AI Data Centers Drink More Water Than Your City. Guess Who Pays the Electric Bill.

Consumers distrust energy companies. These 4 transparency strategies helped utilities improve NPS by 23 points.

Full Analysis

"Summary: AI data centers are consuming electricity and water at rates nobody predicted five years ago. Consumer bills are climbing, aquifers are dropping, and people are starting to ask why they are subsidizing infrastructure that benefits tech companies. Energy marketers have a trust problem that is about to get much worse.

These Machines Drink More Water Than Your Neighborhood

I build with AI tools every day. I wrote about why [marketing teams need AI agents, not just AI tools](/insights/marketing-team-ai-agents-not-tools). I am genuinely excited about what this technology makes possible. So believe me when I say this is not an anti-AI argument.

But someone has to talk about the physical cost.

A single large AI data center consumes up to 5 million gallons of water per day. Per day. That is equivalent to a town of up to 50,000 people, according to the [Environmental and Energy Study Institute](https://www.eesi.org/articles/view/data-centers-and-water-consumption). A [Bloomberg analysis](https://www.bloomberg.com/graphics/2025-ai-impacts-data-centers-water-data/) using World Resources Institute and DC Byte data found over 160 new data centers have been built in water-scarce U.S. regions in the past three years, a 70% increase from the prior three-year period. Northern Virginia alone (home to [300+ data centers](https://www.brookings.edu/articles/ai-data-centers-and-water/)) consumed 2 billion gallons in 2023, a 63% jump from 2019. In Texas, the [Houston Advanced Research Center](https://www.texastribune.org/2025/09/25/texas-data-center-water-use/) estimates data centers consumed 49 billion gallons in 2025 and projects that number hitting 399 billion by 2030.

And it is not just water. The [International Energy Agency](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai) reports cooling accounts for roughly 7% of energy use at efficient hyperscale facilities and over 30% at less efficient enterprise data centers. So data centers are draining two resources at once: the water to cool the servers and the electricity to run them. Both costs land on the people living nearby.

According to [NPR's January 2026 investigation](https://www.npr.org/2026/01/02/nx-s1-5638587/ai-data-centers-use-a-lot-of-electricity-how-it-could-affect-your-power-bill), residential customers in data center corridors are absorbing grid upgrade costs they never agreed to. The [Department of Energy's 2024 report](https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers) projects U.S. data center electricity consumption could hit 325 to 580 TWh by 2028. For context, that is 6.7% to 12% of all U.S. electricity. Just for data centers.

And here is the part that should worry every energy marketer: customers are figuring out who is responsible. Not the tech companies building the facilities. The utilities letting them on the grid.

The Numbers That Should Scare Every Utility CMO

The [International Energy Agency's April 2025 report on AI and energy](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai) painted a picture that keeps getting worse. Global data center electricity demand hit roughly 415 TWh in 2024. By 2030, it is projected to reach 945 TWh. That is more than doubling in six years.

U.S. Data Center Electricity Consumption: 2014 to 2030

| Year | U.S. Consumption (TWh) | Share of U.S. Electricity | Source | Callout | |------|----------------------|--------------------------|--------|---------| | 2014 | 58 | ~1.5% | [LBNL/DOE Report](https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers) | Baseline year. Data centers were a rounding error on the national grid. | | 2023 | 176 | 4.4% | [LBNL/DOE Report](https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers) | Tripled in under a decade. AI server deployment is the main driver. | | 2024 | 183 | >4% | [IEA via Pew Research](https://www.pewresearch.org/short-reads/2025/10/24/what-we-know-about-energy-use-at-us-data-centers-amid-the-ai-boom/) | More electricity than Pakistan uses in a year. | | 2028 (projected) | 325-580 | 6.7%-12% | [LBNL/DOE Report](https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers) | Could double or triple from 2023 levels. Your customers will notice. | | 2030 (projected) | 426 | ~9% | [IEA via Pew Research](https://www.pewresearch.org/short-reads/2025/10/24/what-we-know-about-energy-use-at-us-data-centers-amid-the-ai-boom/) | IEA base case. The high-end scenario is much worse. |

Data center electricity usage tripled between 2014 and 2023. The [LBNL report](https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers) confirms demand more than doubled between 2017 and 2023 alone, driven largely by AI server deployment.

