Data Centers, Communities, and the Mirror We Don’t Always Want to Look Into

Across the U.S., more data centers are being proposed, permitted, and built, and many communities are pushing back. Their concerns are real: noise, land use, water consumption, the strain on local infrastructure, and the simple fact that these massive facilities rarely feel like “good neighbors.”

At the same time, generative AI use is skyrocketing. And whether we like it or not, the two conversations are directly linked.

The Orange Peel Collaborative works at the intersection of people and technology every day, and here’s the truth we don’t talk about enough: we want the benefits of AI, but not always the infrastructure required to run it.

It’s time for a more honest conversation: one that includes real numbers, real tradeoffs, and a real look in the mirror.

What Data Centers Actually Consume: The Facts (Not the Fear)

Data centers are energy-intensive, but the scale is often misunderstood. Let’s anchor the conversation with current, reputable statistics:

• U.S. data centers currently account for ~2–3% of national electricity use. (U.S. Department of Energy)

• Globally, data centers consumed roughly 460 terawatt-hours (TWh) in 2022 which is equivalent to the entire electricity consumption of France. (International Energy Agency)

• One large data center can use between 20–50 megawatts (MW): roughly the electricity needed to power 15,000–40,000 homes.

• Cooling alone can represent 40% of a data center’s total energy usage.

• Water use varies widely, but some facilities consume 3–5 million gallons of water per day, depending on cooling design and climate. (Uptime Institute; NRDC)

These numbers don’t exist in a vacuum: they exist in someone’s community; and that’s where tension grows: we want reliable AI tools and digital convenience, but we don’t want the local impacts. Not near our homes, not on our farmland, not on our water tables. We want data centers…just not here.

Europe’s Energy Offset Models: A Glimpse Into the Possible Future

Europe is ahead of the U.S. in integrating data centers into a cleaner, more circular energy system:

• Denmark and Finland already pipe waste heat from data centers into district heating networks, warming homes and public buildings.

• The Netherlands capped data-center growth until companies agreed to stricter sustainability and community-benefit requirements.

• Ireland temporarily halted new data center approvals to protect grid stability.

These models use data centers to generate community benefits, not just community burdens.

And yes, these kinds of energy-offset systems could be part of the U.S. future. But realistically, we’re a few years out. We don’t yet have the infrastructure, regulations, or public-private alignment to make it standard. A pervasive “cowboy” approach to AI regulation keeps the US perilously perched between innovation and serious environmental and economic consequences.

Let’s Be Honest: AI Use Is Driving Demand. Our Use.

It’s tempting to talk about data centers like they’re an abstract corporate problem. Something “big tech” is doing. But the demand is coming from us as well.From the constant prompting, testing, querying, and generative content creation we now take for granted. And it’s not just professionals. It’s students, hobbyists, creators, and everyday users asking:

• “Write this email.”

• “What should I eat for dinner?”

• “Tell me my horoscope?”

• “Make me 100 variations of this logo.”

• “Explain this like I’m five.”

LLMs like ChatGPT were never designed to replace search engines or function as a digital magic 8-ball. The convenience is intoxicating, and we’re using them far more, and far more casually, than the technology was originally intended for. OpenAI’s Sam Altman has pointed out the difference of using his company’s tech for efficiency versus replacing human decision making- and definitely not as a search engine like Google.

Each seemingly insignificant query carries a real cost: energy, water, and physical infrastructure. So yes, corporations play a role. Governments play a role. Regulators play a role. But individual and collective user behavior is part of the demand signal. It’s uncomfortable, but it’s true.

The “Recycling Paradox”: Individual Choices vs. Systemic Change

There’s a familiar frustration here. It feels a bit like recycling your aluminum cans while billionaires fly private jets daily. Why should your choices matter when the scale seems so mismatched? But the recycling paradox taught us something important: individual action matters most when it’s paired with systemic change.

We need a similar shift with AI use:

• more intentional prompting

• less frivolous querying

• better design from tech companies

• stronger regulation

• infrastructure that reciprocates community benefit

None of this works alone. But together, it looks a lot like a future worth building toward.

So What Do We Do Now?

We need to have a more pragmatic and honest conversation around data centers and generative AI use. The Orange Peel’s approach is actionable steps, in straightforward language. Let’s start here: 

1. Use AI with intention.

Ask whether each query actually requires a large model, or whether a smaller model, a search engine, or your own thinking is enough.

2. Support community-benefit requirements for new data centers.

Cities should be negotiating for:

• local job pipelines

• energy-offset agreements

• green cooling technologies

• community investment funds

• transparent environmental impact reporting

Fun fact: data center companies are often open to these conversations but while people are venting in a Reddit forum, projects move forward before meaningful conversations can be had.

3. Push for U.S. adoption of European-style heat-recapture and grid-stabilizing models.

It works. We just need policy and incentives to catch up.

4. Hold tech companies accountable for building lighter, more efficient models.

The more the industry optimizes, the fewer megawatts we need. From Nvidia to Musk’s various projects to Google- the conversation is nuanced and everyone must shift how we innovate. (Again- we are learning from the recycling model- change is needed at every level.)

5. Acknowledge our role.

We can call out “big tech” all day long. But we also have to be honest about our own usage patterns. You know what to make for dinner. ChatGPT doesn’t need to tell you that.

If We Want a More Sustainable Future, We Have to Want It Enough to Change

Data centers are not disappearing. AI is not disappearing. The question is whether we keep building infrastructure that communities resent or whether we evolve into a system where innovation and responsibility can coexist.Other countries are already building that future. The U.S. can too, but only if we’re willing to treat sustainability as a shared responsibility, not a checkbox or a talking point.

If we want a future where data centers benefit communities, not burden them, then the shift has to start with all of us. From city leaders and tech companies to everyday AI users hitting “enter” on prompts that feel small but add up fast. A more conscious, community-minded future is absolutely possible. But only if we want it to be.

Photo credit: Freepik