Data Centers in North Carolina: What We Fear, What the Engineering Shows, and What We Should Do About It
The gap between public perception and technical reality is wide enough to park a server rack in.
A new data center proposal lands in your county, and within a week the conversation has sorted everyone you know into one of two camps: the people convinced it will drain the aquifer and crash the grid, and the people convinced it is the economic salvation of the region.
A note on where I am coming from, because credibility matters. I spent over 14 years at the National Security Agency. After that, more than a decade in marketing technology, the industry behind the digital advertising and analytics that shape much of modern commerce. For the past several years, I have been working with North Carolina businesses, from family-run shops to large enterprises, helping them put artificial intelligence to work in ways that are practical and responsible.
I am not paid by any data center operator. Not a utility. Not a tech giant. Not an environmental group. Not a chamber of commerce, advocacy organization, or political action committee. I have zero affiliations in this fight. I am writing as a concerned citizen and a technology professional who wants our state to get public policy right.
My core belief, and I will repeat it more than once in this piece, is that good public policy maximizes benefits and minimizes costs. “Costs” are not just dollars. They are the impact on a school district, the view out a kitchen window, the strain on the power grid in August, and whether a community still feels like itself in ten years.
With that out of the way, let’s get to it.
What people are actually worried about
The right place to start any policy conversation is with what regular people are actually feeling. Walk into a town hall when a new data center gets announced, scroll Facebook, read the letters to the editor, and the same themes show up on repeat:
“It’s a giant power-hungry building that’s going to jack up my electric bill.”
“They’re going to drink the lake dry to cool their computers.”
“They’re paving over farmland so some trillion-dollar company in Seattle can train a chatbot.”
“All the benefits leave town. All the headaches stay here.”
And layered on top of all that, because we are living in The AI Moment, there is the looming worry: “These are the buildings that are going to take everyone’s job, right?”
These concerns are not stupid. They are not based on nothing. A modern AI-focused data center campus can draw as much electricity as tens or even hundreds of thousands of homes. Older cooling designs did use staggering amounts of water. Recent surveys show Americans are significantly more likely to think these facilities hurt the environment, raise home energy costs, and damage local quality of life than help.
If you are skeptical of a giant industrial project showing up next to your kid’s elementary school, congratulations, you are paying attention. That is the correct response. Blind cheerleading and blind opposition are both how communities end up regretting things twenty years later.
But here is the part that gets lost: a lot of the loudest fears are aimed at an image of data centers that is already a generation behind the engineering. So let’s update the picture.
What a data center actually is
Strip away the buzzwords and a data center is just a building full of computers, doing the work that makes modern life modern.
Every text you send. Every Netflix episode. Every banking app login. Every Google Maps rerouting around a wreck on I-40. Every telehealth visit. Every photo backed up to the cloud, which, fun fact, is not actually a cloud. It is a building somewhere with really good air conditioning. Sorry to ruin the romance.
When people picture a “data center” today, they are usually picturing one specific kind: the huge ones. The football-fields-of-servers kind. Engineers call these hyperscale facilities, and they are the ones built to train the giant AI models, the ones that need enormous amounts of power, water, and land, and tend to cluster in places like Northern Virginia, Texas, and parts of the West Coast.
But that is not the only species in the zoo. There is a whole other category: smaller, more modular facilities placed closer to where regular work happens. The industry calls these edge data centers. Think of it this way:
A hyperscale data center is the giant industrial bakery that mixes the dough and figures out the recipes.
An edge data center is the neighborhood oven that bakes the loaves people actually eat for dinner.
The technical word for what edge centers do is inference, which is just the term for using an AI model after it has been trained. Inference is what powers the practical stuff: fraud alerts on your debit card, predictive maintenance at a manufacturing plant, smart traffic lights, the chatbot that finally answered your insurance question at 11 p.m.
Here is the thesis I want every North Carolinian to consider:
The future of artificial intelligence will not be won by whoever builds the biggest model. It will be defined by whoever puts AI to work effectively in real businesses, real hospitals, real schools, and real communities.
That means inference. That means edge. And that means North Carolina has a much bigger opportunity than chasing trophy hyperscale projects.
The big fears, meet the actual numbers
Let’s go through the major worries one by one, because the gap between “what people fear” and “what the engineering shows” is wide enough to park a server rack in.
Fear #1: “They’ll suck the power grid dry.”
The honest answer: data center electricity demand is going up, and we should not pretend otherwise. In 2023, U.S. data centers used about 4.4% of the nation’s electricity. Projections put that at somewhere between 6.7% and 12% by 2028, mostly driven by AI.
