How can you lead the AI era while blocking the infrastructure development that makes it possible? That contradiction is now shaping local elections, zoning hearings, and community meetings across the country. It is not going away.
The public debate on data centers has become binary. Data centers are treated either as essential economic progress or an environmental and community threat. I reject both framings. The legitimate concerns about water use, electricity demand, and poor site locations are real.
But they are arguments for building more thoughtfully, not for stopping development. The answer is a better standard for how and where data centers are built.
The demand is not speculative. According to the U.S. Energy Information Administration, electricity consumption is forecast to rise from a record 4,195 terawatt-hours in 2025 to 4,288 terawatt-hours in 2026 and 4,356 terawatt-hours in 2027. AI data centers are an important part of that growing demand.
The global picture makes the pace of growth clearer. The International Energy Agency estimates that electricity consumption from data centers will roughly double from 485 terawatt-hours in 2025 to 950 terawatt-hours by 2030, which will account for around 3 percent of global electricity demand.
Responsible Development
This immense scale is why the debate matters beyond the towns that host these facilities. Data centers already underpin everyday digital life. Banking, logistics, communications, and public services run on them.
AI increases that dependency and the demand for data centers. Infrastructure policy is therefore economic policy, not a local zoning question.
The industry has not helped its own case. Poor operators have damaged public trust, and responsible projects are now caught in the backlash that follows. This damage to public trust is pushing communities to block development, as the country's physical capacity falls further behind its AI ambitions.
I have seen what responsible development looks like. In West Texas, the difference between two neighboring towns is visible from the highway. One has been left behind. The other has new roads, new schools, and upgraded utilities because the project that came in invested in the town as well.
That is not charity. It is the cost of doing business in a community that has to live with the infrastructure long after the ribbon cutting. Tax revenue, local hiring, and infrastructure investment are all part of the calculation, and when they are done well, they change what a town can offer its residents.
Many older data centers used evaporative cooling to save energy, trading lower electricity use for higher water consumption. But evaporation does not require boiling, and closed-loop cooling does not boil water to remove heat.
Direct-to-chip liquid cooling captures heat at the processor and carries it away through recirculating fluid. When paired with dry cooling or air-cooled chillers, this approach can eliminate evaporative water loss from cooling.
Water savings and energy performance depend on the full design and local climate. Communities deserve clear information about both.
Build Better
There is also an opportunity to put that heat to work. When developers and local communities plan together, recovered heat can serve district or campus heating, greenhouses, or aquaculture, reducing demand for separate boilers and water heaters. Projects in Europe, as reported by Microsoft, already demonstrate the potential. The right location, nearby heat users, and good partners make the difference.
The same logic applies to power. The single biggest constraint on data center growth is not land or capital. It is electricity. The grid is not expanding fast enough to meet the demand that AI infrastructure creates.
If the industry builds differently, the outcome can be different. Communities gain infrastructure they would not otherwise have. Utilities gain a customer that pays for the upgrades it needs. The country keeps pace with the technology it says it wants to lead.
What I ask of policymakers is a standard, not a moratorium. Demand measurable commitments on water use, power sourcing, infrastructure investment, and local benefit. Treat each project on its merits rather than treating every data center as the same. The industry should welcome that standard because it can separate the developers who plan to be good neighbors from the ones who do not.
Communities also have a role in shaping how these projects move forward. Engaging early allows residents to raise concerns and seek clear commitments on water use, power demand, and local investment before agreements are finalized.
The choice is not between AI leadership and community protection. It is between building the infrastructure now, with clear standards and community benefit, and building it later, in a country that has already lost the AI lead.
Eric Sonner is founder and CEO of Data Airflow, which supplies cooling infrastructure to hyperscale AI data centers. All opinions are the author’s own.