Hidden Costs of AI

Where is AI?Coming to your Backyard

AI energy consumption is doubling every few years. Every AI query, every model training run, every generated image draws power and water from real places — the cities and towns where we all live. This page breaks down the data center environmental impact: how much energy AI uses, how much water data centers consume, and why the carbon footprint of artificial intelligence is growing faster than most people realize.

Is a data center being planned near you?

Find data center locations using the site developed by Erin Brockovich.

Data Center Map
Take Action

What You Can Do

Individual action matters, especially when it's organized and directed at the right targets.

Know Your Local Data Centers

Find out if there are data centers planned or operating in your community. Local planning applications are public record.

Contact Elected Officials

Demand that your local and national representatives require environmental impact assessments before approving new data center construction.

Support Affected Communities

Amplify the voices of communities already dealing with data center impacts. Their stories rarely reach national media.

Demand Corporate Transparency

Ask the AI companies you use to publish verified, real-time data on energy, water, and carbon, not just annual sustainability reports.

The Scale of the Problem

These AI energy consumption statistics are not projections. They are current, documented, and growing faster than any previous technology transition.

8-25%

increase in residential electricity rates in regions with heavy AI data center concentration

AI electricity demand projected to drive 3-4% annual rate hikes through 2030

500B

gallons of water consumed annually by US data centers — a key data center environmental impact statistic

Equivalent to the water use of 5 million American households

48%

increase in Google's carbon emissions since 2019, a widely cited AI energy consumption statistic

Directly attributed to AI data center infrastructure growth

21%

of Ireland's total national electricity consumed by data centers, a global AI power consumption benchmark

Projected to reach 32% by 2030 per EirGrid

The Infrastructure Behind AI

What Actually Happens When You Use AI

When you send a message to ChatGPT, ask Gemini a question, or generate an image, your request travels to a data center — a warehouse-scale building filled with tens of thousands of specialized chips running at full power, 24 hours a day. AI power consumption per query may seem invisible, but the infrastructure behind it is massive and growing.

These facilities are not like ordinary office buildings. A single hyperscale data center can draw as much electricity as a small city — 10 to 50 megawatts of continuous demand. The chips generate enormous heat that must be continuously removed — either by blowing air through the building or by evaporating millions of gallons of water. This is where data center water usage becomes a critical environmental concern: a single facility can consume 1 to 5 million gallons per day.

Training a large AI model — the process that happens before you ever use it — is even more intensive. A single training run for a frontier model can consume as much electricity as thousands of homes use in a year, and as much water as a small town uses in a month. The carbon footprint of AI is not abstract — it shows up in rising emissions reports from Google, Microsoft, and Amazon.

10–50 MW

Typical power draw of a single hyperscale data center - equivalent to 8,000–40,000 homes

Local Consequences

Six Ways Data Centers Harm Communities

These are not abstract environmental concerns. These data center environmental impacts are documented harms affecting real communities right now — and they are accelerating as AI energy consumption grows.

On the Ground

Three Communities at the Frontline

These are not worst-case scenarios. They are current, documented situations that illustrate the data center environmental impact when AI infrastructure expands without adequate oversight.

Northern Virginia, USA

The World's Largest Data Center Cluster

Loudoun County, Virginia hosts more data center square footage than any other place on Earth. The region's grid operator, PJM, has warned that planned AI campuses will require the equivalent of 40 new power plants by 2030. Local electricity rates have risen sharply, and the county has imposed temporary moratoriums on new construction while infrastructure catches up. Residents report that the economic benefits - primarily property tax revenue - are offset by traffic, noise, and the loss of rural character.

Ireland

A Nation's Grid Under Siege

Ireland has become a cautionary tale for data center policy. With some of the most permissive planning rules in Europe and a strategic location for transatlantic connectivity, the country attracted a massive concentration of hyperscale facilities. Data centers now consume over a fifth of national electricity - more than all urban households combined. EirGrid, the national grid operator, warned that continued growth threatens supply security. The Irish government has since imposed a moratorium on new data center connections in the Dublin region.

Phoenix, Arizona, USA

AI Meets the Water Crisis

The Phoenix metropolitan area is one of the fastest-growing data center markets in the world, and one of the most water-stressed. The Colorado River, which supplies much of the region's water, is at historically low levels. Yet data center construction continues at pace, with facilities permitted to draw millions of gallons per day from the same aquifer system that supplies residential wells. Farmers in the region have reported declining water tables, and Indigenous communities downstream have raised concerns about treaty water rights being undermined by tech industry demand.

48%

Increase in Google's emissions since 2019, directly attributed to AI data center growth

Carbon Footprint of AI

The Climate Pledges That Didn't Survive AI

In recent years, Google, Microsoft, and Amazon each made sweeping commitments to reach net-zero emissions. These pledges were celebrated as proof that the tech industry could lead on climate. Then AI happened.

Google's recent environmental report showed a substantial increase in greenhouse gas emissions since 2019. Microsoft's rose 29%. Amazon's absolute emissions increased for the third consecutive year. All three companies attributed the increases directly to data center growth driven by AI demand.

Google
+48%Recent years
Microsoft
+29%Recent years
Amazon AWS
+3 consecutive yearsRecent years
A Path Forward

What Responsible AI Infrastructure Looks Like

The problem is not that data centers exist - it's that they are being built without adequate safeguards. These are the standards that should be required.

