A Plain-English Guide to Data Centers in Kentucky

I could not tell you the first thing about the Lindsay Clancy trial. I know that it is currently occupying the social zeitgeist, but I do not follow it (as a matter of pride, I try not to follow any popular trial). But, when these trials reach the point where even I get a little curious, I do not wade through the internet. I call my friend Aschley (who is an EXPERT in these matters), tell her to explain the Clancy trial to me like I'm five years old, and in five minutes she gives me an overview so I can stop feeling like I am missing something.

Most of you have now seen the yard signs about data centers, or a post on Facebook or TikTok about water, power bills, or artificial intelligence, and you have not had time to sort out what is true. I would like to be Aschley for you on this one.

First, my bias. I am a construction lawyer in Lexington. I spent 13 years at, the Lexington construction company whose CEO and my friend, Stephen Gray (who is also a heck of a pickleball player), wrote an awesome op-ed advocating for data centers in Kentucky. My clients today are contractors and suppliers, and several of them would make money building data centers. I want these projects in Kentucky. I also would not want one across the road from my house. That is true of power plants, bourbon warehouses, and Louisville Cardinal practice facilities too, and it is not a reason to pretend the question is simple.

In my opinion, data centers are the railroads and interstates of this century. They are being built somewhere, with or without us, and the communities that host them will capture the tax base while the communities that do not will still use the services. But the railroads also got land grants and ran over anyone who was not organized, and the interstates went through the neighborhoods with the least political power. A county fiscal court in Kentucky cannot be expected to negotiate with the Amazons or the Metas on equal terms. That is not a knock on the county. It is why state government has to set the rules, so that every local body is not reinventing them under deadline pressure with a developer's lawyers across the table.

If you want more than five minutes on this, the best overview I have found is a conversation between Derek Thompson, the journalist who hosts The Ringer's Plain English podcast, and Jasmine Sun, a technology writer who spent ten days in July driving through Wisconsin and Michigan, sitting in on village board meetings and talking to the people on both sides of the yard signs. Her written account, "No Data Centers In My Backyard," and the August 11 episode built around it, "Why Americans Hate Data Centers," are the fairest treatment of this subject I have seen. Sun went looking for the water and power arguments and came back with something more useful: what actually turned those towns against the projects, and what the towns that made peace with them did differently. Thompson, who is broadly optimistic about building things, spends the hour pressing her on whether the opposition is right. If you listen to one thing before your county votes, make it that.

With that on the record, here is a “plain” answer to the questions I hear most.

How many are we talking about?

More than most people realize. A Kentucky Lantern survey in July counted more than 15 planned hyperscale projects across 15 counties, from a 1.6 million square foot campus approved in west Louisville to gigawatt-scale proposals in Barren, Mercer, Mason, Greenup, and Boyd counties. Not all of these will be built. Developers file speculative projects in several states at once and build where the power shows up first. But even a third of that pipeline would be the largest industrial buildout in state history.

Will they raise my electric bill?

This is the concern that matters most, and the honest answer is: not yet, but they can, and the protections against it are incomplete.

A Rutgers study of a decade of data from 2,277 operating data centers found no detectable increase in local residential bills through 2024. The risk is in what comes next. A single 100 megawatt facility draws power continuously equal to roughly 80,000 homes. Serving that load means new plants, substations, and transmission lines. Kentucky's utilities have said they will pay for it with data center revenue. LG&E and KU won approval last October to spend about $3 billion on two new gas plants, justified largely by projected data center demand, and proposed a tariff under which data centers commit to a 15-year contract and pay for at least 80 percent of their projected monthly usage whether they use it or not.

That is the right structure. The problem is what happens if the demand does not show up. Once a power plant is in the rate base, the Public Service Commission has limited ability to reassign its cost to the customer who caused it. Everyone else pays. Virginia, which has hosted data centers longer than anyone, now projects residential increases of $14 to $37 a month by 2040 without changes to how costs are allocated. In the PJM market, the price utilities pay to guarantee capacity rose roughly tenfold in one auction cycle, with data center load a leading cause.

