Today’s uproar over data centers reminds me a lot of the angst over Walmart two decades ago. There’s an enormous amount of bad information coming from advocates and opponents alike — and governments too often acting hastily as a result.

I wrote several studies and a book on Walmart to dispense with much of the wrong-headed thinking that surrounded it (see https://scholar.google.com/scholar?hl=en&as_sdt=0%2C15&q=michael+hicks+walmart&btnG= and https://www.google.com/books/edition/The_Local_Economic_Impact_of_Wal_Mart/uBB_ycoGZHAC?hl=en). I’ve started with some work on data centers (see https://michaeljhicks.substack.com/). Here’s what I think we know and don’t know about data centers.

There are at least three types of data centers in the U.S. They support mining of digital currency (Bitcoin, Dogecoin, etc.), cloud computing (where you store photos of your last vacation) and artificial intelligence. AI is driving the explosion of data centers.

Depending whose data you use, there are some 3,500 to 5,500 data centers in the U.S. They range in size from small operations, which are not even noticeable, to vast campuses of over a million square feet. To put that in context, a typical Walmart store might be 175,000 square feet.

I think AI will have a dramatic net benefit to society over the next few decades. Like other general purpose technologies, such as electricity or the internet, it offers to solve problems, eliminate some work tasks and free up human ingenuity and creativity. Economies and workers who can harness AI will do very well in the 21st century.

Data centers are a necessary part of the AI economy, but that does not mean having a data center nearby is an entrée into the exciting new world of AI, nor does it mean that every place should have a data center. For some locations, data centers could offer many benefits, while in other places they may offer nothing but costs.

The Brookings Institution recently released robust commentary and research on the questions and considerations that data centers raise for rural communities (see https://www.brookings.edu/articles/local-implications-data-centers-rural-communities-us/ and https://www.brookings.edu/articles/turning-the-data-center-boom-into-long-term-local-prosperity/). Let me drill down into the practical concerns that lie ahead of state and local policymakers.

First, most data centers have little or no real job creation associated with them. Most of the construction jobs associated with the early phase of operation will come from other construction sites nearby, mainly residential. The jobs probably pay better than home construction but are temporary. The long-term jobs on site are nearly zero and will not measurably benefit a community. Companies that don’t admit this should be viewed with suspicion.

Having a data center won’t magically cause an influx of high-tech jobs. For most AI applications, proximity to a data center is irrelevant. Where firms are sensitive to data processing delays, they will build data centers around clusters of the people they need to work in those businesses.

That's why Loudoun County, Virginia, has become the world's data center capital, with more than 300 facilities. The county has 443,000 residents, roughly 60% of whom hold college degrees, and land that typically commands more than $500,000 an acre.

If fewer than half your adults hold a bachelor's degree — and only a fraction of those hold graduate degrees — you aren't going to attract AI research and development jobs in any meaningful number.

There are places where a data center might crowd out higher value types of commerce. In fact, Loudoun County is already realizing that. In highly educated urban places, or places that are tourism or recreation oriented, data centers could reduce future economic growth.

This does not extend to farmland, however. The U.S. has a dramatic excess supply of farmland that will never again be used for agriculture. All the nation’s data centers could fit on Hoosier farmland not currently in use.

Data centers have some local nuisance effects. They can be noisy, perhaps unsightly, and they emit light pollution. So, like any other industrial use, data centers can harm local home values, but I haven’t seen a study that outlines the cost. Their use of water and energy also imposes some environmental costs, but this depends on the size, type of water use and source of energy.

However, data centers can provide some important benefits. They represent a large property investment, which means they offer enormous fiscal opportunities for many communities. Loudoun County has actually cut its property taxes significantly due to the influx of data centers.

The fiscal benefits depend on the state tax structure and whether data center operators receive local abatements or incentives. State and local governments that offer incentives, or have short duration property taxes (like Indiana), will see little fiscal benefit.

Data center complexes might also include their own power generation facilities, which are typically microgrid gas-fired plants. These facilities typically offer a few more local jobs than just a data center, but they also introduce new environmental damage.

The use of AI is likely contributing to higher electricity costs and is certain to do so in the years to come. Due to the way we regulate energy prices, the effect on residential utility costs is likely to grow regardless of where data centers are built.

A local data center probably won’t affect your electricity prices, but they are going to rise. That's because utilities spread the cost of expanded infrastructure across all ratepayers statewide, regardless of where the data centers that drove that expansion are physically located. A rural county that hosts zero data centers will see its residents' bills rise just the same; although, a genuine expansion of power generation, of all types, could eventually ease that pressure.

Data centers will continue to be built and will help fuel an important AI expansion. But, that does not mean they should be built in every community, or that state or federal governments should force (or prevent) communities into making their own decisions about data centers.

Like Walmart before it, the data center industry deserves neither the breathless boosterism nor the reflexive opposition it tends to receive. It deserves clear thinking.

Note: The views expressed here are solely those of the author, and do not represent those of funders, associations, any entity of Ball State University, or its governing body.