Google canceled a data center project last month after community opposition. When the lawyer confirmed Google was pulling out, the room erupted in cheers. Actual applause for Google not building something.
That should tell you everything about how these negotiations are going.
I've been following the data center boom, and there's this growing pattern of communities just saying "no thanks" to massive tech infrastructure projects. And you know what? I kind of get it.
The tech industry narrative is all "we're bringing economic development!" and "jobs and investment!" But communities are looking at the actual deal and realizing it kind of sucks for them.
The Math Doesn't Math
Here's the pitch tech companies make to local governments:
"We'll build a massive data center in your community. You'll get property tax revenue, some construction jobs, and you'll be part of the AI revolution!"
Sounds great. Until you look at the details:
Property taxes: Yes, the building pays taxes. But it's a building full of servers, not people. Servers don't use schools, libraries, or social services. You're getting revenue without corresponding costs, which seems good.
Jobs: A data center might employ 50-100 people permanently. Most of those are specialized tech roles that won't go to local workers. Construction jobs are temporary. This isn't a factory employing thousands.
Infrastructure strain: Data centers use enormous amounts of electricity and water. That puts pressure on local utilities, potentially driving up prices for residents who were already there.
Few local benefits: Unlike a factory or office building, data centers don't generate much spillover economic activity. Workers aren't shopping locally or eating at restaurants—there just aren't that many workers.
So from a community perspective, you're getting:
- Some tax revenue (good)
- Massive infrastructure strain (bad)
- Minimal local employment (bad)
- Rising utility costs for residents (bad)
- Basically zero community amenities (bad)
That's... not a great deal?
The Water Thing Is Real
Data centers need water. Lots of water. For cooling systems that run 24/7 to keep servers from overheating.
The Great Lakes region is seeing a data center boom specifically because of water availability. But the Great Lakes aren't infinite. Eight U.S. states and two Canadian provinces all draw from them.
One researcher described it as "a giant pitcher of water with straws going into it." Add more straws—data centers and the power plants to run them—and at some point you're drawing too much.
In Georgia, some residents reported problems getting drinking water from their wells after a nearby data center was built. That's not hypothetical impact. That's "my well ran dry because someone built a server farm next door."
If you're a local resident, why would you trade your reliable water access for Google's AI infrastructure? What do you personally gain from that exchange?
The Electricity Trap
Data centers need constant, reliable power. 24/7/365. Which means they can't run on renewables alone (solar doesn't work at night, wind doesn't work when it's calm).
So what gets built? Often: natural gas plants. Sometimes: coal plants getting extended life. Rarely: new nuclear (takes too long to build).
This creates a weird dynamic where communities that might be trying to reduce emissions end up with new fossil fuel infrastructure because AI needs power now.
And residential electricity prices tend to rise when you add large industrial users competing for the same power supply. Especially if utilities need to build new generation capacity.
So you're a homeowner in, say, rural Virginia. Google wants to build a data center nearby. Your electricity bill goes up to subsidize infrastructure for Google's AI. Your water availability decreases. You get... what, exactly?
Oh right, the knowledge that you're helping power ChatGPT. Worth it!
The "Economic Development" Myth
Tech companies love to talk about economic development, but let's be honest about what that means:
Data centers are increasingly automated. They don't employ many people. The jobs that do exist are specialized—network engineers, facilities managers—not entry-level positions for local workers.
Compare that to, say, a manufacturing plant that might employ 1,000 people at good wages. Or a corporate office with hundreds of knowledge workers spending money locally. Data centers aren't that.
They're essentially industrial facilities with minimal local economic impact beyond property taxes and utility bills.
For a community, there are better uses of land, water, and power. A data center is a terrible deal if what you actually need is jobs and economic opportunity.
The Transparency Problem
Here's what makes this worse: tech companies won't disclose how much water and energy they're actually using.
They claim it's competitive information that could give rivals an edge. Which is convenient, because it means communities can't make informed decisions about whether to allow these projects.
"Trust us, we'll be responsible" is not a compelling pitch when you won't share basic operational data.
One researcher noted: "Without more transparency around water and energy consumption, the public doesn't have the information it needs to make informed decisions about data center development."
And that's by design. If communities knew the actual resource consumption, they'd be even less likely to approve these projects.
