Commentary: The next fight over artificial intelligence will not be decided in a laboratory or a congressional hearing. It will be fought in the towns where enormous data centers demand power, water and land—and where residents are being asked to trust that the benefits will eventually reach them.
That is why the emerging backlash against data-center construction deserves more than the usual caricature. Opponents are not necessarily anti-technology, just as supporters are not necessarily indifferent to climate change. The real question is who gets to decide the terms of the AI buildout, and who pays when the promised gains fail to arrive.
Recent reporting indicates that voters in 10 U.S. states and dozens of local jurisdictions are considering measures that would delay, restrict or impose new requirements on data centers. The concerns include electricity and water consumption, higher local infrastructure costs and possible health effects associated with new power generation.Reuters
The case for building
There is a serious argument on the other side. Data centers support advanced computing, cloud services and the development of AI systems that businesses and governments increasingly regard as strategic infrastructure. Construction can create jobs, expand tax bases and attract investment to communities that have struggled to participate in the digital economy.
Some facilities may also help modernize electricity grids. Large, predictable customers can finance transmission upgrades and encourage new generation. If developers pay the full cost of clean power, water protection and grid reinforcement, their demand could accelerate investment that would otherwise take years to materialize.
A blanket ban would therefore be shortsighted. It could push projects to jurisdictions with weaker safeguards, deprive communities of legitimate economic opportunities and leave democratic governments less able to shape a technology that will continue to develop elsewhere.
But “growth” is not a blank cheque
The problem is that the public debate often treats projected benefits as concrete while treating environmental costs as hypothetical. That is backwards. The electricity required by AI infrastructure is immediate; productivity gains are uncertain and may be distributed far from the communities hosting the servers.
One recent analysis cited by the Business Mirror warned that fossil-fueled power construction for data centers could erase years of progress in reducing U.S. electricity-sector emissions. The precise scale of that risk deserves scrutiny, but the direction is clear: an AI boom powered by new gas generation would undermine the climate goals that policymakers claim to support.
Water is similarly local. A facility’s impact depends on its cooling technology, climate and watershed, but those details should be disclosed before approval—not after residents discover that a supposedly digital industry has become a major industrial user of scarce resources.
A better bargain
Our position is neither “build nowhere” nor “build first and regulate later.” Data centers should require a public license grounded in enforceable conditions.
- Developers should disclose projected electricity and water use, including seasonal peaks and drought scenarios.
- New demand should not be met by quietly shifting grid costs to households or small businesses.
- Projects should demonstrate additional low-carbon power rather than merely purchase credits for electricity already on the grid.
- Communities should receive binding commitments on jobs, tax revenue, emergency services and environmental monitoring.
- Approvals should include penalties or closure provisions when operators miss measurable targets.
This approach acknowledges the legitimate promise of AI while rejecting the idea that urgency suspends accountability. It also gives responsible developers a clearer path than endless local opposition: meet the standard, pay the real costs and earn consent.
The technology sector has spent years warning that government must not fall behind innovation. That warning has merit. But democratic government is not supposed to sprint behind industry collecting the bill. It is supposed to decide whether growth serves the public, and under what conditions.
AI may deliver substantial benefits. It may also consume resources at a scale that worsens inequality and climate risk. The only defensible response is to make those outcomes subject to public rules before the concrete is poured. The future should be built—but not at the expense of the people told they have no choice.