As millions of Americans head to the polls today for the 2026 midterm elections, AI infrastructure has emerged as a political flashpoint that cuts across traditional party lines.
While President Donald Trump and national Republican leaders have pushed an aggressive buildout to secure American AI dominance, local candidates from both parties face severe constituent pushback over power-hungry data centres.
Democratic candidates have increasingly made data centre moratoriums and tighter development rules part of their campaigns.
The divide comes as candidates also face pressure to address concerns over electricity prices, water consumption and the costs imposed on local communities, making data centre expansion an increasingly bipartisan political issue.
Thousands of miles away from the ballot boxes, at the Smart City Expo World Congress in Barcelona, two US mayors sat alongside tech giant Nvidia to address this domestic anxiety head-on.
“Public trust is something you have to build and maintain every single day,” said Jane Castor, Mayor of Tampa.
A March Gallup survey found that 71% of Americans opposed building AI data centres in their local area, including 48% who were strongly opposed.
Data centres require large amounts of electricity and water, and residents in several US communities have challenged developments over their potential impact on utility bills, local resources and quality of life. Data Center Watch estimated that at least 75 projects worth around $130bn were blocked or delayed by local opposition in the first quarter of 2026, roughly matching the value of projects disrupted throughout 2025.
An October study by USC Annenberg found that nearly a third of Americans ranked AI and data centres among the country’s three most important issues. Some 77% thought the government was doing too little to regulate data centres, while 71% said they did not trust Big Tech at all to act in society’s best interests when developing them.
Clear AI policies, staff training and accessible information were among Castor’s recommendations for increasing public trust in technology. She described Iris, a system developed in Tampa, as a dashboard that brings together information from services including water quality, wastewater, solid waste, purchasing and permitting for municipal staff. She said residents also needed to be told what information the city collects and how it is used, particularly where surveillance and law enforcement are concerned.
The challenge for cities is that the technology can move faster than the processes required to buy and implement it. “Municipalities don’t move at the speed of light,” Castor said. She explained that procurement and design can take so long that technology is already becoming outdated by the time a city is ready to deploy it.
West Palm Beach Mayor Keith James pointed to a more specific use case. His city is exploring AI to help process public records requests, which can create a substantial administrative workload. He said the project had been welcomed by staff because it could make the process more efficient and help the city respond to residents more effectively.
“The first lesson is finding the problem,” James said. He urged other local authorities to identify operational pain points before deciding where AI might be useful.
The infrastructure challenge
AI systems require substantial computing capacity, which means more data centres and greater demand for electricity and water. Cities looking to expand their use of AI face difficult questions about where that capacity should come from, how much they need, and who bears the environmental and financial costs.
Charbel Aoun, Nvidia’s Smart City & Spaces Director, acknowledged these friction points by outlining three recommendations for cities: right-sizing computing capacity to actual needs, locating facilities where waste heat can be recovered for district heating and other local uses, and establishing shared “centres of excellence” to pool expertise and computing resources rather than allowing departments to develop disconnected systems independently.
Aoun noted that municipal trials frequently fail to deliver lasting utility. “Most pilots end up being a lessons-learned paper that doesn’t go into full production,” he said, arguing that cities should focus on applying shared computing foundations to solve multiple administrative issues sequentially.
As municipalities increasingly turn to technology providers such as Nvidia for the hardware and software underpinning their AI ambitions, questions over vendor oversight remain contentious. Addressing what cities should demand – and withhold – from private vendors, Aoun stated that commercial providers should be treated strictly as suppliers of computing power rather than decision-makers.
“What you should not hand over is decision-making and control,” Aoun said. “Your data, your skills, the relationship with your citizens, that must stay with the city. It has to be by the city, for the city.”
The midterms will determine control of Congress for the final two years of Donald Trump’s presidency, but on the ground in Barcelona, the focus was on local decisions.
The discussion in Barcelona offered a glimpse of the choices facing municipal leaders. Cities want to use AI to improve services and attract talent, while managing the cost of computing infrastructure and retaining control over decisions that affect residents. The technology may be advancing quickly, but local government still has to bring the public with it.