Why Local Opposition Makes Sense
I'm generally pro-tech. I use AI constantly. I write about it professionally. But I completely understand why communities are blocking data centers.
If Google came to my neighborhood and said "we're going to use a significant chunk of your shared water and electricity, drive up your utility bills, employ basically no one locally, and in exchange you'll get... some property tax revenue?"
I'd also show up to the public meeting to oppose it.
The benefits accrue mostly to the tech company and their users (who are globally distributed). The costs accrue mostly to the local community. That's not a fair exchange.
The National vs. Local Interest Problem
Here's where it gets philosophically interesting: data centers might be good for the country or the world, even if they're bad for specific communities.
AI development could generate enormous economic value. Better productivity, new innovations, solving hard problems. That's real value for society broadly.
But "good for society broadly" and "good for this specific small town" are different calculations.
Why should a community in rural Oregon sacrifice their water and power to enable AI development that mostly benefits people elsewhere? What's the compelling case for them to bear the costs while others reap the benefits?
This is a classic externalities problem. The benefits are diffuse and global. The costs are concentrated and local. And we haven't figured out how to make that transaction fair.
What Would Actually Help
If tech companies are serious about getting community support, here's what they could do:
Actual transparency: Disclose energy and water usage. Let communities see the real impact before deciding.
Local benefit sharing: Provide free or reduced-cost internet to the community. Fund local schools and libraries. Create actual community value beyond tax revenue.
Renewable energy requirement: Don't just promise to eventually use renewables. Build the renewable capacity first, then the data center.
Local hiring mandates: Commit to hiring locally for positions that don't require specialized tech skills.
Community veto power: Give communities real input into whether and how projects proceed, not just rubber-stamp approval processes.
Will this happen? Probably not voluntarily. But states could require it as conditions for approving data center projects.
The Bigger Picture
The data center fights happening across the U.S. are a microcosm of a bigger question: who benefits from AI, and who pays the costs?
Right now, tech companies and their users benefit. Local communities hosting the infrastructure pay costs without seeing proportional benefits.
That's not sustainable. Either we figure out how to make the exchange more equitable, or communities will keep saying no.
And honestly? I don't blame them.
The AI revolution is cool and potentially transformative. But it requires massive physical infrastructure—water, power, land, cooling. That infrastructure has to go somewhere. And the communities where it goes have every right to demand a fair deal.
"Be grateful we're choosing your town" isn't a fair deal. "You'll benefit economically" (when you demonstrably won't) isn't a fair deal. "Trust us to be responsible" (while refusing to share data) isn't a fair deal.
My Uncomfortable Take
I think some data center projects genuinely shouldn't be built where they're being proposed.
Not because I'm anti-AI. But because the resource tradeoffs don't make sense for those communities, and we should respect that.
If a small town with limited water is choosing between serving local residents or Google's servers, the answer should be obvious. Residents come first.
If a region with an aging power grid is choosing between reliability for homes or capacity for data centers, again: residents come first.
The tech industry's attitude of "we're doing something important so communities should feel lucky to host us" is both arrogant and counterproductive.
Communities aren't resisting progress. They're resisting a bad deal that asks them to sacrifice their resources without fair compensation.
What Needs to Change
Long-term, we need:
-
Better siting decisions - Build data centers where resources are genuinely abundant, not where land is cheap.
-
Regulatory frameworks - Require transparent disclosure and community benefits as conditions for approval.
-
Alternative models - Maybe data centers should be public-private partnerships where communities share in profits, not just costs.
-
Distributed infrastructure - Instead of massive mega-data centers, maybe we need smaller facilities distributed more evenly.
-
Demand management - Actually asking whether every AI workload needs to be run immediately, or if some could be batched and optimized.
Right now we're in this gold rush phase where companies are racing to build infrastructure as fast as possible, damn the local impact.
That's creating opposition that could genuinely slow AI development. Not because people are Luddites, but because they're being asked to subsidize tech companies while getting nothing in return.
If the tech industry wants communities to support data centers, they need to offer something more compelling than "you should feel grateful we're building here."
Communities aren't stupid. They can do math. And right now, the math doesn't work in their favor.
Maybe that's something worth fixing before the next public meeting erupts in cheers at a cancellation announcement.