AI infrastructure is expanding faster than most local planning processes were designed to handle. Proposals for large data centers are working through county zoning boards, utility interconnection queues, and state permitting offices across the United States, often without much public attention until residents start asking questions.
That is changing. Local opposition to data center projects grew significantly in 2026, with dozens of proposed facilities canceled or delayed after community fights over electricity demand, water use, land use, and the value of promised benefits. Understanding what is actually at stake requires separating the claims that hold up from the ones that do not.
What a Data Center Actually Needs
A large data center is not just a building full of servers. It requires a continuous, reliable supply of electricity — often enough to power tens of thousands of homes — plus water for cooling, reliable network connectivity, and physical space. The power requirement is the most consequential.
A hyperscale campus, the kind built by Amazon, Google, Microsoft, or Meta, can draw as much electricity as a small city. That load has to be available every hour of every day, which means the local utility has to be able to deliver it reliably. In many regions, the grid was not built with that kind of concentrated industrial demand in mind. Interconnection queues — the line of projects waiting to connect to the grid — have grown long in part because of this surge in large-load applicants.
Data centers also use water. Most large facilities use evaporative cooling, which moves heat out of the building by evaporating water. Estimates vary by facility size, climate, and design, but water consumption is a real variable in dry or water-stressed regions, and it has become a flashpoint in some local disputes.
Why Opposition Has Grown
Reporting from Heatmap News found that local pushback against data centers is no longer a fringe concern. Dozens of proposed projects were canceled or delayed in 2026 after communities raised objections at planning meetings, organized opposition campaigns, or pushed for conditions the developers were unwilling to meet.
The concerns tend to cluster around a few themes. Electricity demand is the most common: residents and local officials worry about grid strain, rate increases, and the reliability implications of adding a massive new load to an already-stressed system. Water use comes up in drier areas. Land use is a factor when a proposed campus would replace farmland, wetlands, or open space. And community benefits — particularly the question of how many permanent jobs a facility will actually create — come up almost everywhere.
The pattern points to something the technology industry has not always handled well: social license. A project can be nationally significant and locally unpopular at the same time, and in the United States, local approval processes have real teeth.
What the Evidence Says About Jobs
The job argument is one of the most frequently made by data center developers and one of the most frequently contested by residents. Brookings Institution research from 2026 adds useful nuance to this debate.
The findings are not dismissive of the economic case. Data centers can raise local employment and wages. The construction phase creates significant work. Operational facilities create some permanent employment and tend to generate economic activity in the surrounding area.
But the effects vary substantially by facility type. Hyperscale campuses — the very large facilities operated by major cloud providers — generate more spillover economic activity than colocation facilities, which house servers for multiple tenants but tend to operate with smaller on-site workforces. Industry projections of job creation often use broad multiplier estimates that may not reflect what a specific facility type actually produces in a specific local economy.
The honest read on jobs is that the question is more complicated than either side usually acknowledges. The benefits are real. They are also frequently overstated in the lobbying and permitting process.
How Policy Is Responding
Some governments are beginning to respond to the grid-planning challenge directly. In Australia, state governments have moved toward a "bring your own power" model for data centers. Under the emerging framework, new facilities that want to draw from the grid may be required to invest in enough renewable energy and storage capacity to fully offset their demand. The logic is straightforward: if a data center is going to add significant new load to the system, it should bear some responsibility for supplying that capacity rather than simply absorbing what already exists.
This is not yet a widespread policy in the United States. But it reflects a shift in how regulators are thinking about the problem. Rather than treating data center growth as a benefit to be attracted and accommodated, some jurisdictions are treating it as an infrastructure decision that requires the same planning discipline as any other large industrial development.
The US Environmental Protection Agency has moved in a different direction at the federal level, proposing looser construction permitting for gas plants, data centers, and factories. That proposal reflects a different set of priorities: faster buildout over grid integration planning. The tension between these two approaches will shape data center siting for years.
What Stays Uncertain
Several things about the long-term picture are genuinely unclear.
How much grid stress data center growth will cause in specific regions depends on how fast the buildout continues, what the energy mix looks like, and how aggressively utilities invest in new generation and transmission. Projections vary widely and change quickly.
Whether renewable energy requirements like Australia's model spread to more US states is an open question. Some states have moved to attract data centers with tax incentives and streamlined permitting. Others have raised conditions. The patchwork is not yet resolved.
And the job numbers — for any specific facility in any specific location — are hard to know in advance. The Brookings research points to averages and ranges, not guarantees.
The AI Decoded Read
The decisions being made about data center siting right now are infrastructure decisions in the full sense. They will affect electricity rates, grid reliability, water supply, local tax bases, and land use for decades. They are also, in most of the United States, still made at the local and state level.
That is worth understanding. Zoning hearings, utility commission meetings, state permitting processes, and county commissioner votes are where the actual decisions happen. They are public. They are subject to comment. And they are where the tradeoffs the Brookings research describes — real but often overstated job benefits, real but often underweighted energy costs — actually get weighed.
Whether a specific community should welcome or resist a specific data center depends on factors that vary by location, grid capacity, water availability, and the actual terms of what the developer is offering. The factual record is a better guide than either the industry pitch or the opposition campaign.