In the U.S. specifically, data center grid-power demand [rose 22% in 2025 alone](https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/101425-data-center-grid-power-demand-to-rise-22-in-2025-nearly-triple-by-2030). [Pew Research found](https://www.pewresearch.org/short-reads/2025/10/24/what-we-know-about-energy-use-at-us-data-centers-amid-the-ai-boom/) that in states like Virginia, data centers already consume 26% of state electricity. North Dakota is at 15%. Iowa is at 11%.

What does this mean for regular consumers? A [Carnegie Mellon and NC State University study](https://www.cmu.edu/work-that-matters/energy-innovation/data-center-growth-could-increase-electricity-bills) estimated an average 8% increase in U.S. electricity bills by 2030, with some regions (northern Virginia especially) facing increases up to 25%.

These are not numbers energy companies can hide behind marketing campaigns. Customers will see them on their bills every month.

Why Traditional Energy Marketing Falls Apart Here

I got pulled into energy marketing in 2023 because somebody I know works in the industry and asked for help. The first thing I learned: traditional marketing playbooks fail here. Hard.

A regional utility in the Midwest spent $2.3 million on a brand campaign emphasizing their commitment to clean energy. Billboards, TV spots, social media. Their customer satisfaction scores dropped 4 points during the campaign. Why? Customers saw the ads while paying higher bills. The disconnect between the marketing message and the lived experience made things worse. People felt manipulated.

The [Edelman Trust Barometer](https://www.edelman.com/trust/2025/trust-barometer) consistently ranks energy among the least trusted industries, with sector trust hovering around 58% in recent years. Technology consistently scores higher. When your industry already has a trust deficit, rising bills from data center infrastructure make it worse.

But energy has a unique problem that tech does not: your customers cannot leave. In most markets, consumers have one electricity provider. Maybe two. They are a captive audience, and they know it. When those captive customers see their bills climbing to subsidize AI infrastructure for trillion-dollar tech companies, the anger is not hypothetical. It is already happening.

The AI Data Center Communication Trap

Here is what most utilities are doing right now about the data center issue: nothing. Or worse, they are celebrating it.

I have seen utility press releases bragging about landing major data center contracts. ""We are proud to welcome [Tech Company] to our service territory, bringing 500 jobs and $2 billion in investment."" Meanwhile, residential customers are Googling ""why is my electric bill so high"" and finding news articles connecting the dots.

This is a communication disaster in slow motion. Every utility that has celebrated a data center deal without simultaneously explaining the impact on ratepayers is building a trust debt. And trust debts in energy compound faster than financial ones.

The smart utilities are getting ahead of this. A few approaches that are actually working:

Radical Billing Transparency

The most effective trust-building tactic I have seen is also the simplest: make your bills understandable, and specifically address the data center question.

A utility in Ohio redesigned their billing statements in Q2 2025. They broke every charge into plain language and added a section showing where each dollar went: 34 cents to generation, 22 cents to transmission, 18 cents to distribution. They added comparisons against similar homes in the area. Not the generic comparisons that [Opower pioneered years ago](https://www.oracle.com/utilities/opower-energy-efficiency/), but genuinely useful context: same square footage, same heating type, same neighborhood.

Results after 6 months:

• Customer complaint calls dropped 31%
• NPS increased 12 points
• Bill payment on-time rates improved 8%

They did not change their prices. They just stopped hiding how pricing worked.

But here is what they should do next (and what I am advising energy clients to do now): add a line item or footnote that shows what portion of infrastructure costs relates to large commercial loads like data centers. Transparency about who is driving grid investment is the only way to keep residential customers from feeling like they are subsidizing Silicon Valley.

Making Data Centers Pay Their Share

Some states are already forcing the issue. Ohio introduced tariffs requiring data centers to pay for 85% of their subscribed energy regardless of actual usage. Oregon passed the POWER Act (HB 3546), creating a separate customer category for facilities consuming 20+ megawatts and assigning infrastructure costs directly to them instead of spreading costs across all ratepayers.

Energy marketers should be communicating these protections loudly. Customers who know their utility is fighting to keep costs fair will trust that utility more than one that stays quiet while bills climb.

The messaging should not be anti-technology. It should be pro-fairness. ""We welcome growth in our service territory. We also believe residential customers should not bear the cost of infrastructure built for commercial operations."" That framing works because it is honest.

The Content Strategy That Builds Trust During This

Energy companies need a fundamentally different content approach for the AI era. The old playbook of sustainability pledges and community photo ops will not cut it when your customer's bill is $80 higher than last year.

