Ask Runable forDesign-Driven General AI AgentTry Runable For Free
Runable
Back to Blog
Technology Policy40 min read

Why States Are Pausing Data Centers: The AI Infrastructure Crisis [2025]

A bipartisan wave of data center moratoriums is sweeping across America. Learn why states are halting AI infrastructure development and what it means for the...

data center moratoriumartificial intelligence infrastructureelectrical grid strainAI energy consumptiondata center expansion+10 more
Why States Are Pausing Data Centers: The AI Infrastructure Crisis [2025]
Listen to Article
0:00
0:00
0:00

Why States Are Pausing Data Centers: The AI Infrastructure Crisis [2025]

Something remarkable happened in American politics recently. Democrats and Republicans actually agreed on something. Not taxes, not healthcare, not immigration. They agreed that data centers—the physical backbone of artificial intelligence—are a problem that needs immediate federal and state intervention.

New York just joined at least five other states in proposing legislation to hit pause on data center development. But this isn't just a New York thing. It's happening in Georgia, Maryland, Oklahoma, Vermont, and Virginia. Some of these states are deep red. Others are deep blue. The split doesn't matter anymore. The concern is real, it's visceral, and it's spreading faster than anyone expected.

Here's what's happening: AI companies are ravenous for computing power. That power comes from data centers—massive facilities that consume electricity like small cities. And that electricity has to come from somewhere. As more data centers get built, electric grids strain, energy bills rise, water supplies get tapped, and communities push back.

This isn't a minor inconvenience. We're talking about gigawatts of electrical demand appearing on state grids almost overnight. One utility reported that 10 gigawatts of electric demand connected to data centers tripled in just one year. That's not a trend. That's a crisis pretending to be normal.

The backlash is becoming a defining policy issue of 2025. And it's forcing some uncomfortable questions: Can we actually sustain AI's growth? Should we? Who pays when energy costs explode? And what happens to communities that end up hosting the infrastructure nobody wants?

Let's dig into why this is happening, what it means, and where it's heading next.

TL; DR

  • The Wave: At least six states have introduced moratorium legislation in recent months, with bipartisan support across red and blue states
  • The Scale: Data center electrical demand tripled in one year in New York alone, with 10 gigawatts pending grid connection
  • The Cost: Communities face higher energy bills, water depletion, environmental degradation, and strained infrastructure
  • The Politics: Both Democrats and Republicans are backing restrictions, making data center policy one of the few truly bipartisan issues
  • The Future: Expect more state-level action, industry pushback, and a national conversation about whether current AI growth is sustainable

TL; DR - visual representation
TL; DR - visual representation

Exponential Growth in Data Center Electricity Demand
Exponential Growth in Data Center Electricity Demand

Electricity demand from data centers in New York has grown exponentially, tripling in just one year from 3.3 to 10 gigawatts, with a projected increase to 15 gigawatts. Estimated data reflects the rapid expansion driven by AI demands.

The Sudden Consensus: Why Data Centers Became Political

Political consensus in America is rare. You can count on one hand the issues where Vermont socialists and Florida Republicans find common ground. So when that happens, you know something significant is shifting.

The data center moratorium movement proves this point. In December 2024, Senator Bernie Sanders called for a national moratorium on data center permitting. Days later, Florida Governor Ron De Santis criticized data centers at a policy roundtable. These aren't politicians who typically agree on much of anything. But on data centers, they're singing from the same hymnal.

De Santis's comments were especially pointed. He said he doesn't think people want higher energy bills "just so some chatbot can corrupt some 13-year-old kid online." The line got applause. It resonated because it articulated something people felt but couldn't quite express: this AI boom is benefiting Silicon Valley while everyone else foots the bill.

What's driving this consensus? Three things converge perfectly to create political pressure.

First, the scale is impossible to ignore. Data centers used to be a niche infrastructure issue. Now they're everywhere. New York currently has more than 130 data centers. Several large-scale projects are under construction or proposed, including a 450-megawatt facility built on an old coal plant. The sheer number of these things is new. The speed at which they're multiplying is shocking.

Second, the costs are becoming visible. When utility companies report that electrical demand tripled in a year because of data center connections, that's not a footnote in a report. That's an existential question for grid stability. When communities see electric bills rising and water supplies being depleted to keep servers cool, they notice. They complain to elected representatives. Those representatives, whether Democrat or Republican, hear from their constituents.

Third, the benefit is unclear to most people. Tech companies get rich. Data center operators make money. AI companies train models faster. But what does the person in upstate New York get? Higher bills. Strained infrastructure. Noise, pollution, and environmental damage. The value proposition is one-sided.

QUICK TIP: Follow state legislative tracking sites like NCSL.org to see which data center bills are being proposed in your state. Many are moving quickly through committees.

Add all this together, and you get the perfect political storm. Governors and state legislators feel pressure from constituents. Industry opposition isn't strong enough locally to overcome that pressure. And the issue is simple enough to explain: Too many data centers, too fast, costing too much. Vote yes on the moratorium.

That's why New York State Senator Liz Krueger and Assemblymember Anna Kelles introduced their three-year moratorium bill with confidence. They knew they had momentum. They knew other states were moving. They knew voters were behind them.

The Sudden Consensus: Why Data Centers Became Political - contextual illustration
The Sudden Consensus: Why Data Centers Became Political - contextual illustration

Electricity Consumption of Data Centers vs Households
Electricity Consumption of Data Centers vs Households

A single large data center can consume as much electricity as 700,000 to 1 million households, highlighting the significant power demand of these facilities. Estimated data.

The Electricity Crisis Hidden in Plain Sight

Electricity consumption is the core issue driving moratorium legislation. It's not abstract or theoretical. It's measurable, growing, and threatening grid stability across multiple states.

Data centers are voracious consumers of electricity. A single large facility can draw as much power as a city of 100,000 people. And these facilities run 24/7, unlike most commercial buildings. They need consistent, reliable power—or everything inside shuts down, and your cloud storage disappears.

