Data Centers Are a National Issue

America needs to resolve its data center debate. Cities, counties, and states have delayed or even banned additional data center development. Most recently, Governor Kathy Hochul of New York signed an executive order imposing a year-long ban on large-scale data center construction. Other states seem poised to follow. 

There are political tailwinds behind politicians who don the cape of staunch data center opponents. Nearly half of Americans (48 percent) are strongly opposed to a data center being built in their community. Another quarter (23 percent) are somewhat opposed. In a political climate that rewards short-termism and appeasement to the loudest voices, it’s easy to see why few lawmakers are standing up for what, ultimately, is a nationwide infrastructure project. 

That’s a stance that clashes with America’s historical ability to rally behind long-term infrastructure build-outs that support jobs, drive economic development, and advance national security. The nation has long benefited from forward-thinking leaders daring their political community to invest in the future. Governor DeWitt Clinton of New York did just that in the early nineteenth century. 

Then, as is the case now, the nation sat at the edge of a new technological era. Realization of the full potential of steam power, for example, required substantial investment and development across the nation. As U.S. Representative John C. Calhoun noted, “Let us, then, bind the republic together with a perfect system of roads and canals. Let us conquer space.” Yet, when Clinton went to D.C. to ask for funds, he came back with empty pockets.

Clinton refused to take no for an answer. He persuaded his fellow New Yorkers to join together to build one of the most important infrastructure projects in the nation’s history. Many of those officials likely never realized the full benefits of that project. Nevertheless, they took the long view, and the entire nation was better off as a result.

Governor Hochul and others have embraced the opposite stance. They have decided to see data center development as a state issue. The first finding in her executive order details this faulty premise: 

WHEREAS, New York State is experiencing unprecedented growth in demand for data center development, driven by the need for fo computing infrastructure that supports artificial intelligence (AI), cloud computing, streaming services, and other computing operations.

This is not solely a New York question. At a time when data centers have become targets in international conflicts and a substantial part of our economy, it’s wrong to view this policy question as one that’s exclusive to certain communities and states. If New York does not build data centers, they will be built elsewhere. Whatever harm (justified or not) animated by Governor Hochul will be imposed on others. 

That’s exactly what our founders sought to avoid. Alexander Hamilton, another New Yorker, warned:

The interfering and unneighborly regulations of some States, contrary to the true spirit of the Union, have, in different instances, given just cause of umbrage and complaint to others, and it is to be feared that examples of this nature, if not restrained by a national control, would be multiplied and extended till they became not less serious sources of animosity and discord than injurious impediments to the intercourse between the different parts of the Confederacy.

A nationwide approach to data centers, rather than one in which a single state can opt out of a shared project, must be the path forward. Yet, a national path forward will not be reached so long as those in staunch opposition and those in fervent favor of data centers refuse to compromise.

The facts of this debate are in many ways settled. On jobs, this infrastructure opportunity carries tremendous potential to connect numerous types of workers with meaningful employment. The International Brotherhood of Electrical Workers urged its members to stand up against data center bans, warning that they threaten to "eliminate good-paying union construction jobs." A representative of the Virginia State Building Trades echoed that sentiment in an interview with Politico, contending that "this is good work, strong work, and it's part of essential infrastructure." Opponents of data centers counter that these jobs are often temporary and, therefore, not worth the costs that data centers may otherwise impose on a community. Yet, that's precisely how construction work has always been, as noted by the president of the Maine Building & Construction Trades Council. 

More generally, on economic development, data center construction can lead to many more positive economic outcomes. When communities negotiate with data center developers, they may secure significant tax revenue that leads to immediate and transformative benefits. One Louisiana community, for instance, managed to provide its teachers with massive bonuses thanks to data center development. What’s true for that community is not necessarily true for others. Researchers at the Brookings Institution found that communities that rush into short-term, one-off deals may often leave some economic benefits on the table. Those that instead take a longer-term, regional approach tend to fare better.

Regarding environmental concerns, data centers generally have minimal water needs. Andy Masley has extensively researched the issue and concluded that "the idea that [data centers] use an inordinate amount of water that merits any kind of boycott or national attention as a unique environmental issue is innumerate." Myriad other industries should be at the top of the agenda for those concerned about the nation's water supply. For example, data center water consumption is just three percent of the water required for America's golf courses. In some cases, data centers may be constructed in areas with less plentiful water or a less efficient water infrastructure; those cases may warrant heightened concern. Crucially, data center building will likely occur in such areas where states like New York foreclose development in sections that may be better suited to more sustainable data center infrastructure. 

On electricity prices, Masley and others note that this is a valid concern for many Americans. However, it’s often the case that the potential for increased prices is a product of policy frameworks established well before the current AI moment; the surge in data centers is merely putting pressure on poorly devised laws and outdated grids. Nevertheless, Americans have plenty of sound arguments to make for protections against rate increases, as affordability is a concern for many households. 

