of Land Policy (2025)

""This project will create 10,000 jobs.""

Job creation claims must distinguish between construction-phase jobs (temporary, typically 18-36 months) and permanent operations jobs (ongoing). Construction of major hyperscale data centers often employs 1,500-5,000 skilled workers over an 18-24 month period. However, a review of more than 1,200 U.S. data centers found that even the largest employ fewer than 150 permanent workers, and sometimes as few as 25.

Actual job counts depend on facility size, automation level, contractor vs. employee status, local labor market conditions, and how "indirect/induced" jobs are calculated (which relies on economic modeling assumptions). One study found that for every direct data center job, 4.5 are created elsewhere in the economy.

Communities negotiating incentive packages need accurate baseline expectations. Promised job numbers that include indirect effects using generous multipliers may overstate actual local hiring by 3-5x compared to direct payroll positions. Construction employment is real and substantial, but temporary.

• From Energy Use to Air Quality: The Many Ways Data Centers Affect US Communities — World Resources Institute (2026)
• Big Tech Promised Jobs. Cities Gave Millions. Where Are the Workers? — Good Jobs First
• The Data Center Debate: Fact vs. Fiction — Goldwater Institute (2026)

""Data centers only create a handful of permanent jobs.""

This understates reality for many projects. While highly automated facilities may have small permanent staffs, larger campuses with multiple buildings, diverse tenants, and on-site maintenance needs can employ hundreds permanently. The average data center salary is approximately $160,000, nearly double the national median household income. In Loudoun County, Virginia, data centers generate roughly 38% of county general fund revenue.

Permanent staffing levels vary from ~25 employees for small single-building facilities to 300+ for large multi-building campuses. Construction employment scales with project capital investment. The broader economic impact — grid upgrades, technology infrastructure attracting other businesses — extends beyond direct employment.

Both over-promising and under-stating job impacts harm community decision-making. The real value proposition of data centers is often tax revenue and infrastructure investment rather than direct employment. Communities should evaluate the full economic picture, not just headcount.

• Data Center Salary Survey — DataXConnect (2024)
• Loudoun County Government Financial Reports — Loudoun County, Virginia
• The Data Center Debate: Fact vs. Fiction — Goldwater Institute (2026)

""Data centers will destroy nearby property values.""

The evidence is genuinely mixed. A University of Rochester study using Virginia air permit data found the effect of data center permitting on nearby home prices was "economically small and slightly positive." A Realtor.com analysis of 43 ZIP codes found no statistically significant difference in home prices near data centers. However, a George Mason University-led study found new data centers slowed local home-price growth. In Granbury, Texas, home prices fell 8.2% over the past year near a data center development.

Impact depends on proximity, facility size, noise levels, visual impact, construction activity, and local real estate market conditions. The University of Rochester study covers data centers built before the current hyperscale era — newer, larger facilities may have different impacts. Long-term effects from noise and electricity prices may take years to materialize.

Neither "data centers destroy property values" nor "data centers have zero impact on property values" is supported by the evidence. Communities should demand project-specific noise studies, setback requirements, and property value protections rather than relying on generalizations from either side.

• Not In My Back Yard! The Effects of Data Centers on Housing Prices — University of Rochester (2026)
• The Data Center Boom Is Headed for American Neighborhoods — Realtor.com Economic Research (2026)
• Data centers emerge as real estate's newest pricing wildcard — HousingWire (2026)

""Data centers are silent facilities that won't bother neighbors.""

This is often false for facilities near residential areas. Data centers emit sounds from cooling systems, diesel generators, and fans that can be heard for hundreds of feet. Industrial-sized diesel generators can reach up to 105 decibels. Residents in Prince William County, Virginia, have complained that data center noise routinely exceeds 60 decibels. At xAI's facility in Southaven, Mississippi, 27 natural gas turbines operate 24/7, disturbing nearby neighbors' sleep.

Noise levels depend on facility design, cooling technology (air-cooled systems tend to be noisier than water-cooled), setback distance from homes, presence of on-site gas turbines, and whether acoustic mitigation measures are installed. Some modern facilities with sound barrier systems report noise levels comparable to a household refrigerator. Most community noise ordinances were written for block parties, not industrial facilities, making enforcement difficult.

Noise is one of the most common community complaints and one of the hardest to address after construction. Communities should require pre-construction acoustic studies, enforceable decibel limits at property lines, setback requirements, and ongoing monitoring commitments before approving projects.

• Communities Are Raising Noise Pollution Concerns About Data Centers — EESI (2026)
• From Energy Use to Air Quality: The Many Ways Data Centers Affect US Communities — World Resources Institute (2026)
• Data Centers Explained: Myths, Facts, and Answers — CoreWeave (2026)

""Data centers will crash the power grid.""