The growth rate is the shocking part. New York's example is instructive. Utilities reported 10 gigawatts of electrical demand pending connection to the grid, primarily from data centers. That's not a small increase. That's a massive, potentially destabilizing surge.

Here's the math: A typical American household uses about 10-15 kilowatts of power on average. Ten gigawatts equals the electrical demand of roughly 700,000 to 1 million households. All from data centers. All pending connection. All coming online in the next few years.

And this happened in just one year. The demand tripled. That means three years ago, it was 3.3 gigawatts. Before that, even less. The exponential growth is the real problem.

Why is demand growing so fast? Artificial intelligence. AI companies need enormous computing power to train and run large language models. Open AI, Google, Microsoft, Meta, and dozens of others are all competing to build the biggest, fastest AI systems. That competition drives data center construction.

Microsoft has been particularly aggressive. The company is building data centers specifically optimized for AI workloads. These facilities consume even more power than traditional data centers because AI training requires parallel processing across thousands of GPUs and specialized chips.

DID YOU KNOW: A single training run for a large language model can consume as much electricity as 100 American homes use in a year. As models get larger and more capable, that number keeps climbing.

The problem is that electrical grids weren't built for this kind of surge. Infrastructure planning typically works in 10-20 year cycles. Utility companies plan based on historical demand growth. They build generation capacity, transmission lines, and distribution infrastructure based on those projections.

But data centers don't follow historical patterns. They show up suddenly in a region. They demand connection immediately. They need power reliability measured in four nines (99.99% uptime) or better.

This creates a cascading problem. Utilities have to fast-track infrastructure upgrades they didn't plan for. That costs money. Someone has to pay for those upgrades. Guess who? Current ratepayers. So your electric bill goes up to fund infrastructure serving data center customers.

New York's Governor Kathy Hochul recognized this problem. In response, she launched an initiative requiring data centers to "pay their fair share" for infrastructure upgrades. But pay what? How much is fair? These questions are still being debated.

The electricity problem gets worse when you consider regional variation. Some states have more generating capacity than others. Some have diverse energy sources. Others depend on specific fuel types. A sudden influx of data center demand stresses whatever system exists.

Georgia is a case in point. The state is one of the national hubs for data center development. Georgia has the infrastructure, the real estate, and the utilities interested in hosting these facilities. But Georgia's electrical grid, like most grids, has limits. If too many data centers connect simultaneously, blackout risk increases.

Utilities can brown out, rolling blackouts can occur, and the entire regional economy suffers. That risk isn't theoretical. It's something utilities and state regulators genuinely worry about.

QUICK TIP: Check your state's utility commission website to see if data center interconnection requests are listed. Many states now track these publicly as demand grows.

This is why moratoriums appeal to legislators. A three-year pause sounds extreme, but it solves a real problem: it gives utilities and regulators time to plan. Instead of reactive infrastructure scrambling, they can proactively plan generation, upgrade transmission, and coordinate regional power flows.

The electricity crisis isn't theoretical. It's already here. Moratoriums are the legislative response to a physical reality that grids can't handle.

The Electricity Crisis Hidden in Plain Sight - contextual illustration
The Electricity Crisis Hidden in Plain Sight - contextual illustration

Water, Environment, and the Hidden Costs of Cool Servers

Electricity is the headline problem. But water is where the real environmental damage happens.

Data centers need cooling. Server hardware generates enormous heat. Push too much electricity through densely packed computer chips, and temperatures skyrocket. Servers shut down at 65-70 degrees Celsius. That's hot enough to damage circuits. So data centers need constant cooling.

There are two main cooling approaches: air cooling and water cooling. Air cooling uses large fans to push ambient air across hot hardware. That works okay in cool climates. It doesn't work in hot states like Georgia, Arizona, or Virginia during summers.

Water cooling is more efficient but more problematic. You pump cool water through pipes around the servers, the water absorbs heat, and then that hot water needs to cool down before it can be reused. This requires enormous quantities of water.

A large data center can consume hundreds of thousands of gallons of water daily. Some facilities use millions of gallons per day. This water comes from local sources: rivers, aquifers, municipal supplies. And once used for cooling, it's either evaporated (losing that water entirely) or returned to the source contaminated or at a different temperature.

This matters in water-stressed regions. The American Southwest has been in drought for decades. The Ogallala Aquifer, which supplies water to agriculture across the Great Plains, is being depleted. South Carolina, where water supplies are constrained, is seeing data center development pressure.

Local communities object because they see their water resources being consumed to keep servers cool for companies making billions of dollars. Why should a town's aquifer fund AI training? That's the fundamental question.

Environmental groups have made this argument forcefully. In December 2024, more than 200 national and local environmental organizations signed a letter asking Congress to pass a national data center moratorium. They called data center expansion "one of the biggest environmental and social threats of our generation."

That language is strong. It's not casual. These organizations have spent decades on climate change, pollution, and environmental justice. They're prioritizing data center expansion as a threat comparable to other major environmental issues.

Food and Water Watch, an environmental advocacy organization, specifically drafted the New York moratorium legislation. Their senior strategist, Eric Weltman, explained that while many well-intentioned bills exist to address data center impacts, they're "not adequate." A moratorium is necessary because incremental regulation isn't working.

Why? Because the pace of data center development outpaces regulatory response. By the time a state drafts rules, holds hearings, and implements regulations, data centers have already connected and are operating.

DID YOU KNOW: Some data centers use "water-cooled" technology that recirculates water in closed loops, reducing consumption by up to 90% compared to open-loop systems. However, these systems are more expensive and energy-intensive to operate, which creates a different trade-off.

The environmental impacts extend beyond water. Data centers generate noise, light pollution, and electromagnetic radiation. They fragment habitats and affect local ecosystems. They contribute to heat island effects in surrounding communities.

