The thing actually alarming Governor Hochul isn’t that data centers exist. It’s that their business model assumes the public will build the wires and pipes they need, while they keep the tax break. That’s not Luddism. That’s arithmetic. The editors of National Review took the opposite position in “Hochul’s Silly War on Data Centers”, calling her one-year moratorium on large new data centers a “concession to Luddism and hysteria.” They argue the cloud runs on these buildings, that they’re no different from a steel mill, and that the panic over their energy and water use is mostly myth.
Let me concede what the editors have right: data centers are the infrastructure of the modern economy. Banking, streaming, cloud computing, e-commerce, communication—none of it works without them. The internet lives in those buildings. But that is exactly why the cost question matters more, not less.
Then the editors do the thing that makes their whole case wobble and fall over. They name the tax break—New York’s “generous sales and use tax exemption for internet data centers, a subsidy from the early 2000s” that Hochul wants to repeal—and they call it “unfair to taxpayers” and something that “has no place in a competitive free market.” Good. They know the subsidy exists. They know it’s anti-competitive. They call the tax break unfair. They never follow through on what that unfairness means. You cannot simultaneously admit the industry is built on a special tax handout and then pretend the only thing standing in its way is irrational fear. The handout is the tell. If this were a straightforward industry paying its own way, the state wouldn’t have had to waive the sales tax to get it to show up.
The real engine they’re not naming is the grid. A data center that draws 50 megawatts doesn’t just plug into the wall. It demands new transmission lines, new substations, and, often, new generation. Those costs are not small, and the question of who pays them is the entire fight. The editors mention Hochul’s proposal to make new facilities contribute to a grid-improvement fund—an “ask that, depending on the details, may be necessary,” they allow, gently. But they treat this as a minor side note, not the central structure of the problem. The overwhelming default in American utility planning is that the cost of new capacity gets rolled into the rate base, spread across every residential and small-business customer on the system. So a hyperscale data center negotiates a sweetheart power deal, the state forgives the tax on its equipment, and then your aunt in Binghamton sees her monthly bill tick up to pay for the transmission upgrade. The data center gets the computing; the ratepayer gets the surcharge. That isn’t a myth. It’s the line item.
They say utilities “have long planned for the arrival of energy-intensive industries.” That sounds reassuring until you ask what the planning actually consisted of. Utilities plan for gradual load growth they can model from demographic trends and building permits. The current wave of AI-driven data-center proposals is not gradual. It is enormous, it is arriving faster than the lead time for a new substation, and it is landing in places—Upstate New York, for example—that haven’t seen load growth in decades. The local grid was not built for an aluminum smelter’s worth of demand to appear in eighteen months, and no amount of prior planning changes the fact that somebody has to pour the concrete and string the cable. The question is whether that somebody is the company whose business model created the demand or the person who just wanted to keep the lights on.
Then there’s the water. The editors dismiss the concern by pointing at golf courses and suburban lawns. The comparison is a dodge. A golf course doesn’t suddenly increase its draw by an order of magnitude in a single county, and nobody is proposing to build dozens of new golf courses that each use millions of gallons a day for cooling. The scale is the story. A large data center can consume as much water as a city of 50,000 people. In a place like Oswego County, where the water system already struggles with aging pipes and summer drought, a new 100-MW campus drawing millions of gallons a day isn’t a hypothetical. You don’t have to be a Luddite to notice that there’s only so much groundwater in the bucket.
The editorial ends by scolding Hochul for driving people to Texas and Florida. The job numbers they’re so confident about deserve a closer look. A hyperscale data center might employ a couple hundred people once it’s running, mostly in security, facilities maintenance, and a small IT staff. The construction jobs are temporary. The permanent employment per megawatt is tiny compared to almost any other industrial facility you could name. The tax revenue, once you net out the sales-tax exemption and the abatements and the incentives the state and localities throw at the project, is often a rounding error. The main thing the community gets, in exchange for the land and the water and the transmission lines, is a very quiet, very power-hungry neighbor that employs very few of its residents. That may still be a deal worth making, but you make it worth making by sitting at the table and asking who pays for what—not by calling the governor a Luddite for trying to get a read on the bill.
The constructive thing to do isn’t to ban every data center forever. It’s to make the cost structure honest. End the sales-tax giveaway that the editors themselves call unfair. Require that the cost of any new transmission and distribution infrastructure built to serve a single large industrial customer be charged to that customer, not socialized across the rate base. Make the water-impact studies public and the mitigation binding. And let communities negotiate. If the industry is as strong as the editorial says, it doesn’t need the taxpayer to carry its water—literally or figuratively. A deal where the company keeps the profit and the public carries the risk is not the free market. It is the free lunch, and the rest of us are picking up the check.