Every large project should carry a binding contract that covers the infrastructure built to serve it, an application fee that screens out speculators, and collateral that survives if the operator walks away.

Do they use up our water?

Less than the internet says, and it depends entirely on the design.

Older and cheaper facilities use evaporative cooling, which can consume well over a million gallons per megawatt per year. Many new hyperscale builds use closed-loop or air-cooled systems that use a small fraction of that, and some are close to zero after the initial fill. Sun found that a closed-loop facility uses about as much water as a golf course.

The fair concern is not that a data center will drain the aquifer. It is that nobody should have to take the developer's word for it. Cooling design should be disclosed and locked in as a permit condition, not left to the operator's discretion after approval.

How many jobs?

Fewer than a factory, and anyone who tells you otherwise is selling something.

Construction is the real employment story. A large campus can put a thousand or more tradespeople to work for several years at good wages, and that is exactly why the construction industry, including my clients, is enthusiastic. Once the building is finished, a Brookings analysis of roughly 1,500 facilities found that a county's first large data center adds on the order of 100 to 200 permanent jobs. Toyota employs 10,000 in Georgetown. A data center never will.

What a data center offers instead is tax base with very little demand on services. It does not send children to school or cars onto the roads, and it can pay substantial property taxes on billions of dollars of investment, if the county does not abate them. That is a legitimate trade for a rural county with a declining population. It is a different trade than a factory, and it should be sold as what it is.

Are they loud? Do they pollute?

Compared with a factory, no. A data center has no smokestack, no process emissions, no shift change, and almost no truck traffic once it is built. The main air permit it needs is for the diesel backup generators, which run during outages and periodic testing. If your mental picture is a chemical plant or a coal plant, adjust it. What a data center does have is thousands of fans and chillers running 24 hours a day, and that is the complaint you hear from people who live near one. It is not loud in the way a jackhammer is loud. Measurements Sun reported from the fence line of Microsoft's campus in Mount Pleasant, Wisconsin, came in at 67 decibels, about the level of a lively conversation. The problem is that it never stops. A steady low hum at 2 a.m., every night, on a farm that used to be silent, wears on people in a way that a daytime noise does not, and it carries farther than you would expect over open ground. Kentucky has already had a version of this fight. Neighbors of a bitcoin mine in Wolfe County spent three years complaining about the hum before the operation shut down this summer, and the county's answer was a sound barrier it should have required on day one. Noise limits measured at the property line, setbacks from homes, and equipment enclosures are cheap to build in at the start and nearly impossible to add later. Any local approval should have all three in writing.

Why does Kentucky give them tax breaks?

The legislature exempted data center equipment from sales and use tax for up to 50 years, first for Jefferson County in 2024 and then statewide in 2025, for projects investing $450 million or more, with lower thresholds in smaller counties. The Kentucky Center for Economic Policy estimates that four of the proposed projects alone would forgo more than $2 billion in state revenue. In late August, the Governor called for repealing the exemption, and Republican individuals in both chambers said they are open to shortening or ending it.

Reasonable people disagree here, and I will not pretend the answer is obvious. Every neighboring state offers something similar, and the exemption is what puts Kentucky on the site selection list. On the other hand, national data suggests incentives amount to about 2 percent of a hyperscale project's cost, which raises a fair question about whether they change any decisions. Whatever the General Assembly does, the incentive should at minimum be conditioned on compliance with local rules and on the ratepayer protections above.

Why are some of the deals secret, and why are people so angry about it?

Sun's reporting reached a conclusion that surprised Silicon Valley. Opposition in Michigan and Wisconsin runs about 70 to 30 against, and it is almost identical among Republicans and Democrats. But the residents she interviewed rarely talked about water or AI. They talked about mayors who signed nondisclosure agreements, councils that approved projects before the public knew the name of the company, a village still carrying $183 million in debt from a Foxconn project that delivered a fraction of what was promised.