In densely populated states like New York and Virginia, environmental impacts hit harder because there are more people affected. A data center in rural upstate New York means a whole region dealing with environmental degradation.

The environmental argument for moratoriums is straightforward: pause development long enough to understand and mitigate impacts. Otherwise, by the time we realize the damage, it's too late.

Projected Electricity Consumption by Data Centers
Projected Electricity Consumption by Data Centers

Estimated data shows that if current growth rates continue, data centers could consume up to 1200 TWh by 2031, potentially accounting for 10-20% of the U.S.'s total electricity generation.

Who Pays? The Hidden Cost Shifting Problem

When energy costs rise, who bears the burden? That's the political question driving moratorium support.

Data centers consume electricity, but they don't always pay the full cost of that consumption. Here's why: utility pricing structures are complex. In most states, utilities charge different rates to different customers. Residential customers pay per kilowatt-hour. Large industrial customers negotiate bulk rates.

When a data center connects to the grid and requires infrastructure upgrades (new transmission lines, generator capacity, etc.), those infrastructure costs are spread across all ratepayers. That includes residential customers, small businesses, and other industrial users.

So a homeowner in upstate New York might see their electric bill increase because a data center operator negotiated a cheap rate that doesn't fully cover the infrastructure costs. The homeowner subsidizes the data center through higher rates.

Governor Hochul's "fair share" requirement was an attempt to fix this. Data centers should pay proportionally for the infrastructure they require. That sounds straightforward but is complicated in practice.

How do you calculate fair share? By peak demand? By annual usage? By the infrastructure upgraded specifically for the data center versus existing infrastructure upgraded anyway?

Regulators have to answer these questions. Most haven't yet. So data center operators enjoy rates that don't fully reflect infrastructure costs, and regular ratepayers subsidize the difference.

This cost-shifting explains political opposition. Working-class families struggling with inflation don't want their electricity bills to subsidize AI training. Tech executives make billions from AI products. Why should a utility worker's electricity bill go up to pay for infrastructure serving those companies?

QUICK TIP: If you're worried about how data center expansion affects your area, contact your state legislators and ask about rate recovery mechanisms. How are data centers charged for infrastructure upgrades? Who bears the costs?

Moratoriums appeal to voters because they send a message: we're not subsidizing this. You want to build data centers in our state? Fine. But you'll pay full freight. Maybe that means a moratorium until fair cost recovery is established.

Microsoft's recent initiative is worth noting here. The company announced commitments to be a "good neighbor" in communities hosting data centers. This includes community engagement, local hiring, and infrastructure support.

But corporate good-neighbor programs don't necessarily mean fair cost recovery. They mean Microsoft will do things that improve their public image while still enjoying favorable rates. Communities want structural fairness, not corporate charity.

The cost-shifting debate is the most politically volatile aspect of data center expansion. Because it affects ordinary people's bills. It's not abstract. It's concrete, measurable, and hits household budgets.

The Bipartisan Revolt: Why Red States and Blue States Agree

Political polarization is America's defining characteristic. But not on data centers. Here, strange alliances form.

Virginia is a case study. The state is one of the nation's top data center hubs. Northern Virginia especially—areas near Washington, D. C. have enormous data center infrastructure. Major operators have multiple facilities in the region.

Despite this pro-industry landscape, Virginia's legislature is moving toward restrictions. State Delegate Josh Thomas has led the charge. In his first legislative session in 2024, he found just three other politicians interested in data center reform. That number grew to eight in 2025. By the time 2025's session ramped up, it had grown to 12 or 13.

That's not a fluke. That's a trend. More legislators recognize that data center expansion is a problem requiring action.

Oklahoma and Maryland both have Republican-sponsored data center moratorium bills. These are conservative states with Republican leadership. They shouldn't care about environmental impacts or progressive concerns, right? Except their constituents care. Voters worry about power bills and environmental degradation regardless of party affiliation.

Georgia, another conservative state hosting massive data center development, is seeing political pushback. Residents don't want uncontrolled expansion. They want their electric grids to stay stable. They want water for agriculture and drinking, not just server cooling.

Democratic states like Vermont, Maryland, and New York are also moving. But they're not the only ones. The breadth of support is the remarkable part.

What explains this bipartisan consensus? Three factors:

First, data center impacts are local and visible. Everyone pays electricity bills. Everyone uses water. When data centers affect these basic utilities, it transcends ideology. A conservative farmer in Oklahoma and a liberal environmental activist in Vermont both want stable electric grids and clean water.

Second, tech industry funding is less influential at state legislative level than at federal level. Tech companies have enormous lobbying presence in Washington. At state level, their influence is weaker. Local concerns matter more.

Third, data center benefits accrue to out-of-state companies, while costs hit local communities. This creates a natural coalition against expansion. Residents of Virginia see the benefits going to Microsoft and Amazon while Virginians pay higher power bills. That breeds resentment regardless of political party.

DID YOU KNOW: Tech industry lobbying spending exceeded $148 million in 2023, but most went toward federal policy. State legislative lobbying is significantly cheaper and receives less industry attention, giving residents more influence.

The bipartisan nature of data center opposition is politically significant. It suggests this isn't partisan politics. It's genuine concern about uncontrolled infrastructure development.

Sanders's call for a national moratorium and De Santis's criticism both resonate because they articulate real problems. Sanders frames it as wealth inequality—AI benefits the 1 percent while everyone else subsidizes infrastructure. De Santis frames it as corporate excess harming society. Different rhetoric, same underlying message.

Both appeal because both acknowledge a legitimate problem: current data center development isn't sustainable or equitable.

The Bipartisan Revolt: Why Red States and Blue States Agree - visual representation
The Bipartisan Revolt: Why Red States and Blue States Agree - visual representation

Water Usage in Data Centers
Water Usage in Data Centers

Water cooling in data centers can use up to 10 times more water than air cooling, highlighting significant environmental concerns in water-stressed regions. Estimated data.

