Facts Matter: Setting the Record Straight on Data Centers
I’ve stayed pretty quiet about economic development and the projects that come with it for a long time — mostly because I don’t enjoy back-and-forth arguments built on hypotheticals. But after nearly 20 years working in this field, I think it’s worth taking a step back and explaining what data centers actually do — and what they don’t do. Not about one specific project, but in general, because there seems to be a lot of misunderstanding out there.
First, what data centers do: they store, process, and move digital information. That’s it. They are essentially large-scale infrastructure for data — similar to how warehouses are infrastructure for goods. They are capital-intensive, highly regulated, and designed to operate reliably for decades.
What they don’t do: they don’t automatically raise electric or water rates, they don’t override utility rules, they don’t get special local tax deals just by being built, they don’t drain “all the resources” of a community — and they do not cause cancer, miscarriages, or other health conditions. There is no scientific or medical evidence supporting those claims, and data centers are not classified as hazardous facilities.
On utilities — approving zoning does not automatically raise rates. South Dakota is part of the Midcontinent Independent System Operator (MISO) grid, which operates under a regulated system. Large users don’t trigger rate increases just because they show up.
That’s very different from places people often cite, like the PJM market in parts of the eastern U.S., where competitive capacity markets can allow projected demand alone to influence system-wide pricing. That mechanism simply doesn’t exist here.
For example, if rates were ever to increase because of a specific customer, the electric utility serving the area — such as Xcel Energy in Sioux Falls — would first have to determine that new infrastructure is needed, decide not to charge that customer directly, file a public rate case, and receive regulatory approval. That process is separate from zoning, transparent, and nothing about it is automatic.
The same goes for capacity and infrastructure. Utilities and MISO do not connect projects unless power, water, and wastewater capacity already exist — or can be added and paid for by the customer. Local governments and counties are not writing checks to fund private electric infrastructure.
When people point to California rolling blackouts, that’s a completely different grid structure, climate-driven peak demand, and decades of policy decisions. South Dakota does not operate under that model.
Now, on jobs — because I keep seeing “75 jobs” dismissed like it’s no big deal. It actually is. These are high-skill, high-wage jobs, not seasonal or entry-level positions. Even conservatively, 75 jobs at an average salary of $80,000–$100,000 equals $6–7.5 million in annual payroll flowing into the local economy every single year. That supports housing, retail, healthcare, childcare, and local businesses — year after year.
On top of that, data centers add significant property tax base. A large data center campus can represent hundreds of millions of dollars in taxable value, supporting schools, counties, cities, and other taxing entities — without the traffic, school enrollment growth, or service demands that come with residential development. Undeveloped land contributes very little; developed industrial property helps pay for the things communities already rely on.
On taxes — I keep seeing references to “50-year tax breaks.” Local zoning decisions do not include tax abatements unless they’re approved separately in a public vote. What people are usually talking about is House Bill 1005 (HB1005), a proposed state bill that has not passed. Even if it did, it would apply only to state sales and use tax on certain equipment, not property taxes, utilities, land, buildings, or infrastructure.
And this isn’t new tax policy. Manufacturers have received sales tax exemptions on qualifying equipment for decades.
For example, a manufacturer buying a $900,000 CNC machine avoids about $58,500 in sales tax — one time, at purchase. A data center buying a $1.2 million server rack would avoid about $78,000 — again, one time at purchase. No cash subsidy. No ongoing “tax holiday”.
The reason “50 years” shows up in HB1005 is because facilities like manufacturers and data centers replace equipment repeatedly over decades. The bill simply creates a long eligibility window so companies don’t have to re-qualify every few years. Each exemption still applies only once, at the time equipment is purchased. Legislators use long windows to provide certainty for capital-intensive investments — not to give away decades of taxes.
Finally, on risk — every industrial project carries risk. Manufacturing plants close. Warehouses go vacant. Processing facilities shut down. If that happens, the owner — not taxpayers — remains responsible. That’s not unique to data centers.
I’ve been around long enough to see this firsthand. Over the last 20 years in and around Watertown, I’ve watched site-selecting companies and manufacturers expand, contract, change ownership, relocate, or even close. That’s part of economic reality — not a failure of zoning or utilities, and not something unique to one industry.
That experience is also why communities don’t stop pursuing development just because risk exists. Economic development has never been about guarantees — it’s about managing change while building tax base, jobs, and long-term capacity. Communities that continue to invest in quality development are the ones that can absorb change, reinvest in sites when businesses evolve, and support schools and infrastructure without overburdening residents. Doing nothing doesn’t eliminate risk — it just limits opportunity.
It’s completely fair to debate whether a certain type of development is the right fit for a community. But those conversations should be based on how zoning, utilities, taxation — and basic economics — actually work, not on fear or hypotheticals.
