You Better Start Believing in Datacenters, You're on One
There’s a famous line in Pirates of the Caribbean where Captain Barbossa tells Elizabeth Swann, “You better start believing in ghost stories, Miss Turner. You’re in one.”
Something similar is true of data centers. You better start believing in them, because you’re already living inside one. Every time you stream a show, check a bank balance, use maps, send an email, or- yes; ask an AI a question, you’re relying on racks of servers sitting in large, windowless buildings that most of us never see. Those buildings have quietly been essential infrastructure for decades. And right now a lot of communities are trying hard to keep more of them from being built.
This resistance is real, growing, and surprisingly bipartisan. It’s worth taking seriously; but it’s also worth putting in context before we decide the answer is simply “no more.”
You’re already on one
For most of the internet’s history, data centers have been background infrastructure. They stored email, hosted websites, and ran the cloud services that companies quietly depended on. The AI boom merely changed the scale; training and running large models requires far more concentrated computing power, which means far more electricity and cooling. The result is a rapid build-out of much larger facilities.
This demand isn’t optional, however- as detractors would have you believe. The same tools people use every day including search, translation, coding assistants, medical imaging analysis, and logistics optimization all run on this infrastructure. Refusing to build the capacity doesn’t eliminate the need though, it just makes the services more expensive, slower, or more dependent on someone else’s grid. Do we really want to offload the most critical functions of our global networks and data systems to foreign countries? Do you want your medical records and email living in a datacenter in China?
The backlash is real
Local opposition has moved from scattered complaints to an organized force. Polls in 2026 have shown large majorities of Americans would rather not have a data center near their home. In the first quarter of this year alone, trackers reported roughly $130 billion worth of projects delayed or blocked. Dozens of counties and towns have passed temporary or permanent pauses and now New York became the first state to impose a broad moratorium.
The concerns people raise are understandable. These facilities draw a lot of power. In places where the grid is already tight, that can put pressure on rates for everyone else, and some designs use substantial amounts of water for cooling, which matters in dry regions. The buildings are large, the cooling systems can be noisy, and construction can change the character of rural or semi-rural land. Diesel generators for backup power raise air-quality worries. And the permanent job counts are often modest relative to the capital invested and the tax incentives sometimes offered.
None of that is imaginary. When a community sees a multi-hundred-megawatt project proposed nearby with limited transparency about water or power impacts, pushback is rational.
Putting the numbers in perspective
Scale still matters. Data centers currently account for a mid-single-digit percentage of U.S. electricity use, with projections that the share will rise meaningfully by 2030. That is a real planning challenge for utilities and grid operators. It is not however, by itself, an argument that the underlying demand is illegitimate. The same growth is already driving new generation projects, including nuclear restarts and expansions in some regions.
Water use follows a similar pattern. Direct consumption by data centers is measurable and can be significant at the local level, especially with evaporative cooling. Nationally it remains a small fraction of agricultural irrigation and is often lower than the water used by golf courses. Closed-loop systems, air cooling, and recycled water can shrink the footprint further. The right response to water stress is better design and siting rules, not treating every project as an existential threat to the aquifer.
The climate framing has shifted
For years, the most extreme climate scenarios—those projecting 4–5°C or more of warming under “business as usual” shaped a great deal of public urgency and policy language. Those pathways have now been set aside by the researchers who develop the official scenarios feeding the next round of major assessments. They were judged implausible given actual emissions trends and energy system changes- and worse still, entral estimates under current policies sit lower, in a range that is still serious but no longer apocalyptic in the way earlier messaging often suggested.
This matters for the data-center debate because a lot of the broader environmental energy that once justified maximalist opposition to large infrastructure projects has lost some of its force. Local impacts on land, water, and power rates remain legitimate topics. Treating every data center as a climate emergency does not follow as cleanly as it once seemed to.
What host communities actually get
The fiscal picture is not one-sided. In Loudoun County, Virginia, home to the densest concentration of data centers in the country, the facilities generate on the order of $1.3 billion in annual tax revenue in recent budget projections, a large share of local funding. That revenue has helped the county cut residential property tax rates substantially over the past decade while expanding schools, public safety, and recreation. Other jurisdictions have seen similar patterns: construction employment is substantial, the tax base from servers and buildings can be transformative, and the land footprint is often modest relative to the revenue produced. There are real trade-offs. Permanent operational jobs per facility are limited. Poorly structured incentives can shift costs onto other ratepayers. Transparency about water and power needs has too often been lacking. Those are arguments for better deals and clearer rules, not for freezing development.
A better path than “just say no”
Opposition to new technology has a long history of mixing legitimate local grievances with broader discomfort about change. Data centers are large, industrial, and somewhat mysterious to people who don’t work in the industry. It is easy for them to become a proxy for worries about AI, Big Tech, rising bills, or the feeling that decisions are being made somewhere else.
Blanket moratoriums solve the local political problem at the cost of capacity the broader economy still needs. The more useful approach is narrower and more practical: require transparent disclosure of power and water needs, prefer designs that minimize both, co-locate with new generation where possible, enforce noise and visual standards, and structure tax arrangements so that host communities capture clear net benefits. None of that requires pretending the facilities are optional.
We are going to need more computing capacity, more electricity, and more careful planning about where both go. The alternative is not a quieter, greener status quo. It is slower services, higher costs, and a growing gap between the places willing to build and the places that are not.
You don’t have to love the look of a data center. You do have to recognize that the digital world you already inhabit runs on them. Starting from that fact makes the conversation about how and where to build them a lot more productive than treating every project as a fight to be won or lost.