The Scale of the Problem: By the Numbers

When politicians talk about pauses and moratoriums, they're responding to numbers that are genuinely alarming.

Let's look at what we know: New York has over 130 data centers currently operational. The state has 10 gigawatts of electrical demand from data centers pending grid connection. That's roughly equivalent to 700,000 to 1 million households' total electricity consumption.

Virginia has proposed over 60 data center-related bills in its legislature in 2025 alone. Sixty bills. That suggests legislators think the problem requires comprehensive legislative response, not just tweaks.

Georgia is one of the top three data center states nationally. The state hosts major facilities for Amazon, Google, and others. New facility proposals keep arriving.

Montana saw communities imposing local moratoriums on data center development. Oregon is seeing similar local action. Communities are acting because state legislatures are slow.

All told, at least 14 states have towns or counties that have imposed data center moratoriums at the local level. Fourteen states! And that's just local action. At the state legislative level, at least six states have introduced formal moratorium bills.

The growth rate is what matters. Data center electricity demand has grown exponentially over the past three years. Five years ago, this wasn't a major issue. Ten years ago, data centers were just beginning the infrastructure race for AI. Now, it's one of the fastest-growing sources of electricity demand.

Consider the timing: Bernie Sanders called for a national moratorium in December 2024. Within days, Ron De Santis criticized data centers. Within weeks, multiple state legislatures introduced bills. Within months, major environmental organizations with deep expertise in energy policy had made data center expansion a top priority.

That kind of policy mobilization happens when an issue reaches critical mass. Nobody mobilizes unless they think something is broken and needs fixing.

QUICK TIP: Track data center development in your state by checking your Public Service Commission or Public Utilities Commission website. Most states now list pending data center interconnection requests publicly.

The numbers show that data center expansion is accelerating. Moratoriums are the regulatory response. Whether they'll actually slow development or just shift it elsewhere is still unclear.

The Scale of the Problem: By the Numbers - visual representation
The Scale of the Problem: By the Numbers - visual representation

How Long Is a Moratorium? Three Years and What Happens Next

The New York bill proposes a three-year moratorium on data center permitting. Three years is a long time to pause development, but it's not forever. What happens during those three years? And what happens after?

During a moratorium, state regulators would study impacts. New York's bill requires the Department of Environmental Conservation and Public Service Commission to issue reports on data center impacts and suggest new regulations or orders to minimize those impacts.

That's the theory. In practice, three years is enough time to examine environmental impacts, understand grid stress, and develop regulatory frameworks. It's not enough time to solve underlying problems. Three years is a pause button, not a solution.

Which raises the question: what comes after the pause? The bills don't say. Do restrictions sunset? Do they get renewed? Do new regulations kick in?

Most likely scenario: After the moratorium, states will implement new regulations rather than lift restrictions entirely. Rules about environmental review, infrastructure cost recovery, and grid impact assessment will probably become standard.

Data center operators know this. They're already preparing. Microsoft's "good neighbor" initiative is essentially pre-emptive compliance. If Microsoft gets ahead of regulations and demonstrates responsible practices, state regulators might impose lighter restrictions.

The industry group Data Center Coalition's response confirms this. A spokesperson said the industry "recognizes the importance of continued efforts to better educate and inform the public...through factual information about the industry's responsible usage of water and our commitment to paying for the energy we use."

Translation: We'll fight the moratorium, but we're also preparing for stricter regulations. We'll try to look responsible and hope for lighter-touch oversight.

DID YOU KNOW: Some countries are already implementing data center regulations. Ireland limited data center development to reduce strain on Dublin's power supply. France proposed regulations on AI infrastructure power consumption.

The three-year timeline is also significant politically. It's long enough that current legislators won't be blamed if growth continues. It's short enough that voters see it as action, not endless study. It's the Goldilocks solution: not too much, not too little.

But the deeper question is whether moratoriums address the core problem. They slow things down, but do they prevent the underlying issue? If AI companies can build data centers in states without moratoriums, moratoriums just shift development geography.

Some states might become data center hubs. Others might impose restrictions. Operators will chase the path of least resistance, locating in states with lighter regulations. That could create regulatory arbitrage, where the most environmentally and energy-conscious states drive data center development to less regulated states.

This is why some activists push for federal action. A national moratorium can't be circumvented. It has to be addressed at the national policy level. But federal legislation moves slower than state legislation. Moratoriums might already be in place before Congress acts.

How Long Is a Moratorium? Three Years and What Happens Next - visual representation
How Long Is a Moratorium? Three Years and What Happens Next - visual representation

Potential Consequences of Unrestricted Data Center Expansion
Potential Consequences of Unrestricted Data Center Expansion

Estimated data suggests severe consequences if data center expansion continues without restrictions, with grid failures and water depletion being most critical.

The Technology Company Response: Damage Control and Preparation

Microsoft, Google, Amazon, and other tech giants didn't anticipate this backlash. They thought data center expansion would be treated like normal industrial development. Permission granted, facility built, everyone moves on.

Instead, they hit a wall of political opposition. Communities don't want data centers. Environmentalists oppose expansion. Legislators listen to constituents and introduce restrictions. The tech industry's infrastructure strategy collided with reality.

Microsoft's response is instructive. The company rolled out a public commitment to be a "good neighbor" in communities where it operates. This includes community engagement, local hiring, infrastructure investment, and environmental responsibility.

It's a good public relations move. It shows Microsoft listening to concerns and committing to responsible practices. It also signals to regulators that Microsoft will comply with future restrictions and work with local communities.

But it's also a form of damage control. Microsoft's infrastructure plans require building many data centers. If too many states impose moratoriums, those plans get derailed. So the company is trying to build goodwill proactively.

Other companies are probably developing similar strategies. Open AI, Anthropic, and other AI labs will likely face similar pressure. They'll need to make public commitments to responsible data center development.

Industry groups are also responding. The Data Center Coalition, representing facility operators, is emphasizing "responsible usage of water" and the industry's commitment to "paying for the energy we use." Again, it's an attempt to get ahead of the narrative. We're already doing the right things, they're saying. Moratoriums aren't necessary.