The anger is about process, and the fix is not complicated. Developers routinely ask local officials to sign nondisclosure agreements before the first meeting, so that a magistrate or council member cannot tell constituents which company is behind the project, how much power it will draw, or what it has asked for. That is a reasonable practice for a private land deal. It is not a reasonable practice for a public body deciding whether to bring a gigawatt of load into a county. Microsoft announced in March that it will no longer require nondisclosure agreements from local governments anywhere in the world, which tells you the industry can live without them. Kentucky should make that the rule for everyone: no confidentiality agreements with public officials, and the company name, the load, the cooling design, and the incentive request on the table before any vote. A developer that will not negotiate in public is telling you something about the deal.

The flip side is also true, and the developers who do this well know it. The projects that get built are usually won long before anyone talks about zoning or tax breaks. The hyperscalers that show up early, sit through the fiscal court meetings that have nothing to do with them, hire local, buy local, and let people meet the person whose name will be on the permit are the ones that have earned enough trust to survive the first bad headline. Sun saw the contrast in Wisconsin: in Mount Pleasant, Microsoft held dozens of public meetings and dropped its nondisclosure agreement, and the project moved forward even in a village that had been burned by Foxconn. Toyota did the same thing in Georgetown 40 years ago, and Scott County has been reaping the benefit ever since. The rules I am describing are a floor. A developer that treats them as a ceiling, and does the minimum the statute requires, will get the reception it deserves.

Kentucky is already seeing the same pattern. Moratoriums in Bell, Barren, Mercer, and Fayette counties. Lawsuits by developers against Cave City and Simpson County over local ordinances. Residents in west Louisville who spoke for hours against a project that the planning commission approved 6 to 1 because zoning is the only question it is permitted to ask.

So what should we actually do?

Treat this like the infrastructure decision it is, not a zoning fight and not a referendum on artificial intelligence. The state should adopt one framework so that no county has to negotiate from scratch against a company with more lawyers than the county has employees: cost-responsibility contracts approved by the PSC so the data center pays for the plants and lines built to serve it, public negotiation with no nondisclosure agreements, cooling design and power load disclosed before any local vote, decommissioning bonds so that a failed project does not become a concrete slab, and community benefit commitments that are written down and enforceable rather than promised at a ribbon cutting. Then let the counties decide, with those protections already in place. In the many Kentucky counties with no zoning at all, there is no local vote to begin with, which is exactly why the rules have to come from Frankfort.

The wrong answer is the one New York chose. In July, Governor Hochul enacted a statewide moratorium on new hyperscale data centers for up to a year while the state writes environmental standards and a community benefit framework. Every one of those is a reasonable thing to want. None of them requires shutting the door to get it. A developer with a site under contract and billions in financing does not wait a year to see what Albany decides. It goes to Pennsylvania or Ohio, and the rules New York eventually writes will govern projects that are no longer coming. Kentucky does not need a freeze to write good rules. It needs the General Assembly to write them in January.

Then build them. Kentucky has the land, the water, the grid, and the trades. Stephen Gray is right that these facilities will be built and the benefits will land somewhere. The communities that come out ahead will be the ones that said yes with conditions, in writing. The ones that lose will be the ones that said no until the developer sued, and the ones that said yes before they read the contract.

So before you put up the yard sign, in either direction, ask your county judge three questions. Who is the company? Who pays for the power lines? And what happens if they leave? If nobody can answer those, the problem is not the data center. It is the deal.

Nate Simon is the founder of Simon Law, PLLC, a construction law firm in Lexington. His clients include contractors and suppliers that stand to benefit from data center construction. He also uses the internet, which as far as he can tell is how the data centers got him. He has not weighed in on the Clancy trial and does not intend to.