Communities can grow with new data centers
Data centers can create lasting value for communities.
This is a column about technology. See my full ethics disclosure here.
Almost daily, we see another headline for a new data center being built. While Louisiana cheers Meta’s new $10 billion facility as “progress.” West Texas officials call Vantage’s $25 billion campus a “economic growth driver,” and North Dakota celebrates a new $3 billion build. But it doesn’t take long to realize what’s happening here as a clear pattern emerges.
The scale of these projects makes the headlines irresistible, but the celebratory tone obscures the underlying imbalance. Ten or twenty billion dollars sounds like a community lifeline investment, but most of that capital goes into concrete, cooling, and servers. The actual payroll and human capital investment is minuscule compared to the factories and mills of the past. A single data center facility can house millions of dollars of hardware but employ fewer people than a mid-sized grocery store. For a town that might be desperate for jobs, the math rarely pencils out once the ribbon-cutting ceremony is over.
In lieu of railroads or steel mills, for a growing number of rural areas, the anchor employer is quickly becoming a data center. Instead of wages and long-term economic stimuli, local community leaders are locking in business-friendly utility contracts.
The trap of the company town is that the company holds the leverage, and the town holds the bill. When the math changes, the company leaves. The town can’t. Today, the story is no different; instead of a coal seam, the resource being extracted is cheap power and water.
The dependence on cheap utilities turns these towns into captive suppliers. Once a data center locks in a preferential rate with the local power grid and negotiates favorable water rights, every other resident bears the residual costs. Soon, residential power bills inch upward to cover infrastructure upgrades. Farmers and small businesses find themselves in a bidding war for the same aquifers that now feed cooling towers. And because these contracts are often written for decades, local governments have little leverage to renegotiate terms when conditions change. All of this occurs while the profits generated from the data center are often routed far away from the community in which they are created.
That’s the risk with AI campuses. They create construction jobs, a handful of long-term roles, and a mountain of utility commitments. When costs spike or demand shifts, the burden stays local.
Another under-discussed piece is resiliency. Unlike a steel mill that leaves behind a physical plant or a railroad that at least laid down tracks, data centers leave little usable infrastructure once they go dark.
The specialized cooling, power delivery systems, and server racks aren’t assets a community can readily repurpose. When a data center operator decides to consolidate workloads or move compute elsewhere, what’s left behind is a shell of an enormous concrete block that requires ongoing maintenance or demolition. Even if the hardware that’s left behind is still usable, it’s likely to be obsolete or near end-of-life. Ultimately, communities will likely end up with liabilities rather than assets.
The outcome isn’t inevitable, because there are smarter ways to run the play. Some policy advocates argue that local communities could and should own equity shares in the industries they host, and they should hold a larger stake proportional to the long-term environmental and economic risks they face. These dividends could be put towards longer-term community improvement, along with oversight efforts to ensure the natural resources are protected.
Some have compared this approach to the model of sovereign wealth funds in resource-rich countries. Norway, for example, built a national fund off oil revenues that continues to support public welfare long after the original wells decline. While U.S. towns won’t wield the same scale, the principle holds: if your resources are being consumed for global gain, the community deserves a slice of the future upside. Equity stakes, resource royalties, or mandated community reinvestment programs could serve as modern guardrails against the hollowing out that follows when the boom cycle ends.
Communities can still be business-friendly while requiring new transparency laws and regulations. Transparency cuts both ways. Just as communities need visibility into the long-term costs, investors would benefit from a clearer social contract. The endless disputes over zoning, subsidies, and environmental impacts drag projects into years of litigation. If negotiations began with binding frameworks already in place, costly delays over commitments on water use, emissions offsets, and community payouts could be minimized and face less pubic resistance. What corporations forgo in short-term discounts, they would gain in long-term legitimacy.
I know this sounds anti-American, but protecting our natural resources is one of the most American things we’ve done. We enjoy some of the most beautiful National Parks in the world.
Local leaders can and should prioritize the long-term interests of their communities by establishing economic protection frameworks. For example, in California, residents are already absorbing the costs from power companies’ wildfire mistakes, which have led to state legislators drafting new laws to protect residents from footing the bill for “power-guzzling” data centers. As data centers continue to expand in square footage, residents should be compensated for the extraction of their vital resources and protected against the inevitable penultimate chapter of every company town story.
There are options. We can either accept AI something that extracts resources and exports profits, or we can demand a model closer to community partnership that creates wealth for many. The company town can provide a short-term boon, but it invariably leads to economic pain and sorrow for those in its wake.
The first wave of AI data centers looks a lot like the old company towns. The next wave doesn’t have to.
This is Part 3 of a 4 part series. Catch up on Part 1, or read the last chapter.