Hostility and opposition to data centers was palpable four months ago. It has gotten much worse since then. Americans’ increasing aversion to new data centers borders on mass hysteria. A recent Gallup survey found that Americans would rather have a nuclear power plant built near them than an AI data center — by over 20 percentage points!
This animus toward new data centers, however, mostly rests on misinformation and social panic. After all, the biggest concentration of data centers pre-AI boom was in one of the wealthiest counties in the country (Loudoun). Northern Virginia, long dubbed “Data Center Alley,” has had a significant number of data centers for decades, without any noticeable harm to standards of living, quality of life, or electricity and water availability. Ironically, the demonization of new data center construction flies in the face of the fact that newer data centers are both more efficient and more valuable than older ones.
But even Republicans such as Texas Gov. Greg Abbott, who initially cheered AI data centers, are sounding a more cautious note. If this caution removes tax and regulatory subsidies to data center builders, well and good! But if it means killing projects over bad “vibes” or adding needless red tape and bureaucratic approvals, we have moved from caution to paranoia.
Opposition to data centers tends to focus on three things: water, electricity, and noise. People assume that adding a new data center near where they live will make their lives worse because it will be a nuisance, use up their water, and increase their electricity prices. But none of these concerns is insurmountable, and they are sometimes simply imaginary. People’s concerns about water usage are often off by an order of magnitude.
Newer AI facilities are undeniably massive — often spanning millions of square feet and drawing up to a gigawatt of power. But the panic over their resource footprint, whether water or electricity use, is way overblown. A single conventional hyperscale facility can consume between 1 million and 5 million gallons of water per day. While that may sound like a lot, it isn’t much different from a golf course or a couple of acres of farm land.
Across the United States, direct data center water consumption totals an estimated 17 to 35 billion gallons annually. By comparison, U.S. golf courses consume roughly 500 billion gallons per year, residential lawns use up to 3.3 trillion, and agriculture tops 40 trillion. Furthermore, major operators such as Google, Meta, Amazon Web Services, and Microsoft are rapidly adopting closed-loop liquid cooling systems that reduce on-site water usage.
In drought-prone regions, the fundamental problem is not data centers but government agencies failing to use market prices to allocate water efficiently in accordance with supply and demand.
People reasonably worry about high electricity prices. But rising electricity prices precede the data center boom. They have been driven by a combination of green crusading — shutting down coal power plants prematurely, slowing or stopping the building of new power plants, and dramatic market distortions through wind and solar credits and subsidies.
Bureaucratic management of utility companies has also led to chronic underinvestment in new infrastructure and slower adoption of new technology. If people want to see their electricity prices come down, they should push for regulatory and governance reform of utilities, not block data center development.
Noise concerns are similarly misguided. Standard industrial zoning frameworks have successfully managed noise generated by highways, airports, and factories for decades. If data center hum genuinely wrecked local quality of life, Loudoun County, Virginia — home to the world’s densest concentration of facilities — would not consistently rank among the wealthiest, highest-demand housing markets in the nation.
WANT LOWER ELECTRIC BILLS? LET UTILITIES BUILD POWER PLANTS AGAIN
Legitimate concerns exist when developers lobby local officials for tax abatements or free-ride on public grids. The solution is simple: Create and enforce clear, user-pays rules. Rather than panicking, concerned citizens and policymakers should make sure developers fully fund their power, water, and necessary infrastructure upgrades — from substations to transmission lines.
Like with any market transaction, both parties can benefit from the data center boom. But blanket obstructionism ignores economic trade-offs and risks hamstringing American leadership in artificial intelligence. Instead of giving in to populist panic or doling out special corporate carve-outs, policymakers should let private capital build AI infrastructure without unnecessary red tape.
Paul Mueller is a senior research fellow at the American Institute for Economic Research.