No Roadblocks, No Special Favors: A State Framework for Data-Center Growth

States can welcome AI infrastructure while protecting ratepayers, respecting local property rights, and avoiding industry-specific subsidies.

What to Know

• U.S. data centers used about 4.4% of the country’s electricity in 2023; their share could reach 6.7%–12% by 2028.
• Major electricity users can protect households through long-term contracts, minimum bills, and financial commitments that keep project costs with the companies creating them.
• Georgia’s House passed a data-center ratepayer-protection bill on February 17, 2026, but it did not receive final Senate approval.
• A Tennessee developer says it received only 10 minutes’ notice before Hawkins County considered a blanket ban on data centers and cryptocurrency mining.
• Montana enacted its Right to Compute Act in 2025, protecting the ability to own and use computational resources.

Artificial intelligence depends on physical infrastructure: servers, cooling equipment, power plants, transmission lines, and large industrial buildings. The U.S. Department of Energy reports that data centers used about 176 terawatt-hours of electricity in 2023, or 4.4% of total U.S. electricity consumption. DOE projects they could use 325–580 terawatt-hours in 2028.

Those figures explain why data-center policy now affects more than technology companies. States must decide who pays for new power infrastructure, how communities regulate local development, and whether public policy should favor, restrict, or treat data centers like other large private investments. The State Policy Network’s August 2026 review describes one emerging approach: no roadblocks, no special favors.

Welcome Investment Without Picking Winners

The “no roadblocks, no special favors” framework rejects two opposite approaches. One is a blanket ban or indefinite moratorium that blocks data-center development regardless of a project’s location, infrastructure plan, or local effects. The other is government selecting preferred companies through special tax breaks, public guarantees, or company-specific deals.

The Commonwealth Foundation’s position is that Pennsylvania should streamline permitting and expand energy supply while avoiding industry-specific tax exemptions and corporate welfare. That does not mean data centers should operate without rules. They still need to meet generally applicable standards for zoning, safety, water, environmental compliance, and utility service.

Data-center types and scale. U.S. Department of Energy.

This distinction matters for households and taxpayers. A subsidy may appear to attract investment at no immediate cost, but its cost can surface later through forgone revenue, public infrastructure spending, or higher utility rates. At the same time, unclear permitting rules can delay infrastructure that businesses and consumers already use through cloud services, banking systems, logistics tools, and AI applications.

A neutral policy applies clear rules to every large project. It welcomes private investment when companies meet those rules, while requiring the companies to accept the financial risks of their own decisions.

The largest practical test of neutral treatment is electricity: who pays when a project requires major new power infrastructure?

Build the Power, but Keep the Cost With the User

Electricity is the central test of this approach. Data centers can require new generation, substations, and transmission. When utilities build those assets for a proposed facility that later shrinks, delays, or leaves, existing customers can be left paying for infrastructure they did not request.

The John Locke Foundation has proposed Consumer-Regulated Electricity legislation in North Carolina. Its “build, bring, or buy” model would let large users build their own generation, bring a dedicated source of power, or voluntarily buy electricity outside the traditional utility rate-base model.

This is a proposed policy model, not current North Carolina law. Its purpose is to give large users alternatives to asking a utility to build new infrastructure and recover the cost from all customers. If a data center needs unusual amounts of electricity, the company could arrange and finance a dedicated supply of power.

FERC’s four pillars for large-load grid reform. Federal Energy Regulatory Commission.

That model would still require reliability, safety, permitting, and interconnection standards. On-site or dedicated generation does not automatically resolve questions about backup power, emissions, transmission access, or emergency operations. But it changes the starting point: the company requesting the capacity should face transparent costs before households do.

Once states identify who should pay, they must decide which pricing and contract tools will make that responsibility enforceable.

Use Prices and Contracts to Protect Ratepayers

Building additional power will not be enough on its own. States also need rules that determine who bears the financial risk when a large user changes plans.

The Goldwater Institute’s May 2026 report recommends real-time pricing, peak-based charges, and interruptible-service agreements. These tools would give large electricity users stronger incentives to reduce demand when the grid faces the greatest stress. For example, a facility that can shift some computing work or use on-site resources during a peak period could avoid the highest prices.

Real-time pricing addresses short-term grid stress. Long-term contracts address a different risk: a company leaving before it has paid for the generation and infrastructure built to serve it.

FERC options for connecting large electricity users. Federal Energy Regulatory Commission.

According to the Georgia Public Policy Foundation, the Georgia Public Service Commission adopted minimum billing and longer contract terms for large-load customers in late 2025. These provisions aim to require major customers to keep paying for infrastructure even if they reduce operations or leave the state early.