That sounds alarming. It is a real planning challenge. But here is the context the alarmists leave out.
Throughout the 2010s, the amount of computing being done exploded, and yet total data center electricity use stayed surprisingly flat. The industry got dramatically more efficient. Better servers. Smarter software. Buildings designed by people who, frankly, hate wasting money on electric bills as much as you do.
Engineers have a stat called Power Usage Effectiveness, or PUE. The plain-English version: if a data center uses one watt of electricity to actually do computing, how many extra watts does the building burn on lights, fans, cooling, and overhead? A PUE of 2.0 means you waste a watt for every useful watt, basically a 1990s SUV. The industry average is now around 1.5, and the best operators are running at 1.1 or lower, the data center equivalent of a Prius pulling a Tesla in its trailer.
New designs keep pushing it further: liquid cooling that runs coolant right to the chip, immersing whole servers in special non-conductive fluid (yes, really, it looks like baby oil and is roughly as unsettling), free-air cooling in cool climates, AI systems that manage other data centers to save energy, and waste heat that gets piped to nearby buildings or greenhouses.
Demand is rising. But “data centers waste electricity” as a flat statement is about as accurate as saying “cars get bad gas mileage.” Which cars? From when?
Fear #2: “They’ll drink all our water.”
This one has real history. Older evaporative cooling systems, the kind that work like a giant swamp cooler, really did use hundreds of thousands to millions of gallons of water per day at the biggest sites. That is a legitimate thing to be mad about.
But the technology has moved fast. Closed-loop cooling recycles the same water over and over, like the radiator in your car. Liquid immersion uses essentially zero water, those servers are bathing in something that is not H₂O. Many sites use reclaimed wastewater, the stuff that already went down a drain, so they are not competing with drinking water at all. Some modern data centers use less water per square foot than a typical office building. Read that twice. Your local accountant’s office may be a bigger water hog than a well-designed data center.
Modular edge data centers, the smaller, distributed kind I am advocating for North Carolina to prioritize, are inherently easier on the environment. Smaller footprint. Less land. Cooling sized for local conditions. Closer to users, which means less electricity wasted shoving data across long-distance transmission lines.
There is also a resilience angle nobody talks about: if one giant hyperscale site has a problem, that is a big problem. If one node in a distributed network has a problem, the rest pick up the slack. You do not put all your digital eggs in one very large, very thirsty basket.
The right policy is not “block them all” or “approve everything.” It is: require best-in-class technology, pair them with clean energy where you can, demand transparent reporting on actual water and power use, and reward operators who invest in efficiency.
Fear #3: “AI and data centers are going to take everyone’s job.”
This one deserves real care, because it is where perception and reality diverge the most.
Here is the line I want you to remember: AI isn’t usually the layoff story. It’s frequently the cover story.
When a big company announces layoffs and cites “AI transformation,” go look at the actual numbers. Nine times out of ten, you will find post-pandemic over-hiring that needed correcting, investor pressure to cut costs, offshoring decisions that were already in motion, or a CEO who needed a sexier press release than “we hired too many people in 2021.”
The actual jobs picture for data centers:
Construction. Big projects bring thousands of well-paying construction jobs for one to three years. Real money flowing through real local economies.
Operations. Once running, a single facility employs fewer people, usually 50 to 200. Modest, but not nothing. And those are skilled, stable, year-round jobs.
Multiplier effects. Counties hosting data centers have seen roughly 4–5% private sector employment growth over time, with bigger jumps in construction and information sectors.
Enabled economy. This is the part economists are bad at counting but families feel. Data centers power tools that let a small-town clinic run AI-assisted diagnostics. They let a 12-person manufacturing shop in Hickory compete with a giant in Shenzhen. They let a marketing agency in Wilmington serve clients across the country with two employees and a good idea.
AI accelerates trends that were already happening. It is not, despite the headlines, a robot army marching to your cubicle. The smart response is investing in workforce transitions: education that lines up with what employers actually need, retraining for people whose roles are changing, and a regulatory environment that does not try to freeze 2015 in amber but does help people move into higher-value work.
We have been here before.
Imagine you are a blacksmith in 1905. Or a stable hand. Or one of the thousands of people whose entire livelihood depended on horses, the people who fed them, shod them, harnessed them, cleaned up after them, sold buggies, repaired wheels. Henry Ford rolls in with a loud, smelly contraption that breaks down every six miles. The newspapers are full of pieces about the death of an entire way of life. And honestly? Those pieces were not wrong about the disruption. A lot of jobs did go away.