Grid reliability depends on whether generation and transmission capacity keeps pace with load growth. A conventional data center draws as much electricity as 10,000-25,000 households; a hyperscale AI facility can use as much power as 100,000+ homes. LBNL projects data center energy could reach 12% of U.S. electricity consumption by 2028. However, data centers provide steady, predictable baseload demand that can help utilities plan. New grid infrastructure built for data centers becomes permanent regional capacity serving all customers.

Reliability risk varies by region. Areas with grid constraints due to insufficient infrastructure are more likely to experience strain. ERCOT (Texas), PJM (Virginia/Pennsylvania), and other regions with rapid data center growth face the most acute challenges. On-site generation, demand-response programs, and battery storage can reduce grid strain.

Grid reliability is a legitimate concern that requires proactive planning — not alarmism or dismissal. The question is not whether data centers strain the grid, but whether utilities, developers, and regulators are investing in the generation and transmission needed to serve growing demand while maintaining reliability for all customers.

• 2024 United States Data Center Energy Usage Report — Lawrence Berkeley National Laboratory
• Data Centers and Their Energy Consumption — Congressional Research Service (R48646)
• From Energy Use to Air Quality: The Many Ways Data Centers Affect US Communities — World Resources Institute (2026)

""Data centers strengthen the grid by paying for infrastructure upgrades.""

This can be true when developers fund their own infrastructure. Utilities typically require developers to fund substations, transmission upgrades, or dedicated lines at private expense. These improvements often become permanent public infrastructure assets. Many states now require large-load customers to pay for system expansions. However, if a data center closes or uses less electricity than projected, ratepayers may foot the bill for infrastructure built to serve them.

The outcome depends on the specific tariff structure, infrastructure funding agreement, and whether the data center meets its projected load. Well-structured agreements protect ratepayers; poorly structured ones create stranded cost risk. The growing trend toward special tariffs and cost-allocation requirements (Maryland, New Jersey, Oklahoma, Virginia) reflects recognition of this risk.

Both extremes — "data centers crash the grid" and "data centers always strengthen the grid" — are oversimplifications. The right answer is project-specific: Does this developer fund its own infrastructure? Are there clawback provisions if projected load doesn't materialize? Are existing ratepayers protected from stranded costs?

• Data Centers Explained: Myths, Facts, and Answers — CoreWeave (2026)
• The Data Center Debate: Fact vs. Fiction — Goldwater Institute (2026)
• Data Centers and Their Energy Consumption — Congressional Research Service (R48646)

""Data centers are a drain on local government resources.""

The evidence generally points the opposite direction. Data centers produce some of the highest tax-revenue-per-acre ratios of any land use. In Loudoun County, Virginia, data centers generate roughly 38% of county general fund revenue. For every $1 in public services consumed, data centers pay approximately $26 in taxes — six times more than traditional manufacturing. However, Maine enacted LD 713 in 2026 prohibiting data centers from state tax incentive programs, signaling that not all states view data centers as uniformly beneficial.

Fiscal impact depends on the tax structure, whether tax incentives are offered, the cost of public services required, and whether the host community or a broader region captures the tax revenue. Communities that offer aggressive tax abatements may see reduced fiscal benefit. The tax revenue benefits the host community while infrastructure costs may be spread across the utility service area.

Data centers can be significant fiscal contributors, but the benefits depend on the deal structure. Communities should evaluate whether incentive packages erode the fiscal benefit, whether tax revenue is committed to long-term community needs (schools, infrastructure), and whether the fiscal benefit is sustainable if the facility closes.

• The Data Center Debate: Fact vs. Fiction — Goldwater Institute (2026)
• Loudoun County Government Financial Reports — Loudoun County, Virginia
• Maine LD 713 — Maine Legislature (2026)

""Data centers are clean, green facilities with minimal environmental impact.""

This overstates reality. About half of the electricity used by U.S. data centers comes from fossil fuel power plants, which themselves use significant water for steam generation. On-site natural gas turbines produce air emissions. Data centers cover hundreds of acres with impermeable surfaces, removing land from agriculture or nature. However, leading operators are investing in renewable energy, closed-loop water systems, and water replenishment projects. The environmental footprint varies enormously by facility design and energy source.

Environmental impact depends on the electricity source (grid mix vs. on-site gas vs. contracted renewables), cooling technology, land use, and whether the operator invests in mitigation. A data center powered by on-site gas turbines has a very different environmental profile from one powered by contracted solar and wind with closed-loop cooling.

Environmental claims should be evaluated facility-by-facility, not industry-wide. Communities should ask about the specific energy source, cooling technology, emissions profile, land use impact, and any environmental mitigation commitments. Both "data centers are green" and "data centers are environmental disasters" are oversimplifications.

• Data Drain: The Land and Water Impacts of the AI Boom — Lincoln Institute of Land Policy (2025)
• Data Centers Water Footprint — AGU (2026)
• From Energy Use to Air Quality: The Many Ways Data Centers Affect US Communities — World Resources Institute (2026)