In summary, the nation needs more data centers. Different locations present different cost-benefit analyses for data center construction. States opting out of this national project increase the odds for projects being built in areas that may be less well-suited to such projects or may lack the political will and capital to push back on such construction. That’s precisely why this generational infrastructure conversation should occur at the federal level. It’s also why the entrenched coalitions that have formed around this debate need to compromise. A status quo in which governors unilaterally foreclose their states from data center construction will not serve the nation’s long-term interests.

Three such coalitions have emerged: absolutists in opposition, realists for a path forward, and purists in favor. Absolutists in opposition ground their firm stance in concerns over water usage, land use, and electricity rates. Purists in favor remind people that data centers have been and must be part of our growing digital economy. The realists accept that the gap between those two groups will not be closed through spontaneous consensus on the aforementioned issues. 

Realists must now present tangible, viable legislative frameworks for Congress. As others and I have proposed elsewhere, one possible path may look like a cap-and-trade program for data center development. A quota for the total amount of new data center capacity could be established annually. That quota should not be set by whoever holds executive power at a given moment. An independent national commission that is also a statutory body with fixed membership rules, conflict-of-interest prohibitions, and a mandate insulated from the whims of the next administration would establish the total amount of new capacity for each of the next five years. Industry representatives would hold seats on the commission, but so would environmental scientists, public health researchers, and representatives of the rural and low-income communities most likely to host this infrastructure. Quotas would be grounded in documented infrastructure needs rather than industry projections, which have a track record of overstating demand and understating cost.

The commission would then divide that national quota into regional caps. No single state would absorb a disproportionate share of the build-out. A transparent formula would account for grid capacity, water availability, and, crucially, cumulative burden: the degree to which a region has already shouldered the externalities of prior national economic projects. A cap-and-trade approach borrows a lesson from environmental law: when a national project imposes localized costs, the fairest response is to cap the total burden, distribute it deliberately, and let those who bear more than their share be paid by those who bear less. Data center development fits that mold precisely.

Compensation is where the compromise will be won or lost. Absolutists in opposition have little reason to trust promises of community benefits; they have watched too many corporate commitments evaporate once the ribbon is cut. That skepticism is not unreasonable. Today, compensation is a product of one-off negotiations between individual communities and sophisticated developers. As the Brookings researchers found, that arrangement routinely leaves communities with less than they bargained for, and some communities lack the expertise or leverage to bargain at all. The deeper problem is that no one in those negotiations has an incentive to price the true costs. Environmental and community harms are hard to quantify, slow to materialize, and easy to lowball when the parties doing the pricing prefer a smaller number.

A national framework can succeed where community-by-community dealmaking has faltered. Under this proposal, every approved project would trigger an independent cost assessment before ground is broken. Assessors, selected through a process insulated from industry influence and bound by methodologies published in advance for public comment, would price the full footprint of each facility: water draw, strain on local power infrastructure, effects on surrounding property values, and the gap between jobs promised and jobs delivered. That assessment would set a compensation baseline that developers must meet, converting community benefits from a matter of negotiating luck into a matter of right. 

The states, meanwhile, would retain a real choice in how much of a role they would like to play in this national endeavor. Each state would vote on how much of its allotment to accept. Those that host fewer projects would contribute a proportionally larger share to a national development fund flowing directly to host communities–no more free-riding on infrastructure built in someone else's backyard. Just as Hamilton warned against states whose unneighborly regulations impose costs on their fellow states, a state that opts out of hosting should not opt out of paying. Nevertheless, choice requires guardrails. No state could take on dramatically more than its share without a supermajority vote of its legislature; that floor exists precisely to protect states whose fiscal desperation might otherwise invite the very harms the absolutists fear. 

* * *

Governor Hochul rightfully observed that "data center development can be unpredictable." However, she failed to acknowledge that her executive order is only contributing to that unpredictability. A patchwork of state bans, moratoria, and one-off deals guarantees that data centers will continue to land wherever resistance is weakest rather than where the nation's interests are best served.

Reasonable minds will quarrel with the commission's design, the formula behind the regional caps, or the mechanics of independent cost assessment. Those quarrels are welcome. The proposal's chief virtue is not any single mechanism but its framing: it treats data center development as a national project that demands a nuanced approach, one that takes the concerns of host communities seriously, rewards the states that say yes, and asks something of the states that say no. Realists should feel free to improve upon it, discard it, or advance alternatives of their own. What they cannot do is cede the field.

The country has been here before. The Erie Canal was not the only conceivable route to the West, and DeWitt Clinton's critics derided the project as folly. What set Clinton apart was not the perfection of his plan but his refusal to accept that a generational infrastructure question could be answered with inaction. Today's absolutists and purists can keep shouting past one another while governors foreclose their states one executive order at a time. Or the realists can put a framework on the table and dare Congress to take the long view. The nation was better off the last time someone did.

Kevin Frazier leads the AI Innovation and Law Program at the University of Texas School of Law and is a Senior Fellow at the Abundance Institute.