The backlash against data centers has more to do with electricity than Big Tech admits

James O'Donnell's article, published in MIT Technology Review (in the edition of The Algorithm newsletter dated November 9, 2026), argues that opposition to AI data centers in the United States is not just a culture war, as the tech industry portrays it, but is largely explained by the electricity they consume. The author writes after spending a few days in Ashburn, Virginia, in the corridor known as Data Center Alley, the world's largest data center hub and, he notes, the place where the backlash now sweeping the country began to take shape.

The starting point is a statement by Chris Lehane, OpenAI's chief global affairs officer, who wrote two days after the midterm elections that data centers "have become a symbol of public anxiety about affordability, corporate power, and people's loss of control over their future." In other words, a culture war. O'Donnell disagrees: after his visit, he concludes that the backlash has more to do with electricity than Big Tech admits. The industry can improve its image, but it cannot escape the physical consequences of its enormous appetite for energy.

According to the author, there are two ways to power the data centers AI companies want: connect to the grid or generate electricity on site. In Loudoun County, Virginia, where data centers are everywhere but have been built at a steadier pace than elsewhere in the country, most are connected to the existing grid. An Equinix facility he visited has large diesel generators as backup but draws its power from the local utility, Dominion Energy, whose electricity comes largely from natural gas (cleaner than coal or oil, but worse than renewables). The average capacity of a data center in that part of Virginia is now 54 megawatts, more than the consumption of Denver International Airport.

Elena Schlossberg, a prominent anti-data center activist in Virginia, says she bangs her head against the wall when the industry frames the fight as a culture war: "it was always about the power." The author acknowledges that some protesters are driven by a broader anti-tech populism, but argues that the concrete consequences of drawing so much power from the grid are what have most shaped the protests. The protests have led the state to require data centers to pay the full cost of new transmission lines, rather than spreading it among other users. Developers told him they are preparing for rules that would disconnect them from the grid during peak demand. Schlossberg also fears indirect effects of a strained grid, such as greater use of diesel generators or Dominion expanding fossil fuels instead of renewables.

Most AI data centers being built elsewhere in the country do not follow this model. The desired scale is so large that companies are setting up in rural areas with almost no grid infrastructure, forcing them to bring their own power ("behind-the-meter" generation) through natural gas turbines, with diesel generators as backup. The article notes that Anthropic has reportedly agreed to pay $3.4 billion for gas turbines, described as the largest purchase of power generation by an AI company. Even if grids were nearby, they could hardly supply the power required: the smallest data centers attractive for AI training are 300 megawatts, large but not unprecedented (one developer told him about a former steel complex already equipped for that load), while the ideal ones are around 6,000 megawatts, the estimated demand of the city of Chicago, which is served by more than 20 power plants within the city itself and dozens more outside it.

Generating their own power may seem attractive because it doesn't strain the grid, but at that scale it inevitably affects neighboring communities. When public utilities build infrastructure of this size, communities expect to have a say in where it goes; now private companies are undertaking equivalent projects. Gas turbines emit pollutants at any scale, and diesel generators even more so. In the rush to get facilities up and running, some companies have become, in the author's words, "bad neighbors": he recalls that, according to the Environmental Protection Agency (EPA), Elon Musk violated air quality laws by installing dozens of turbines to power a data center in Memphis.

For O'Donnell, attributing the backlash to a culture war obscures a simple fact: there is a limit to how pleasant living next to a small city of computers can be made. Companies can negotiate better deals, increase tax revenue, hire local workers and involve the community more, as the protests are already pushing them to do. But asking a small town to host the equivalent of a Walt Disney World, an international airport, a steel mill and the power plants to run them should come as no surprise if it breeds mistrust.

The author notes that electricity is not the only cause of the opposition. Fear of job losses, existential risks, the concentration of corporate power and misinformation (most facilities use far less water than people believe) also play a role. Nor does he consider consuming large amounts of electricity bad in itself: many advances of modern life have required it. But one developer told him the industry sees its ambitions constrained less by public opposition than by a lack of electricity, something that, according to O'Donnell, should give us pause.

As for what comes next, the author concedes that newly elected politicians may not be as opposed to data centers as they were on the campaign trail. But if new data centers need their own generation, they will go where fewer people are able to resist, such as the Permian Basin, with an easily tapped pipeline network and a small population. They could also try their luck abroad: the day after the elections, Google announced a $15 billion investment in data centers in Malaysia. And there is one place where Big Tech might find no limits: a few hours after the election results, Elon Musk posted that "data centers in space look pretty attractive right now."