We’re asking: “Are you for or against data centers?”
But perhaps that is already the wrong question.
What infrastructure architecture does the United States need for AI—and where does a data center belong within that architecture?
A data center is increasingly discussed almost as though it were an isolated industrial development:
Here is 500 MW or 1 GW of demand. Can we find the land, electricity, water, transmission and permits for it?
Developer: We need capacity.
Utility: We need generation and transmission.
Community: Who pays, and what happens to our water, power bills and environment?
Politician: Stop the data center.
Industry: America will lose the AI race.
Everyone can be partly correct while nobody has solved the system.
And today’s reporting increasingly reflects exactly that political collision. Investors may still view community opposition as a manageable local-relations problem even as moratoria, restrictions and political resistance spread. Axios: Data center investors are overlooking the political risks
The Missing Layer Is Architecture
Instead of beginning with:
“Where can we build another data center?”
Start one level higher:
What compute capacity will the country actually need? How much should be hyperscale versus distributed? Where does existing generation capacity exist? Where can new generation be added responsibly? What transmission must be built? What cooling technology fits each climate? Where is water plentiful or scarce? Which workloads require proximity and which don’t? Where can waste heat or other infrastructure be integrated? What happens during grid emergencies? Who pays for the incremental infrastructure? And what does the host community actually receive in return?
Only after answering those questions should the map tell us where the data centers belong.
That’s why the emerging backlash should not simply be framed as an obstacle to AI.
Some of it may be NIMBYism. Some may be political opportunism. Some reflects legitimate environmental and economic concerns.
But underneath it may also be a planning signal.
The public may effectively be saying:
“You are asking our communities to absorb the consequences of an infrastructure system nobody appears to have designed as a system.”
That is much more interesting.
And here is the irony: political opposition itself is now becoming another infrastructure bottleneck. Data Center Watch has documented billions of dollars in proposed projects that have been delayed, blocked or withdrawn amid local opposition. Data Center Watch: Data Center Opposition Report
Perhaps We Should Change the Question
Perhaps the debate should not be whether America needs more data centers.
Perhaps the question is whether we should continue building them one project at a time without first designing the infrastructure architecture they collectively require.
We don’t need a data center on every corner.
We need a compute infrastructure plan.