But the industry's message isn't resonating. Legislators are moving ahead with moratorium bills anyway. That suggests industry arguments aren't persuasive enough to overcome constituent pressure for restrictions.

QUICK TIP: Pay attention to industry statements and public commitments from tech companies operating in your state. These often reveal what regulations they're expecting and trying to shape through public relations.

Long-term, the tech industry will likely adapt. They'll build more efficient data centers. They'll invest in renewable energy to power facilities. They'll locate new data centers in countries or regions with fewer restrictions. They'll optimize AI models to require less computing power.

These adaptations will take time. But they're coming because the regulatory environment is changing. Data center expansion as a business model is becoming politically untenable in many states.

The Technology Company Response: Damage Control and Preparation - visual representation
The Technology Company Response: Damage Control and Preparation - visual representation

Environmental Groups and the New Anti-Data Center Coalition

Two hundred organizations signing onto a letter against data center expansion isn't random. It reflects genuine concern from groups with deep expertise in environmental issues.

Food and Water Watch didn't pull the New York moratorium proposal out of thin air. The organization spent months analyzing data center impacts, consulting with other environmental groups, and designing legislation that would actually address problems rather than paper over them.

Eric Weltman's comment that existing bills are "not adequate" is important. Environmental groups aren't anti-technology. They're sophisticated policy analysts. They understand that incremental regulations often fail to achieve goals. They recognize regulatory capture, where regulated industries gradually reshape regulations in their favor.

So a moratorium is a sledgehammer approach. It's blunt. It's not precise. But moratoriums work for buying time to develop better regulations that can't be easily circumvented.

The coalition of environmental groups includes organizations focused on climate, water, environmental justice, and land use. The breadth of support is significant. It suggests multiple environmental concerns converge on data center expansion.

Climate groups care about electricity consumption and carbon emissions. Water groups care about depletion and pollution. Environmental justice groups care about siting facilities in or near low-income and communities of color. Land use groups care about habitat fragmentation and development pressure.

All these concerns exist simultaneously. Data center expansion creates real harms across multiple dimensions. Environmental organizations are responding by building coalitions and pushing for comprehensive restrictions.

Federal environmental organizations are also getting involved. That suggests data center expansion is becoming a major environmental issue at the national level. When national groups make something a priority, it usually reflects judgment that the issue is significant and requires urgent response.

DID YOU KNOW: The AI field's electricity consumption is expected to exceed 10% of global electricity generation by 2030 if current trends continue. That would make AI the largest electricity consumer in the world, surpassing transportation and heating.

The environmental coalition will be important in shaping future data center regulation. These groups have influence with state regulators. They provide technical expertise. They mobilize voters. When environmental groups push for restrictions, legislators listen.

Expect environmental organizations to continue driving the anti-data center agenda. They'll push for stronger moratoriums, tighter regulations, and better environmental review processes. They'll provide technical expertise to support restrictions. And they'll mobilize public opinion against expansion.

Environmental Groups and the New Anti-Data Center Coalition - visual representation
Environmental Groups and the New Anti-Data Center Coalition - visual representation

Potential Federal Policies for Data Centers
Potential Federal Policies for Data Centers

Energy efficiency standards and environmental review requirements are estimated to have the highest impact on regulating data centers at a federal level. Estimated data.

States as Laboratories: Virginia, Georgia, and the Data Center Wars

Virginia and Georgia are where the data center story gets real. Both states are major hubs for data center development. Both have significant political pressure to restrict expansion. The outcomes in these states will shape national policy.

Virginia is particularly interesting. The state hosts enormous data center infrastructure in Northern Virginia, near Washington, D. C. Companies like Amazon Web Services, Google, Microsoft, and others have major presence. The region generates economic activity and tax revenue.

But infrastructure strain is real. The electrical grid in Northern Virginia is under stress. Water supplies are constrained. Traffic and congestion are getting worse. Residents are frustrated.

State Delegate Josh Thomas has become the leader of data center reform in Virginia. Starting with just three other interested legislators, he's built a coalition that might reach 12 or 13. That's significant momentum.

Thomas is pushing multiple approaches. Some bills might impose moratoriums. Others might focus on environmental review, cost recovery, or community benefits agreements. The goal is comprehensive legislative response rather than single bills.

Georgia faces similar dynamics. The state is a top-three data center hub nationally. Facilities are operating and proposed throughout the state. But political pressure for restrictions is growing.

Georgia is interesting because it's Republican-led. The state isn't known for environmental restrictions or progressive regulation. Yet data center opposition is finding traction. That's partly because rural communities care about electricity rates and water supplies, regardless of partisan affiliation.

Both states will become laboratories for how legislatures approach data center regulation. The bills Virginia and Georgia pass will likely be models for other states. Successful approaches will be copied. Failed approaches will be avoided.

QUICK TIP: Watch Virginia and Georgia legislative tracking sites closely. Bills passed in these states often become models for other states' data center policy.

The outcomes in Virginia and Georgia will also signal to the tech industry whether regulatory restrictions are durable or easily circumvented. If Virginia and Georgia impose strict regulations that actually reduce data center expansion, other states will likely follow. If regulations are weak or easily avoided, the moratorium model might fail nationally.

The stakes in these two states are high. The political outcome will determine not just regional data center policy but national trends.

States as Laboratories: Virginia, Georgia, and the Data Center Wars - visual representation
States as Laboratories: Virginia, Georgia, and the Data Center Wars - visual representation

The Cost of Inaction: What Happens Without Restrictions

Let's consider the alternative. What happens if moratoriums don't pass? What happens if states don't restrict data center development?

Electrical grids face cascading failures. That's the literal answer. Overload a grid with demand it wasn't designed to handle, and blackouts result. The 2003 Northeast Blackout demonstrated this. A single failure cascaded across multiple states, affecting 55 million people and costing $7 billion.