Policy

Genuine Renewable Energy

Require data centers to match consumption with real-time renewable generation, not paper certificates, and prioritize siting near existing clean energy sources.

Technology

Water-Efficient Cooling

Mandate air-cooled or closed-loop cooling systems in water-stressed regions. Prohibit evaporative cooling in areas under drought restrictions.

Governance

Community Benefit Agreements

Require binding agreements that deliver local jobs, infrastructure investment, and utility bill relief to communities hosting data center campuses.

Policy

Mandatory Transparency

Require public disclosure of energy, water, and carbon data for all data centers above a threshold size - with independent verification.

Regulation

Efficiency Standards

Set minimum Power Usage Effectiveness (PUE) and Water Usage Effectiveness (WUE) standards for new construction, with phase-in requirements for existing facilities.

Regulation

Strategic Siting Requirements

Prohibit data center construction in water-stressed regions, near residential areas without noise mitigation, and on prime agricultural land without environmental review.

Stop the Infrastructure of Unchecked AI

How to Stop a Data Center From Being Built

AI's power grab requires massive physical infrastructure — data centers that consume water, electricity, and land on a staggering scale. The AI energy consumption statistics are clear: data center environmental impact is accelerating. These facilities are approved one zoning board at a time. That is also how they can be stopped.

2–5M

gallons of water per day

consumed by a single large data center

100MW+

of electricity demand

per facility, equal to 80,000 homes

1,000+

new data centers

approved in the U.S. over the past 3 years

78%

of approved data center permits were never publicly opposed

Show Up to Zoning Hearings

Local Government

Data centers live and die on zoning approvals and conditional use permits. A packed hearing room full of opposed residents sends an unmistakable signal to planning commissioners, and it creates a public record that can be used in legal challenges.

  • Search your county's planning portal for pending data center applications
  • Sign up to speak during the public comment period, even 2 minutes on record matters
  • Bring neighbors: a crowd of 30 carries far more weight than a single objection
  • Ask commissioners to require an independent environmental impact study before any vote

5M+

gallons of water consumed daily by a single large AI data center

Make Water Usage the Headline

Environmental Pressure

A single large data center can consume millions of gallons of water per day for cooling. In drought-stressed regions, this is a winning argument with local officials, farmers, and conservation groups who would not otherwise engage on AI policy.

  • Request the applicant's projected daily water draw from the planning department
  • Contact your local water utility and ask how the facility will affect municipal supply
  • Connect with agricultural associations and conservation districts who share this concern
  • Submit formal comments demanding a water-use impact study before permits are granted

40%

of all new U.S. electricity demand projected to come from data centers by 2030

Challenge the Energy Grid Impact

Utility Pressure

Data centers are the single fastest-growing source of electricity demand in the United States. Many utilities are quietly planning coal and gas plant extensions to meet this demand, creating an opening for ratepayer and clean-energy advocates.

  • File a public comment with your state Public Utilities Commission on data center load forecasts
  • Contact your state legislators on the energy or utilities committee
  • Ask your utility provider how data center load growth will affect residential electricity rates
  • Support Integrated Resource Plan (IRP) processes that cap data center energy priority

3x

more likely to stop a permit when 3+ community groups formally oppose it

Build a Community Coalition

Organizing

Isolated objections are easily dismissed. A coalition that spans neighborhood groups, farmers, teachers, environmental groups, and local businesses is politically impossible to ignore, especially in local elections where margins are small.

  • Connect with existing neighborhood associations and environmental groups in your area
  • Create a shared online space (group chat, mailing list) for coordinated action
  • Identify elected officials up for re-election who rely on your coalition's vote
  • Draft a joint community letter and circulate it for signatures before the next hearing

Local news

coverage doubles public hearing attendance on average

Get It In the Local Press

Media Strategy

Local journalists are hungry for stories about corporate interests conflicting with community needs. A well-placed news story creates political pressure that no lobbying budget can fully neutralize, and it recruits new allies who only learned about the project after reading about it.

  • Write a concise, fact-based press release and send it to local reporters covering planning or environment
  • Offer a spokesperson and residents willing to go on record with personal stakes in the issue
  • Highlight contrasts: gallons of water used vs. local drought restrictions, energy costs vs. rate hikes
  • Follow up after the hearing with a news hook: 'Community demands vote delay pending full impact review'

NEPA review

challenges delay projects by an average of 18–24 months

Use Legal & Regulatory Tools

Legal Action

Environmental review laws, wetlands protections, noise ordinances, and state utility regulations all create formal channels to challenge data center siting. You do not need to win a lawsuit, filing a challenge delays the project and raises its cost enough to matter.

  • Request a full Environmental Impact Statement under NEPA or your state equivalent before any permit
  • Contact an environmental law clinic or the Earthjustice regional office about standing to challenge
  • File formal objections with your state environmental agency citing water, air, and noise impacts
  • Demand full public disclosure of the applicant's utility agreements and grid interconnection plans

AI Energy Consumption Is Not Invisible

The data center environmental impact is measurable, documented, and growing. Every AI product has a physical footprint. The communities bearing that footprint deserve a say in how it's managed. Demanding transparency on AI energy consumption and accountability from AI companies is not anti-technology — it's pro-community.