Data center backlash is more about electricity than Big Tech admits, MIT Technology Review writer argues

The piece appeared in The Algorithm, MIT Technology Review's weekly AI newsletter. Its author spent a few days in Ashburn, Virginia, in Data Center Alley, described as the world's largest data center hub and arguably the place where the backlash now sweeping the country first took shape. The visit coincided with voting in the US midterms.

The argument is a response to what the author calls accepted wisdom in Big Tech. Two days after the midterms, OpenAI's chief global affairs officer, Chris Lehane, wrote that data centers "have become a proxy for public unease about affordability, corporate power, and people losing control over their futures." In other words, a culture war. The author came away convinced instead that the backlash is more about electricity than Big Tech admits. The industry can improve its image, the piece says, but it cannot escape the physical consequences of its appetite for electricity.

The author splits the problem in two: a data center can connect to the grid or generate power onsite. In Virginia's Loudoun County, where data centers are everywhere but have been built at a steadier pace than the breakneck expansion elsewhere, most are grid-connected. One Equinix facility the author entered has massive diesel generators for backup but takes its power from the local utility, Dominion Energy, much of whose electricity comes from natural gas. The average data center operating in this part of Virginia today has a capacity of 54 megawatts, more than the power draw of Denver International Airport.

Elena Schlossberg, a prominent activist against data centers in Virginia, told the author she is "always banging my head against the wall" when the industry frames the fight as a culture war. "It was always about power," she says. The author concedes that some Virginia protesters seem more animated by broader anti-tech populism than Schlossberg is, but says the particular consequences of drawing so much power from the grid have done the most to shape how the backlash has played out. Protests led to new state rules that data centers must pay fully for the cost of new transmission lines to the grid, so the cost is not spread among everyone else. Developers told the author they are preparing for new rules that would cut them off from grid power during peak demand. Schlossberg also worries that a strained grid will mean more diesel generator use, or that Dominion will expand fossil fuel inputs rather than renewables.

Elsewhere the picture differs. The overwhelming majority of AI data centers being built outside Loudoun are not designed to use the grid. The scale wanted is so large that companies are going to rural areas with virtually no grid infrastructure, and bringing their own "behind the meter" power from natural-gas turbines, with diesel generators as backup. Anthropic has reportedly agreed to pay $3.4 billion for gas turbines, described as the largest power generation purchase by an AI company. Even with a grid nearby it would struggle: the smallest data centers attractive for AI training are 300 megawatts, which is not unprecedented for an industrial site (one developer is eyeing a former steel mill already equipped to draw that much from the grid), but the ideal facilities are more like 6,000 megawatts. That is the estimated draw of Chicago, which is powered by more than 20 power plants within the city alone and dozens farther out.

Onsite generation might look attractive because it spares the grid, the author says, but generating power at this scale affects surrounding communities, and communities have traditionally expected a say when public utilities build infrastructure of this size. Natural-gas turbines of any scale release air pollutants, and diesel generators are worse. In the rush, companies have become horrible neighbors; the author recalls that Elon Musk violated air quality laws, according to the Environmental Protection Agency, when he set up dozens of turbines to power a data center in Memphis. The author's image: if a small town is asked to welcome a Walt Disney World, an international airport, a steel plant and the power plants to run them all, should anyone be surprised by its trepidation? Companies can offer better deals, with more tax revenue, local hiring and community involvement, as the protests are pushing them to do, but there is only so pleasant you can make a small city of computers to live around.

The author adds limits. Electricity is not the sole reason for opposition: fears of job losses, existential risks and concentrated corporate power all fuel the backlash, as does misinformation (most facilities use far less water than many people think). Consuming a lot of electricity is also not inherently bad. One developer told the author that industry insiders see their ambitions as constrained less by public opposition than by a lack of electricity.

Looking ahead, the author says newly elected politicians' actual stances on data centers may be less negative than those they campaigned on. Regardless, data centers that need onsite power will go where the fewest people can resist them, possibly the Permian Basin, where natural-gas lines are easy to connect to and populations are sparse. Companies might also go abroad: Google announced the day after the elections that it will spend $15 billion on data center projects in Malaysia. The piece ends with Elon Musk, who posted hours after the midterm results: "Data centers in space looking pretty appealing right now."

Key facts

The author argues the data center backlash is more about electricity than Big Tech admits, disputing OpenAI's Chris Lehane, who called data centers a proxy for public unease about affordability, corporate power and loss of control.
In Loudoun County, Virginia, most data centers use the grid; the average one operating there has 54 megawatts of capacity. Protests produced state rules that data centers pay fully for new transmission lines, and developers say they expect peak-demand cut-off rules.
Most AI data centers built elsewhere bring their own gas-turbine power. The smallest attractive for AI training are 300 megawatts and the ideal ones about 6,000 megawatts, the estimated draw of Chicago; Anthropic has reportedly agreed to pay $3.4 billion for gas turbines.
Onsite generation brings air pollution (turbines release pollutants, diesel is worse) and community pushback; the author expects such sites to go where the fewest people can resist, such as the Permian Basin, or abroad, as with Google's $15 billion Malaysia plan.
The author says electricity is not the only cause: job fears, existential risk, corporate power and misinformation also feed the backlash.

Why it matters

The piece challenges the industry's preferred story that opposition to data centers is a culture war over progress. If the author is right, public relations cannot fix it, because the friction comes from the physical load: either strain on the grid and its costs, or turbines and generators built next to communities. The author also notes that insiders see their ambitions as limited more by a lack of electricity than by public opposition. The scale makes the point concrete: a typical Virginia data center draws 54 megawatts, while the ideal AI site is about 6,000 megawatts, the estimated draw of Chicago.

Who it affects

Residents near data center sites are the central group: grid-connected neighbors in Virginia face transmission costs and grid strain, while rural communities elsewhere face gas turbines and diesel backup generators. It also affects AI companies and developers, who the author says are preparing for rules that could cut them off from grid power at peak demand and who may move to sparsely populated places such as the Permian Basin or abroad. Utilities such as Dominion Energy, whose electricity is largely gas-fired, and state regulators who set the cost rules are also in the picture.

How to use it

This is an analysis, not a product or tool. As a lens, it suggests asking of any data center project whether it draws on the existing grid or brings its own power, since the piece says each route provokes a different kind of backlash. For grid-connected sites the issues it names are who pays for new transmission lines and what happens at peak demand; for onsite sites they are air pollution and the community's say over where large power infrastructure goes.

How solid is it

This is a reported opinion piece from MIT Technology Review's weekly AI newsletter, The Algorithm, built on a few days in Ashburn, Virginia, plus conversations with an activist (Elena Schlossberg) and unnamed developers. The central conclusion is the author's own judgment ("I came away convinced"), not a measured finding. The Anthropic $3.4 billion turbine purchase is labelled "reportedly". The Musk air quality claim is attributed to the Environmental Protection Agency. The author names no specific developer behind the steel mill idea or the electricity-constraint remark, and gives no date for the expected peak-demand rules.

Risks and caveats

The author says electricity is not the sole cause of opposition: job losses, existential risks, concentrated corporate power and misinformation also play a part, and some Virginia protesters seem more driven by broader anti-tech populism than Schlossberg. Heavy electricity use is not inherently bad either. The predictions are speculation: newly elected politicians may be less negative on data centers than their campaigns suggested, and onsite-powered sites may drift to places like the Permian Basin. Musk's space post is quoted as a closing note; the source does not say it was serious or part of a plan.

“It was always about power”

— Elena Schlossberg, a prominent activist against data centers in Virginia