Data center demand could trigger similar events. Multiple facilities connecting suddenly, peak demand exceeding grid capacity, and the entire regional grid collapses. That's not theoretical risk. That's something utilities actually worry about.

Water supplies become depleted. In water-stressed regions, agricultural activity declines. Communities face water rationing. Ecosystems collapse due to lack of water. All to cool servers processing AI requests.

Electricity rates rise substantially. Ratepayers subsidize infrastructure upgrades needed to serve data centers. Lower-income households, who spend higher percentages of income on utilities, get hit hardest. This is an equity issue. Wealthy tech executives benefit from cheap computing. Working-class people pay higher electricity bills.

Environmental degradation accelerates. Habitats get fragmented. Local ecosystems get disrupted. Noise and light pollution worsen. Communities face cumulative environmental stress.

Business uncertainty increases. If data center expansion triggers grid failures or environmental crises, it becomes a systemic risk. Investors worry. Companies operating in affected regions face uncertainty. Economic stability declines.

When you sum these consequences, inaction becomes obviously untenable. States can't afford to let data center expansion proceed without restrictions. The costs are too high, the risks too severe.

So moratoriums become rational policy response to genuine crisis. They're not anti-technology. They're pro-grid-stability, pro-environment, pro-equity.

The Cost of Inaction: What Happens Without Restrictions - visual representation
The Cost of Inaction: What Happens Without Restrictions - visual representation

What Might Come After Moratoriums: Regulatory Frameworks

If moratoriums pass and become the norm, what replaces them? States will develop data center regulations. What might these look like?

Environmental review requirements will likely become standard. Data centers will require environmental impact assessments examining electricity impacts, water consumption, noise, habitat effects, and cumulative regional impacts.

Cost recovery mechanisms will probably require data center operators to pay full cost of infrastructure upgrades. No more subsidies from residential ratepayers. Operators pay for the transmission lines, generation capacity, and grid upgrades their facilities require.

Community benefit agreements might become required. Data centers would commit to local hiring, workforce development, property tax contributions, and community investment. This ensures local communities benefit, not just tech companies.

Energy source requirements might mandate that data centers use renewable energy or purchase renewable energy credits. This addresses climate concerns by ensuring facilities don't increase carbon emissions.

Water restrictions might limit data center water consumption or require alternative cooling technologies. Some states might ban water-cooled facilities entirely, requiring air cooling or more innovative approaches.

Capacity limits might impose maximum data center electricity consumption or maximum facilities per region. This prevents any single state from being overwhelmed by data center demand.

These regulations wouldn't ban data centers. They'd make data center development more expensive, more transparent, and more equitable. Operators could still build but would have to bear full costs and consider local impacts.

DID YOU KNOW: Some data centers are experimenting with alternative cooling technologies, including AI-optimized liquid cooling and machine learning systems that predict cooling needs. These could reduce water consumption by 99% compared to traditional cooling.

The regulatory framework that emerges from moratoriums will likely become national model. If California, New York, and Virginia develop effective regulations, other states will copy them. Federal regulations might follow, establishing baseline requirements for all states.

Industry will adapt. Companies will build more efficient facilities, invest in renewable energy, and locate where regulations are feasible. The unbridled growth period is probably ending. The regulation period is beginning.

What Might Come After Moratoriums: Regulatory Frameworks - visual representation
What Might Come After Moratoriums: Regulatory Frameworks - visual representation

The National Conversation: Federal Action and Long-Term Policy

State moratoriums are important, but they're ultimately temporary. Long-term, federal policy will determine how data center expansion is managed nationally.

Bernie Sanders's call for a national moratorium was significant partly because it elevated the issue to federal level. Federal action would set baseline requirements all states must follow. It would prevent regulatory arbitrage where companies locate in the least-regulated states.

What might federal policy look like? Several approaches are possible.

A temporary federal moratorium, like some states are proposing, would pause development while federal regulators study impacts and develop long-term policy. This is the simplest approach and likely what Sanders intended.

Energy efficiency standards might require data centers to meet minimum efficiency requirements. This is already done for other industrial facilities. Similar standards for data centers could reduce electricity consumption and water usage.

Carbon pricing or tax incentives might make renewable-powered data centers cheaper than coal-powered facilities. Federal policy already uses this approach for solar and wind. Similar mechanisms could favor efficient data centers.

Water usage standards might limit data center water consumption or require conservation technologies. Federal environmental law already regulates water usage by industries. Data centers could be included.

Environmental review requirements might become federal mandate. All major data center projects could require federal environmental review, not just state review. This would catch regional impacts that states might miss.

QUICK TIP: Follow congressional committees dealing with technology, energy, and environment. Federal data center legislation is likely to emerge within the next 2-3 years as state moratoriums demonstrate the issue's importance.

The challenge with federal action is political. Tech companies have enormous lobbying presence. They'll resist federal restrictions that increase costs or slow development. Congress moves slowly. By the time federal legislation passes, states may have already established divergent approaches.

So federal action likely follows state action rather than leading it. States will experiment with different approaches. Successful models will emerge. Then federal legislation will establish baseline requirements based on what worked.

That process could take 5-10 years. In the meantime, state moratoriums and regulations will shape data center development. The states that move first—New York, Virginia, Georgia, others—will set precedent.

The National Conversation: Federal Action and Long-Term Policy - visual representation
The National Conversation: Federal Action and Long-Term Policy - visual representation

The Sustainability Question: Can AI Infrastructure Scale Sustainably

Here's the fundamental question underlying all this: Is current AI growth sustainable?

Data centers consume enormous electricity. AI training is particularly power-intensive. As AI models get larger and more capable, electricity consumption grows exponentially. This growth trajectory is obviously unsustainable. At some point, you run out of electricity and water.

So either AI growth slows, or infrastructure innovation dramatically increases efficiency, or we find new approaches entirely.

