AI and Energy: This Week's Data Center Backlash

This week's feature goes deep on the regulatory turn in Texas and Seattle. Here is everything else that moved in the AI and energy story: the polling that turned the politics, the water map driving the fights, a reality check on what data centers actually do to your electric bill, and the demand curve that has utilities rewriting their rate books.

---

Public Support for the Data Center Boom Has Collapsed

A Reuters/Ipsos poll released June 11 put hard numbers on a mood that politicians have been reading for months. Just 33% of Americans agreed that building data centers at a rapid pace is mainly a good thing, while 64% disagreed. Only 14% said they would be comfortable with one built near them, and 57% said they would oppose a nearby facility outright. The opposition is not partisan noise: 77% of respondents, in similar shares across Republicans, Democrats, and independents, said they worry artificial intelligence will make electricity more expensive.

The survey of 4,531 people ran for six days and closed June 9. Reuters reporters Valerie Volcovici and Jason Lange tied the result directly to the November midterms, noting that data center construction has become an issue campaigns are now running on. Roughly half of those polled also worried AI could eliminate jobs in their household, so the electricity-cost anxiety sits inside a broader unease about what the AI buildout does to ordinary people.

For anyone tracking the environment and sustainability politics of AI, the poll is the clearest signal yet that the energy-and-water cost of artificial intelligence has crossed from a specialist concern into a kitchen-table one. That shift is what gives local moratoriums and state directives their political cover. When opposition to a nearby data center runs at 57% and crosses party lines, a city council freeze or a governor's directive stops being risky and starts being popular. The numbers also help explain why industry groups have shifted from promoting jobs and investment to defending the sector's basic legitimacy. The question of whether AI is sustainable, in the literal sense of whether communities will keep tolerating its footprint, is now a polling question, and the early answers are not friendly to unrestricted growth.

Source: Reuters / Ipsos

---

Most New US Data Centers Are Headed for Drought Country

An analysis published June 8 by The Guardian mapped where the AI buildout is actually landing, and the picture sharpens why water keeps showing up alongside energy in every new data center fight. About two-thirds of the 809 data centers planned across the United States, roughly 517 facilities, are slated for areas that were in drought at some point over the past year. The clustering is heaviest across the West, the South, and the interior, the same regions where water supply is already stretched.

The overlap is not an accident. Developers chase cheap land, fast permitting, and available power, and those incentives keep pointing them toward dry, low-cost regions where adding a facility that can draw millions of gallons a day collides with communities that have none to spare. The AI environmental impact on water has tended to get less attention than the energy headline, partly because water draw is harder to see and was rarely reported. That is exactly the gap the new disclosure mandates aim to close.

The siting data reframes the whole AI and energy conversation as an AI and water conversation in much of the country. A facility's thirst is not a rounding error: large sites can consume up to millions of gallons daily for cooling, and concentrating them in drought zones turns a national compute story into a local resource conflict. For sustainability and water-utility planners, the map doubles as a forward indicator. The drought-heavy regions with the densest data center pipelines are where the next round of permitting battles, rate disputes, and usage-reporting fights will likely break out. It also explains why water-efficient cooling, rather than carbon alone, has become a front-line demand in the new wave of state proposals.

Source: The Guardian

---

A Fact-Check Complicates the "Data Centers Are Spiking Your Bill" Story

Not every claim in the backlash survives scrutiny, and a June 12 PolitiFact analysis by Caleb McCullough is a useful corrective for anyone repeating viral cost figures. Senator Elizabeth Warren had cited a 267% increase to argue that data centers are driving up consumer electricity bills. PolitiFact rated the framing Mostly False, and the reason matters for how the energy-and-AI debate gets argued.

The 267% figure referred to wholesale electricity prices at grid nodes near data centers between April 2020 and April 2025, not to residential bills. Wholesale power is only 30% to 50% of a typical consumer's bill; the rest is transmission, distribution, and taxes. Over five years, the national average residential electricity cost rose about 42%, with sharp regional variation: prices climbed 94% in Washington, D.C., 74% in Maryland, 73% in Maine, and 58% in New York between March 2021 and March 2026.

The fact-check still leaves data centers on the hook. PolitiFact noted that PJM's independent market monitor attributed a 174% increase in capacity costs for the 2025-26 delivery year to data center demand. The honest version is more specific than the slogan: data centers are measurably pushing up the wholesale and capacity components of power costs in the regions where they cluster, and those costs do reach consumers. The headline percentages in circulation, though, conflate wholesale spikes with the final bill. For environment and sustainability advocates, the lesson is practical. The case for metering and regulating AI's energy appetite is strong enough on the real numbers to stand without inflated ones, and overstated claims hand the industry an easy rebuttal.

Source: PolitiFact

---

The Demand Curve That Has Utilities Rewriting the Rate Book

The regulatory scramble makes more sense against the raw demand math. Gartner forecast on June 10 that data center electricity consumption will grow 26% in 2026, a single-year jump that lands on grids already running tight in summer. That growth is the most important driver of new electricity demand in the United States, and it is arriving faster than utilities can build generation and transmission to match.

Utilities are responding by changing how large computing loads are priced. On June 12, Seattle City Light detailed a new rate class for data centers requesting new or expanded service of 10 megavolt-amperes or more, roughly the draw of 2,000 homes. The structure is built so those customers pay the incremental cost of procuring and delivering the power they require, take full financial responsibility for the related infrastructure over the life of the contract, sit in a service queue tied to supply availability, and invest in conservation and demand-response measures. The utility's stated reasoning was blunt: existing residents and local businesses funded the grid over decades and should not absorb the cost of large new loads that could relocate later and strand the investment.

This is the market-side companion to the legislative action. Where lawmakers debate moratoriums and reporting mandates, utilities can move now by simply repricing service so the artificial intelligence energy boom funds its own grid. Special large-load rate classes are spreading across US utilities for the same reason, and they may end up shaping the AI buildout's environmental footprint as much as any statute, by making the cheapest siting decisions more expensive and pushing operators toward efficiency, on-site generation, and locations with real power headroom. For sustainability teams, rate design is becoming a quiet lever on where and how much AI infrastructure gets built.

Source: Gartner; Seattle City Light

---

Sources

• Americans wary of AI-driven data center boom, Reuters/Ipsos poll shows, Reuters / Ipsos
• Majority of US's new AI datacenters to be built on drought-hit land, The Guardian
• How much have data centers increased electricity prices? (Warren fact-check), PolitiFact
• Gartner Says Data Center Electricity Consumption to Grow 26% in 2026, Gartner
• Getting Ahead of Data Center Power Demands, Seattle City Light

Related: AI and the Environment: Texas Makes Data Centers Pay