But look at what came next.
The car did not just replace the horse. It invented the road trip. It invented travel tourism, entire towns, hotels, diners, gas stations, national parks visited by ordinary families. It let goods and services move roughly 100 times faster without anything like 100 times the cost. It created entire industries that nobody in 1905 could have written down on a piece of paper because the words for them did not exist yet.
Fast-forward several decades. We invent the internet. Also a massive disruption. Also accompanied by a lot of doom-saying.
Then you combine the two, cars plus the internet, and you get Uber. Lyft. DoorDash. Uber Eats. Industries that employ millions of people, that did not exist in any form a generation ago, and that depend entirely on technologies people were once terrified of.
And then ask yourself a question worth sitting with:
Where would we have been during the COVID-19 pandemic without those technologies?
How many small restaurants survived because DoorDash existed? How many elderly people got groceries delivered because Instacart existed? How many essential workers got to their shifts because Uber and Lyft existed? How much of normal life kept limping along because the internet let us work, school, and see our doctors from home?
The technologies that earlier generations feared became the technologies that carried us through a global crisis.
I am not saying AI is going to be exactly like that. But every major technology shift in American history has followed the same shape. Real disruption. Real fear. Real job losses in specific places. And then, if we engage thoughtfully instead of either panicking or pretending nothing is changing, entirely new industries, new jobs, and new capabilities that the doom-sayers could not have predicted with a crystal ball and a winning lottery ticket.
The blacksmiths who learned to fix engines did fine. The ones who insisted nothing was changing did not. That is the lesson. That is always the lesson.
Fear #4: “It’ll wreck our community.”
This concern is the most legitimate of all when projects are poorly sited. A massive industrial facility plopped next to a neighborhood by a developer who could not find the town on a map? Yes, that is bad. That is correctly bad.
But it is also the most fixable through plain old good governance: smart zoning that puts these things where they fit, noise standards that modern designs can absolutely meet, scaling that grows as a community can absorb it, real public-private partnerships (not the kind where the public part is “shut up and accept it”), and requirements that operators contribute to local infrastructure they are using.
Community impact is not destiny. It is policy.
Why North Carolina specifically, and why edge is our lane
This is where North Carolina has a real choice to make.
We have reliable power. We have a skilled workforce. We have universities cranking out engineers. We have a business climate that is the envy of much of the country. Companies want to be here.
So the question is not whether data centers come. It is what kind and how.
The temptation is to chase trophy hyperscale projects, the giant ones with the big ribbon-cutting photos and the press releases full of round numbers. Those have a place. Allow them where they fit. Make them follow strict efficiency standards. Take the wins where they make sense.
But the bigger opportunity, the one that distributes prosperity instead of concentrating it, is to lead the country in modular, distributed edge data centers for AI inference.
Picture this: smaller facilities spread across the state, urban and suburban and rural. Each one creating construction jobs locally, operations jobs locally, and supporting local AI applications. Health systems running diagnostics with low latency because the compute is in the region, not in Virginia. Small businesses tapping into AI tools that are fast and affordable because the data does not have to travel halfway across the country. Better data privacy because more information stays close to home. Per-site impacts that are manageable, not the kind of footprint where one site changes the whole character of a county.
This is the model where the benefits do not all flow to coastal headquarters. It is where rural counties get a real seat at the table. It is where Raleigh and Charlotte do not have to be the only places that matter.
And this is the model where AI actually helps small businesses compete instead of just enriching the four companies you can name without thinking.
What good policy actually looks like
We have identified the concerns. We have corrected the worst myths. Now: what do we actually do?
Good policy is evidence-based, not vibes-based. It looks at honest cost-benefit math, with both columns filled out.
The benefits column includes: economic growth and construction jobs; the enabled economy, small businesses, hospitals, schools, and farms using AI; innovation in healthcare, agriculture, education, and public safety; national resilience (we should not want all of America’s compute to live in three zip codes); and quality of life improvements from practical AI applications.
The costs column includes: resource use (mitigated by modern technology and smart standards); local impacts (managed by zoning, siting, and community agreements); grid demand (addressed by generation planning and efficiency requirements); and worker transitions (handled by proactive education and retraining).
Both columns are real. The job of policy is to honestly weigh them.
Specific recommendations I would love to see North Carolina embrace:
Streamlined but rigorous permitting that favors efficient, modular, edge-focused designs. Fast for the good projects, hard for the lazy ones.
Performance-based incentives. Tax breaks tied to real metrics. Hit water targets, hit PUE targets, commit to workforce training, partner with clean energy, contribute to community funds. Earn the incentive. Do not just receive it.