Infrastructure innovation is happening. Data centers are getting more efficient. Chip designers are creating more power-efficient processors. Cooling systems are improving. AI models are getting better at producing results with less computation.

But efficiency improvements typically follow an S-curve. Big initial improvements taper off. You can't get better and better indefinitely. Eventually, you hit physical limits.

Consider: electricity is ultimately limited by generation capacity. The U. S. generates what, roughly 4,000 terawatt-hours annually? If data center demand grows at current rates, it could consume 10-20% of that within a decade. Add transportation, heating, manufacturing, and you exceed generation capacity. You can't solve that with efficiency alone.

Water is similarly limited. Water-stressed regions can't increase supply indefinitely. Water-cooled data centers hit physical limits quickly in places like California, Arizona, or the South.

So the uncomfortable conclusion: current AI growth rates are unsustainable. Something has to change. Either AI growth slows, or we radically transform energy and water infrastructure, or we move data centers to the few regions with abundant water and electricity.

Moratoriums acknowledge this reality. They're saying: we can't accommodate this growth at current rates. We need to slow down, figure out sustainable approaches, and then proceed more deliberately.

That's the underlying political argument. Not "AI bad." But "uncontrolled growth unsustainable."

DID YOU KNOW: If AI electricity consumption continues growing at current rates, it would exceed all of Denmark's electricity generation by 2030. Denmark would need to double its generation capacity just to power new data centers added in a year.

The sustainability question also affects the tech industry's long-term strategy. If growth becomes constrained by infrastructure limits, that changes incentives. Companies will invest more in efficiency. They'll shift toward specialized chips that reduce power consumption. They'll develop smaller, more targeted models rather than massive general-purpose models.

These changes would be positive. Forced efficiency drives innovation. Constrained growth forces better decisions about which AI applications actually matter versus which are frivolous.

Moreover, sustainable infrastructure is good business. Facilities that use less electricity cost less to operate. Investors increasingly care about sustainability. Companies operating sustainably have better public image. So forcing efficiency serves long-term business interests too.

The moral of the story: moratoriums aren't anti-technology or anti-innovation. They're forcing the tech industry to build sustainably instead of recklessly. That's healthy correction.

The Sustainability Question: Can AI Infrastructure Scale Sustainably - visual representation
The Sustainability Question: Can AI Infrastructure Scale Sustainably - visual representation

International Parallels: How Other Countries Are Responding

America isn't alone in facing data center expansion pressure. Countries around the world are grappling with the same issues.

Ireland provides an instructive example. Dublin is a major data center hub for European operations. Google, Amazon, Meta, and others have facilities throughout the country. The Irish government got concerned about electricity strain and policy pressure to restrict expansion.

The government imposed restrictions on new data center development, particularly in Dublin. Companies could still build but faced more hoops to jump through. The goal was to slow development until infrastructure could catch up.

France has taken different approach, focusing on AI infrastructure power consumption. The country's energy regulator has proposed restrictions on power-intensive AI applications. The goal is to ensure AI development remains compatible with France's carbon reduction targets.

Germany, similarly focused on energy transition, is pushing data centers toward renewable energy. Facilities are required to use renewable sources or invest in renewable development. This aligns data center growth with climate goals.

Denmark and other Nordic countries, with abundant renewable electricity and cooler climates, are attracting data centers. Companies locate there to take advantage of cheap renewable electricity and natural cooling. This is regulatory arbitrage at the international level.

These international precedents matter for American policy. They show that countries can restrict data center development without collapsing their tech sectors. Companies adapt. They build where restrictions are feasible. They invest in efficiency.

QUICK TIP: Watch international data center policy, especially in Europe and Asia. Regulations that work internationally often become models for American policy a few years later.

The international experience suggests American moratoriums are probably feasible. They'll slow development but not stop it. Companies will adapt. Innovation will continue. But it'll happen more sustainably.

International Parallels: How Other Countries Are Responding - visual representation
International Parallels: How Other Countries Are Responding - visual representation

Timeline and What Comes Next: 2025 and Beyond

We're in early 2025. Data center moratorium bills are moving through state legislatures. The momentum is clear. More states will likely introduce similar legislation. Several moratorium bills will probably pass.

By end of 2025, I expect at least three states will have passed moratorium legislation. New York is likely. Virginia is possible. Georgia might. Others could follow.

Throughout 2025, we'll also see regulatory action. States without moratorium legislation might implement restrictions anyway through regulatory channels. Public utilities commissions might establish new interconnection requirements. Environmental regulators might impose stricter review standards.

Industry will respond with legal challenges, lobbying, and public relations campaigns. Tech companies will highlight job creation and economic benefits. They'll emphasize efficiency improvements and renewable energy commitments. This political battle will intensify through 2025.

Federal attention will grow. Congress will hold hearings. Federal agencies will convene task forces. By late 2025, there might be federal legislation introduced addressing data center policy. Passage is unlikely in 2025, but the stage will be set.

International policy will continue diverging. Some countries will welcome data centers (maybe offering incentives). Others will restrict. This creates opportunities for companies to play jurisdictions against each other.

In 2026, federal legislation might pass if political momentum builds. Federal policy would establish baseline standards all states must follow. This would prevent regulatory arbitrage and establish coherent national policy.

By 2027-2028, we'd probably see new data center regulations become standard in most states. Moratoriums might expire, getting replaced by permanent regulatory regimes. The wild west of data center development would end. The regulated era would begin.

This timeline assumes continued state-level momentum and eventual federal action. If that happens, data center expansion gets managed. If not, states proceed with restrictions anyway, creating a patchwork of different state approaches.

Either way, the era of uncontrolled data center growth is ending. That's the important takeaway. The specific regulatory approach isn't yet determined, but restrictions are coming.

Timeline and What Comes Next: 2025 and Beyond - visual representation
Timeline and What Comes Next: 2025 and Beyond - visual representation

FAQ

What is a data center moratorium and why are states imposing them?