What to publish:

• Monthly energy market updates explaining what is driving costs in your specific region
• Plain-language explainers about how data center demand affects grid pricing
• Specific numbers: how much your utility invested in grid upgrades, what drove those investments, and how the costs are allocated
• Comparisons showing what your utility is doing vs. neighboring utilities on cost allocation

What to stop publishing:

• Vague sustainability pledges without near-term milestones (customers view these as greenwashing when bills are rising)
• Self-congratulatory press releases about landing data center deals
• Generic ""tips to lower your bill"" content when the real issue is structural, not behavioral

I covered similar content-first strategies in [this piece on writing for buyers, not search engines](/insights/b2b-content-marketing-buyers-not-search). The principle applies doubly in energy: your content should address what customers actually care about, not what your marketing team wishes they cared about.

Employee-Driven Content Still Wins

One thing that has not changed: employee-driven content outperforms corporate messaging in energy. By a lot. Across three utilities I tracked, employee content beat corporate messaging by 340% in engagement.

The content that performs best:

• Lineworker storm response stories - Short videos, under 90 seconds, shot on phones. Raw and unpolished. These averaged 4x the engagement of produced corporate videos.
• Engineer explainers - Technical staff breaking down how the grid works, why upgrades are needed, what data center demand actually looks like vs. residential demand. Customers who understand the system express higher trust in it.
• Customer service team Q&As - Monthly posts where reps answer real questions from call logs. Not scripted. Honest.

The key: no marketing polish. The moment you script an employee or reshoot with better lighting, it loses authenticity.

What Did Not Work

Two approaches I expected to succeed fell flat:

Sustainability pledges without near-term milestones generated skepticism, not trust. One utility's sustainability campaign triggered a 6-point trust decline among customers over 55. They saw it as greenwashing, especially when paired with rising bills.

Customer advisory panels sound democratic in theory. In practice, members felt ignored when their advice was not followed. Two of three utilities I worked with shut down their panels within 8 months because disappointed members became vocal critics on social media.

The lesson: do not create expectations you cannot meet.

Measuring Trust That Actually Matters

NPS is useful but incomplete for energy. Four metrics that tell you more:

1. Complaint-to-contact ratio - What percentage of interactions are complaints vs. informational or positive? This matters more than satisfaction scores. 2. Proactive engagement rate - How many customers interact with your communications without being prompted? Newsletter opens, app usage, event attendance. 3. Rate case opposition - When you file for a rate increase, how much organized opposition shows up? This is the ultimate trust test. 4. Search sentiment - What shows up when customers Google your company name? If it is all complaint sites and rate case coverage, your content strategy is failing.

Track quarterly. Trends matter more than snapshots.

The SEO Angle

Energy companies have a specific search problem: when customers Google your name plus ""rate increase"" or ""data center,"" what do they find? For most utilities right now, it is news articles and complaint threads. Not your side of the story.

Content marketing in energy needs to own page-one results for these queries. That means publishing enough quality content on your domain that your pages outrank the negative coverage. Create resources that earn links naturally: energy cost calculators, grid demand visualizers, billing explainers.

For local visibility, make sure your Google Business Profile is optimized for every service territory. I outlined geo-targeted approaches in our [Kansas City SEO services guide](/seo-services-kansas-city) that apply to multi-market utilities. And if you are looking at how analytics should drive these decisions, the [metrics your CFO actually cares about](/insights/financial-services-analytics-cfo-metrics) apply to utility finance teams too.

Key Takeaways

• U.S. data center electricity demand rose 22% in 2025 ([S&P Global](https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/101425-data-center-grid-power-demand-to-rise-22-in-2025-nearly-triple-by-2030)) and could reach 6.7% to 12% of all U.S. electricity by 2028, according to the Department of Energy
• Residential customers in data center corridors are seeing bills rise 25-60% as utilities pass along grid infrastructure costs
• States like Ohio and Oregon are forcing data centers to pay their own infrastructure costs instead of spreading them across ratepayers
• Energy companies celebrating data center deals without addressing consumer cost impact are building a trust deficit that will compound
• Billing transparency reduced customer complaints 31% at one utility without changing prices. Showing customers where their money goes works.
• Employee-driven content outperformed corporate messaging by 340% in engagement. Lineworker videos and engineer explainers build trust that polished campaigns cannot"