America’s rush to build the enormous data centres required by artificial intelligence is encountering an increasingly powerful obstacle: the communities expected to host them. Texas says new requirements have halted as many as 1,800 proposed projects, while restrictions are spreading across states and hundreds of local authorities as residents question who ultimately pays for AI’s rapidly growing appetite for power, water and infrastructure.
INR Report: Based on reporting by David Lindfield for Slay News HERE
Artificial intelligence may exist in the digital world, but the infrastructure powering it is very physical.
Behind generative AI, cloud computing and increasingly sophisticated machine-learning systems are vast data centres filled with processors that require enormous quantities of electricity, cooling equipment, water and supporting infrastructure.
Across the United States, the scale of that physical expansion is beginning to collide with local concerns over electricity supply, water availability, power prices and neighbourhood development.
The political response is becoming difficult for the technology industry to ignore.
Texas Draws a Line
Texas Governor Greg Abbott says his administration's new approach has effectively halted as many as 1,800 proposed data centre projects, while imposing requirements intended to protect surrounding communities.
Under the Texas approach described in the source report, developers must avoid draining local water supplies, overwhelming the electricity grid or disrupting surrounding communities.
Abbott's office has gone further, arguing that companies benefiting from the AI boom should carry more of the infrastructure burden themselves.
His administration says data centres should invest their own money, provide their own power, reuse water and ensure that their operations add rather than subtract capacity from the electricity grid.
That represents a significant change in the debate.
The question is no longer simply whether a data centre creates investment and jobs. Governments are increasingly asking whether the economic benefit is sufficient to justify the resources these enormous facilities consume — and whether ordinary electricity and water customers should indirectly subsidise their expansion.
AI’s Infrastructure Problem
The extraordinary computing requirements of modern AI have transformed data centres from relatively obscure pieces of internet infrastructure into major industrial developments.
For communities hosting them, the concerns are straightforward.
A large facility can require substantial electricity generation and transmission capacity. Cooling systems may require significant water resources. New substations, transmission lines and other infrastructure may also be required.
The source report says the US Energy Information Administration estimates that restrictions in Texas could dramatically reduce projected electricity demand in the state.
That illustrates the scale of what is being proposed.
The AI revolution is therefore becoming an energy and infrastructure story as much as a technology story.
Industry Says It Can Be a Good Neighbour
The data-centre industry disputes the suggestion that expansion must come at the expense of local communities.
The Data Coalition, representing companies in the sector, says operators recognise their responsibilities to host communities and are committed to protecting residential electricity customers, strengthening grid reliability and practising responsible water stewardship.
Oracle offered a similar position in the source report.
Those commitments will increasingly be measured against actual outcomes.
If companies can generate or finance additional electricity capacity, recycle water and pay the full infrastructure costs associated with their facilities, some of the strongest objections may be reduced.
If the costs instead flow through to households, taxpayers or already-constrained infrastructure, political opposition is likely to grow.
New York and Pennsylvania Tighten the Rules
Texas is not acting alone.
According to the report, New York moved to halt construction of new data-centre projects exceeding 50 megawatts, affecting at least a dozen proposed facilities.
Pennsylvania has chosen a different model.
Governor Josh Shapiro signed an executive order requiring AI companies seeking state support to obtain necessary permits, cover their electricity costs, hire and train local workers and comply with environmental requirements.
Shapiro's position is that companies seeking government support must also secure acceptance from the communities where they intend to build.
The significance is the political breadth of the response.
Abbott is a Republican governor. Shapiro is a Democrat. Yet both are confronting essentially the same issue: how much responsibility should enormously wealthy technology companies carry for the local consequences of the infrastructure their businesses require?
More Than 530 Local Governments Reportedly Restrict Development
Resistance is also moving below state level.
The source reports that more than 530 counties and local governments across the United States have banned or restricted data-centre construction.
That local resistance may prove particularly important.
National governments can pursue ambitious AI strategies, but data centres ultimately have to be constructed somewhere. They require land, electricity connections, water arrangements, planning permission and local infrastructure.
The communities supplying those resources therefore possess considerably more influence over the AI boom than the industry's digital image might suggest.
Poll Finds 75% Oppose Data Centres Near Their Neighbourhoods
The political pressure appears to have substantial public support.
An August survey conducted by Embold Research for Heatmap found 75 percent of registered voters strongly or somewhat disapproved of data-centre construction near their neighbourhoods.
Only 15 percent strongly or somewhat approved.
The opposition reportedly crosses conventional political divisions.
That creates a difficult calculation for governments. Data centres promise investment, construction activity, technology development and potentially substantial economic benefits.
But the benefits can be national or corporate while the immediate costs — power infrastructure, water consumption, construction disruption and land use — are concentrated locally.
That imbalance is increasingly becoming the centre of the political argument.
Who Pays for the AI Revolution?
This may ultimately be the most important question raised by America's experience.
AI companies require enormous computing resources. Those resources require electricity generation, transmission infrastructure, water, land and capital.
Someone has to pay for all of it.
If a data-centre developer finances its own power requirements, contributes additional generation to the grid, pays for transmission upgrades and manages its water requirements sustainably, the economic equation looks very different from one in which infrastructure costs are spread across existing households and businesses.
Texas appears to be attempting to establish precisely that distinction.
Rather than rejecting AI development outright, Abbott's approach effectively says the industry can expand — provided communities do not surrender the reliability of their electricity, water or neighbourhood infrastructure to accommodate it.
Does This Affect New Zealand?
Potentially, very much so.
New Zealand is a much smaller electricity system than the United States. Any future expansion of AI-scale computing infrastructure therefore raises questions that should be answered before, rather than after, major projects are approved.
The American experience provides a useful policy framework.
For any significant New Zealand data-centre development, the measurable questions should include how much electricity the facility requires at peak demand, what additional generation and transmission capacity it will finance, how much water it consumes, who pays for infrastructure upgrades, how many permanent local jobs it creates, and what happens when electricity supply becomes constrained.
Those are more useful measures than simply announcing the dollar value of an investment.
There is also a larger strategic opportunity.
New Zealand's renewable electricity resources could potentially make the country attractive for certain forms of computing infrastructure. But that advantage only remains an advantage if additional industrial demand does not increase costs or reduce reliability for households and existing businesses.
America's rapidly developing backlash therefore carries a straightforward lesson.
The argument does not necessarily need to be AI versus communities.
The more important question is whether governments establish the rules early enough to ensure that companies profiting from the AI infrastructure boom pay the full cost of the resources and capacity they require.
The United States is discovering what happens when communities believe that bargain has not been clearly established.
Source: David Lindfield, Slay News, “Thousands of AI Data Center Projects Halted as Backlash Spreads Across America,” 30 August 2026.
Independent reporting. Original context. Credited sources.