Public-private partnerships for strategic site selection so we put facilities where they make sense and not where they create conflict.
Transparency requirements. Citizens and officials deserve actual data, power use, water use, emissions, not promotional brochures.
Workforce investment so North Carolinians are the ones building, operating, and benefiting from these systems. Community colleges. Technical training. Apprenticeships. The works.
Allow hyperscale where it fits, but make distributed inference our signature play. Be the state that deploys AI better than anyone else, not just the state that hosts the boxes.
This is not pro–data center. This is not anti–data center. This is pro–smart policy.
This is not a red-versus-blue issue. Please stop trying to make it one.
I want to land this hard, because the partisan sorting is already underway.
I have watched data center debates get forced into partisan boxes by people on both sides who would rather have a clean tribal villain than a complicated policy problem. That is lazy, and it is going to cost North Carolina dearly if we let it continue. There is nothing inherently conservative or liberal about wanting reliable power, clean water, good jobs, transparent government, and a state that competes in the next economy. Those are North Carolina values. They are American values. They show up at the dinner tables of people who voted very differently last November and will again next November.
The worst outcomes come from simplistic narratives. “Build everything, ask nothing” gets you grid problems, water fights, and angry communities. “Block them all” gets you a state that watches the future happen somewhere else. Maximum benefit and minimum cost requires thinking, evaluating, and adapting.
So here is my direct ask, and I am going to be a little more forceful about it than I usually am, because the moment calls for it.
To the folks who are anti–data center. Your concerns about water, power, noise, traffic, and what your community is going to look like in ten years are legitimate. You are not being paranoid. You are being a responsible citizen. But please, please, do not let “no” be the end of the sentence. Showing up to a hearing to shout “absolutely not” and then going home is not how good policy gets made. It is how policy gets made without you in the room, which is the worst of all possible outcomes. Sit down at the table. Bring your list of concerns. Bring your line in the sand on what you absolutely will not accept. Make policymakers earn every concession. That is how communities get water guarantees, noise limits, real tax benefits, and zoning that protects what you love about where you live. Your power is greatest when you are engaged, not when you are absent.
To the folks who are pro–data center. You see jobs, investment, tax base, and a chance for your community to be part of the next economy. Those are real, and they matter. But please, please, do not let “yes” be the end of your sentence either. Cheering for every project and waving away every concern is how communities end up with broken promises, broken pavement, and broken trust. The best thing you can do for the industry you support is demand it be excellent. Demand high efficiency standards. Demand workforce commitments. Demand transparency. Demand community benefit. When you push for excellence, you make the case for the industry easier, not harder. You disarm the loudest opponents by giving them less to be afraid of.
To everyone. Get in the room. Sit at the table. Bring your concerns and your willingness to negotiate. The people who do not show up do not get a vote on the future of their community. The people who show up only to throw rocks do not get one either. The people who show up with a clear head, a clear list, and a willingness to make a deal, those are the people who shape what gets built and what does not.
This is how maximum benefit and minimum cost actually happens. Not by accident. Not by slogan. By people on every side of this taking it seriously enough to engage.
We have done this before. Electricity. The telephone. The car. The internet. Every one of those came with real disruption and real fear, and every one of them came out better because societies engaged with them thoughtfully instead of running from them or rubber-stamping them.
We can do this one right too. But only if we actually do it.
The bottom line
Data centers are not going away. They are infrastructure for the next economy in the same way power lines and fiber were infrastructure for the last one. The question is what kind, where, and under what terms.
The gap between perception and reality is bridgeable, with better information, with a willingness to distinguish between different kinds of facilities, and with policies focused on real outcomes rather than easy slogans.
North Carolina has a chance to lead. Not by hosting the most boxes, by pioneering the responsible, distributed, broadly beneficial model that other states will copy in five years.
Here is what I am asking of you, fellow citizen:
Engage with your local officials. Show up. Ask hard questions. Do not outsource the future of your community to whoever shows up loudest.
Demand data and nuance. If someone is selling you a one-sentence answer, they are selling you something.
Support workforce programs. This is how the benefits actually reach families.
Push for incentives that reward efficiency and broad access, not vague promises and ribbon cuttings.
Talk to your neighbors, including the ones who disagree with you. This issue does not divide along the lines people expect, and the conversation gets better when more of us are in it.
This is your power grid. Your water. Your tax base. Your community. Your state. The decisions getting made right now will shape what North Carolina looks like for the next twenty years.
Do not sit this one out.