Georgia House Bill 1063 passed the House on February 17, 2026, but it did not receive final Senate approval before lawmakers adjourned. The bill would have required new large-load contracts to include terms that prevent other customers from absorbing the incremental cost of serving a data center.

The Georgia debate shows a real policy trade-off. A statute can create a durable, enforceable protection against cost shifting. Regulators, however, may want flexibility to adjust contracts as forecasts, technology, and local grid conditions change. The South Carolina Policy Council has proposed another version of the same principle: long-term power-purchase agreements requiring data centers to finance the generation built for their facilities.

Electricity costs are only one part of the public debate; communities also need clear ways to address the local effects of individual projects.

Protect Communities With Clear Rules, Not Blanket Bans

Ratepayer protections do not answer every local concern. Residents may reasonably ask about noise, water use, traffic, building size, backup generators, transmission lines, and the effect of a project on nearby land uses.

Local governments can address these concerns through clear standards for setbacks, sound levels, water reporting, emergency planning, design, traffic, and public hearings. These rules protect nearby property while giving developers predictable expectations.

The Beacon Center of Tennessee’s case against Hawkins County shows what can happen when a government moves from regulation to a categorical ban. The dispute involves ExoticRidge, a proposed cryptocurrency-mining data-center facility that said it would generate its own electricity rather than use the local grid.

Beacon says Hawkins County had no zoning code and that the developer received only 10 minutes’ notice before county officials considered a resolution banning all data centers and cryptocurrency-mining operations in September 2025. The county adopted the ban by a 10–2 vote, according to the complaint filed in the case.

The county’s concerns about noise, industrial activity, and rural character are legitimate policy questions. The dispute is whether a total prohibition, rather than specific and generally applicable rules, is an appropriate response. On March 31, 2026, the Beacon Center of Tennessee filed a federal civil-rights lawsuit on ExoticRidge’s behalf, alleging that the county singled out a lawful business and violated constitutional and property-rights protections. The Southern Environmental Law Center later offered Hawkins County pro bono representation in defense of the ban, adding an environmental and local-control perspective to the case. The litigation remains unresolved.

Janet Mills, 75th Governor of Maine

Maine faced a statewide version of this debate. The Legislature approved a temporary moratorium on large-scale data centers, but Gov. Janet Mills vetoed the measure because it did not exempt a proposed $550 million data-center project in Jay. Reuters reported that Mills said she supported a temporary moratorium and would have signed the bill with a carveout for the Jay project, which is expected to reuse existing infrastructure and support local jobs and tax revenue. Her veto was therefore a dispute over that specific project exemption, not a rejection of ratepayer or environmental safeguards.

These land-use questions concern a facility’s physical impact, while AI policy raises a separate question: how far should governments regulate the technology the facility supports?

Regulate Harmful Conduct, Not Computing Itself

Data centers support AI, but data-center rules are not the same as rules governing AI systems. A state can regulate power costs, land use, safety, and consumer protection around a facility without treating the ability to develop or use computing tools as harmful by default.

The Frontier Institute developed model Right to Compute legislation and helped advance Montana’s law in 2025. The law protects the ability to own and use computational resources for lawful purposes. It applies a strict-scrutiny standard when state or local governments restrict that right, requiring any restriction to serve a compelling public interest and use the least restrictive practical approach.

NIST framework for managing AI risks. National Institute of Standards and Technology.

The law also includes safety guardrails. It requires operators of AI-controlled critical infrastructure to maintain risk-management policies and retain the ability to disable AI control when necessary for safe operations. Supporters argue that this model allows governments to address fraud, theft, discrimination, safety risks, and other harmful conduct directly without treating computing itself as harmful by default.

States still face an open question: should AI policy develop through separate state laws, or could a patchwork of different rules create unnecessary compliance costs for businesses operating across the country? The answer may differ by issue. States can act quickly on local power and zoning concerns, while nationally operating technology firms may benefit from greater consistency in some areas.

What Policymakers Should Consider

The following framework gives campaign leaders five clear policy and messaging lines:

Wrap Up

Data-center policy is increasingly energy policy, consumer-protection policy, land-use policy, and AI policy at the same time. The Department of Energy’s projection that data centers could account for as much as 12% of U.S. electricity use by 2028 makes the cost-allocation question urgent.

States do not need to choose between a blank check for large technology companies and a permanent roadblock to new infrastructure. They can set neutral rules, require companies to finance the costs they create, and protect nearby communities through enforceable standards.

The durable test is straightforward: welcome private investment, avoid special favors, and make sure households, small businesses, and local communities do not carry costs that belong to the project requesting them.