A data center moratorium is a temporary pause on permitting new data center development, typically lasting 3-5 years. States are imposing them because data center expansion is straining electrical grids, depleting water supplies, and raising electricity costs for residents. Moratoriums give regulators time to study impacts and develop sustainable long-term policies before more facilities are built.

How much electricity do data centers actually consume?

Data centers consume enormous amounts of electricity. A single large facility can draw as much power as 700,000-1 million households. In New York alone, data centers have 10 gigawatts of electrical demand pending grid connection—representing a tripling of demand in just one year. This explosive growth is driving grid instability concerns and power supply challenges.

Why are both Democrats and Republicans supporting data center restrictions?

Data center impacts are non-partisan. Everyone pays electricity bills and relies on water supplies. When data center expansion raises power costs and strains infrastructure, voters from both parties object. Working-class families struggle with inflation and higher utility bills. Farmers need water for crops, not server cooling. Environmental concerns transcend political boundaries. This creates natural bipartisan opposition to uncontrolled expansion.

What happens to communities if data center development continues without restrictions?

Communities face multiple consequences: electrical grids could fail during peak demand (causing cascading blackouts), water supplies get depleted affecting agriculture and drinking water, electricity rates rise for all customers, environmental degradation accelerates, and tax benefits often don't match infrastructure costs residents bear. Lower-income households suffer most since electricity represents larger budget percentages for them.

Are data centers moving to states without moratoriums?

Yes, regulatory arbitrage is happening. Companies are shifting development toward states with lighter restrictions. This is why a national approach might be necessary—individual state moratoriums can't prevent companies from moving. However, even states attracting data centers are likely to face similar pressure and restrictions eventually as impacts become visible.

What alternatives exist to water-cooled data centers that consume so much water?

Air cooling is used in cooler climates but is less efficient and expensive. Liquid cooling with closed-loop systems can reduce water consumption by 90% but costs more to operate. Immersion cooling—submerging hardware in coolant—is emerging as an alternative. AI-optimized cooling that uses machine learning to predict needs is also developing. These alternatives exist but cost more, which is why operators prefer water cooling if it's available.

Will federal legislation replace state moratoriums?

Likely yes, eventually. Federal legislation would establish baseline requirements all states must follow, preventing regulatory arbitrage. However, federal action typically follows successful state experimentation. States will probably implement moratoriums and develop regulatory frameworks first. Federal legislation will follow 2-5 years later, establishing national standards based on what worked at state level.

How do data center costs impact electricity rates for regular consumers?

When data centers require infrastructure upgrades (transmission lines, generation capacity, grid modifications), those costs are often spread across all ratepayers through rate increases. This means households subsidize data center operations even though they don't benefit directly. Governor Hochul's "fair share" requirement attempted to make data centers pay directly for infrastructure they require, though implementation details are still being worked out.

What role do environmental groups play in the moratorium movement?

Environmental organizations are major drivers. Over 200 national and local groups signed a letter calling for a national data center moratorium. They provide technical expertise on environmental impacts, draft legislation, and mobilize public opinion. Environmental groups have been effective at framing data center expansion as an environmental justice issue affecting water supplies, electricity, and air quality in affected communities.

What's Microsoft's "good neighbor" commitment and does it address core problems?

Microsoft announced initiatives including community engagement, local hiring, and infrastructure support in data center communities. This is partly genuine good faith and partly damage control—Microsoft's expansion plans require many new facilities, and moratoriums threaten those plans. However, corporate good-neighbor programs don't necessarily address structural problems like cost recovery and fair electricity pricing. Communities want regulatory fairness, not corporate charity.


The data center moratorium movement represents a fundamental shift in how America approaches infrastructure for artificial intelligence. For years, data center expansion happened with minimal oversight. Now, it's becoming a major political issue with genuine bipartisan support for restrictions.

The reasons are straightforward. Data centers consume extraordinary amounts of electricity and water. They strain grids. They raise costs for everyone else. And the benefits accrue primarily to wealthy tech companies and executives, while costs hit working people through higher utility bills.

Moratoriums aren't perfect solutions. They're blunt instruments. But they're proportionate response to genuine crisis. States can't afford uncontrolled data center expansion. The infrastructure can't handle it. The environment can't sustain it. Communities can't afford it.

This will reshape AI infrastructure development. Companies will build more efficiently. They'll invest in renewable energy. They'll develop technologies that require less computing power. Some will locate data centers in countries with more favorable regulatory environments. All of this is coming.

The era of reckless data center expansion is ending. The era of sustainable, regulated infrastructure is beginning. That's the story here. Not anti-technology sentiment, but pro-sustainability correction. When a system is growing unsustainably, correction is necessary. That correction is happening now.

FAQ - visual representation
FAQ - visual representation


Key Takeaways

  • At least six states have introduced data center moratorium legislation in early 2025, with truly bipartisan support from both Democratic and Republican lawmakers
  • Data center electrical demand tripled in New York in a single year, with 10 gigawatts pending grid connection—representing a scale crisis for regional power infrastructure
  • Environmental groups have mobilized comprehensively, with over 200 national and local organizations calling for a national moratorium on data center development
  • Moratoriums address three interconnected problems: electrical grid stability, freshwater depletion for cooling systems, and cost-shifting where residents subsidize infrastructure serving profit-generating data centers
  • Tech companies are responding with public commitments to responsible development, but underlying regulatory frameworks determining fair cost recovery are still being developed across states

Related Articles

Cut Costs with Runable

Cost savings are based on average monthly price per user for each app.

Which apps do you use?

Apps to replace

ChatGPTChatGPT
$20 / month
LovableLovable
$25 / month
Gamma AIGamma AI
$25 / month
HiggsFieldHiggsField
$49 / month
Leonardo AILeonardo AI
$12 / month
TOTAL$131 / month

Runable price = $9 / month

Saves $122 / month

Runable can save upto $1464 per year compared to the non-enterprise price of your apps.