Over 230 environmental and community organizations are calling for a national pause in approving new data centers, saying the rapid buildout to feed artificial intelligence and crypto is leapfrogging energy and water systems while driving up residents’ bills.
In a letter sent to federal lawmakers, the groups, which include Friends of the Earth, Food & Water Watch, and Greenpeace, are calling for a halt on permits and building until stronger standards govern electricity use, water consumption, air pollution from backup generators, and community protections. They cast the surge as a public-interest issue, not just a tech sector story.
Why Campaigners Are Calling for a Pause
The coalition points to a growing body of evidence that dense server farm clusters are straining power grids and driving up prices. Consumer anxiety is also increasing: a recent Sunrun-commissioned survey of homeowners found a large majority are concerned that local data centers will drive up their utility bills. Retail electricity prices increased by about 13 percent over the last year, the largest annual surge in a decade.
Analysts predict that power demand from U.S. data centers will more than double by the mid‑2030s to over 100 gigawatts on industry estimates (from about 40 gigawatts of load today). A lot of the new projects are aimed at Mid-Atlantic states like Virginia, Pennsylvania, Ohio, Illinois, and New Jersey — states with easy access to the grid, land to build on, and tax incentives — but many of these projects are also moving into rural counties where infrastructure is more fragile and water resources are more precious.
Water is a flashpoint. Operators increasingly promote air‑cooled or recycled-water systems, but hyperscale facilities may still use millions of gallons a year, depending on the climate and cooling design. In public sustainability reports, Google reported using over 5 billion gallons of water in a recent year across operations, and Microsoft reported using more than 1.7 billion gallons — demonstrating the scale even as companies push for efficiency gains.
The groups are after a few things at the federal level: binding 24/7 carbon‑free procurement, rather than annual matching; hard caps on runtime for diesel generators; transparency around real‑time energy and water use data; heat‑reuse analysis for district energy projects; and prioritized siting rules that would see brownfields and grid-ready locations favored over greenfield development.
The Strain Is Evident in Communities Near Data Hubs
Protests have ramped up where multigigawatt projects pursue approvals. In Detroit, protesters gathered outside the utility DTE as regulators consider a proposal to feed OpenAI and Oracle’s proposed 1.4‑gigawatt complex. Residents raised concerns about bills, use of freshwater, and traffic; the utility stressed economic development and new generation plans.
In Wisconsin, a city’s common council meeting discussing the formation of a 902-megawatt site associated with OpenAI and the Oracle-led Stargate initiative turned “ugly,” prompting arrests. Northern Virginia — home to “Data Center Alley” — has been struggling for years with noise from backup generators, land-use fights near residential areas, and frequent warnings from Dominion Energy about transmission bottlenecks when clusters outpace expansions.
Governments outside the U.S. have attempted to apply brakes in the past. Ireland’s regulator essentially curbed new hookups around Dublin until grid constraints ease, and Netherlands officials briefly put a hold on new hyperscale builds as part of developing a national strategy — examples that activists point to when arguing that some type of managed pause can direct smarter siting without permanently freezing growth.
Industry Pushback and the Reality of the Grid
Data center operators and their customers caution that a blanket moratorium could thwart productivity gains in AI, cloud services, and digital infrastructure. Companies point to long‑standing efficiency metrics such as improvements in power usage effectiveness (PUE), increasing extension of power purchase agreements for wind and solar, and the proliferation of 24/7 carbon‑free goals. Google has committed to 24/7 clean power by 2030, and Microsoft’s goal is to be carbon negative during a similar timeframe, with pilot projects that include heat reuse and advanced cooling.
But grid planners face a timing complication. It can take a decade or longer to permit and build transmission lines. Interconnection queues in places like PJM and ERCOT are backed up, and even when new renewable plants are ready to go, congestion can limit their ability to produce output. EPRI and other researchers have warned that data center load is growing fastest where the grid is already tight, exacerbating the issue.
Some utilities are testing demand-flexing strategies — moving AI training to off-peak hours, installing their own on-site storage, and hooking into grid-enhancing technologies to wrangle additional room out of existing wires.
Environmental groups contend that voluntary measures and matching annual renewables will not ensure the hourly reductions in emissions that communities require, particularly during heat waves, when peaks are pushed by buildings and servers alike.
What Comes Next for Data Centers, Communities and Policy
The moratorium request pressures lawmakers and federal agencies to draw up guardrails: uniform impact assessments on water and air, emissions performance standards for megawatt-sized loads, and siting incentives that prioritize grid‑ready zones. State utility commissions will be central to those deliberations as they balance long‑term resource plans and connection approvals linked to single campus loads that can rival entire towns.
The deeper question isn’t whether computing is growing but how and where. A temporary hold, the groups argue, would require projects to demonstrate that they add firm clean capacity, conserve finite water resources, and bring local benefits before shovels hit dirt. Whether policymakers choose to slam the brakes on or to impose a more stringent case‑by‑case regime will determine just how fast the next wave of AI infrastructure comes online — and who picks up the tab.
