Rahm Emanuel published a Wall Street Journal piece this week arguing that Democrats can learn from Texas on energy, invoking concerns about tech billionaires and hyperscalers driving up electricity costs for ordinary ratepayers and warning that the data center boom represents politically unsustainable hunger games.
Mark Mills, founder and executive director of the National Center for Energy Analytics, joined Dan Proft on Chicago’s Morning Answer to address the data center resistance movement, the electricity cost question, the nuclear parallel, and the specific claims about water usage that are fueling local opposition to data center siting across the country.
Mills said the central claim driving data center resistance, that these facilities are responsible for rising electricity costs, is simply not supported by the data. The data centers currently being built have not been built yet. They are not on the grid. They are not consuming electricity at scale. Attributing current electricity affordability problems to data centers is like blaming gasoline expenditures on a car that is still in the factory. What has actually driven electricity costs higher in state after state is almost entirely the result of state and federal mandates and subsidies pushing more solar and wind onto the grids, which increases operating costs, requires more transmission infrastructure, and creates reliability problems that drive further expense. The causation is being deliberately misattributed, he said, and what the politicians promoting the data center narrative are really hoping to do is harvest tech company money to cover up the mistakes they have already made in energy markets.
On the specific numbers, he cited tech entrepreneur David Sacks’s back-of-the-envelope calculation that a one-gigawatt data center requires roughly fifty billion dollars in capital expenditure and generates twenty-five to thirty billion dollars in annual revenue, with electricity costs running one to two billion dollars a year, meaning electricity represents approximately five percent of operating costs. Mills said the revenue estimate is slightly optimistic but the framework is correct, and it reveals something important: data center operators are essentially insensitive to electricity prices. What matters to them is not whether power costs five percent or ten percent of revenue, but whether they can get power at all. This explains why somewhere between a third and a half of data centers currently under construction are being built as private grids disconnected from the public grid entirely, mostly powered by natural gas turbines. The bad news for politicians hoping to force the hyperscalers to subsidize their past policy mistakes is that companies building private grids have no obligation to fund public grid remediation.
On the nuclear parallel, Mills said this is the same fight with the same actors. The anti-Vietnam War movement of the 1960s and 1970s steered its political machinery toward environmentalism and especially anti-nuclear activism, and succeeded in essentially killing the domestic American nuclear industry. No new nuclear power plants were built in the United States over the past two decades except for two in Georgia, completed at great expense after extraordinary effort by Georgia Power. Meanwhile China, using technology pioneered in the United States, is now building more nuclear power plants than any other country in the world. The same environmental movement still opposes nuclear energy. They have not changed. And the data center resistance movement is, he said, running the identical playbook, using similar fear tactics around water, electricity costs, and community character to stop the construction of infrastructure that the country’s economic future depends on.
On water usage concerns, which Mills said are generating significant anxiety at the local level, he said the claims being made are not merely wrong but egregiously wrong. Data centers do not have to net consume water because they can use closed-loop cooling systems that recirculate rather than consume. More importantly, the agricultural land on which many data centers are being sited uses vastly more water than any data center would. He cited the specific case of Kevin O’Leary’s Utah data center project, where agricultural irrigation of the same land uses approximately two hundred times more water than the planned data center will consume. If water table preservation is the genuine concern, the focus should be on agricultural water use, not data centers. Several hyperscalers have already recognized this and are investing in local water projects that would make them net creators of clean water for surrounding communities rather than net consumers.
He acknowledged that the populist appeal of data center resistance should not be dismissed. People who live near large construction projects have real concerns about aesthetics, noise, construction traffic, and community character. Nobody wants to be told they are being manipulated by bad actors, even when they are. He said the nuclear industry faced identical social license challenges decades ago and the data center industry is going to have to take local community engagement seriously rather than simply labeling opponents as Chinese Communist Party assets, as O’Leary has done. That framing may be partially accurate but it is not persuasive to someone with a valid quality of life concern, and the industry cannot afford to repeat the errors that allowed the environmental movement to kill nuclear development for forty years.