Data centers are not the primary driver of rising electricity bills in the United States, according to new research; instead, state-level energy policies that deprioritize affordability are responsible for rate increases. A study from the Institute for Energy Research found no statistically significant correlation between the number of data centers in a state and its current electricity prices. In fact, states with the highest concentrations of data centers have electricity prices virtually identical to the national average.

The confusion is understandable. Data centers, especially those powering artificial intelligence (AI) systems, consume enormous amounts of electricity. A single ChatGPT query uses about 0.34 watt-hours, and at scale, ChatGPT consumes roughly 850 megawatt-hours per day across 2.5 billion queries, equivalent to the annual electricity consumption of about 29,000 U.S. homes. With AI demand growing exponentially, data centers have become a convenient scapegoat for policymakers facing constituent complaints about high energy costs.

Why Are Electricity Rates Actually Rising?

The real culprit is state policy, not data center expansion. Research from the Institute for Energy Research concluded that "the discrepancy in electricity rates across the country is driven more by state-level policies than by the proliferation of data centers". States that have embraced aggressive renewable energy mandates, 100 percent carbon-free targets, premature coal and nuclear retirements, rooftop solar cost-shifting, and restrictions on natural gas infrastructure consistently deliver the nation's highest electricity prices.

North Carolina provides a stark example. The state's utility law previously prioritized adequate, reliable power as affordably as possible. However, policymakers have shifted priorities in recent years by requiring utilities to add increasing amounts of renewable energy and to shut down coal plants, with eventual natural gas plant closures planned. These ideological mandates have destabilized the energy market without building sufficient replacement capacity, creating a perfect storm for ratepayers.

How Much Electricity Do AI Data Centers Actually Consume?

The scale of AI data center power consumption is genuinely unprecedented. Global AI data centers already consume roughly three times the electricity used by Bitcoin mining operations, which themselves use between 138 and 175 terawatt-hours annually. The International Energy Agency (IEA) projects that all data centers globally will consume 415 terawatt-hours in 2024, growing to 945 terawatt-hours by 2030 and approximately 1,200 terawatt-hours by 2035, with AI's share of that power rising from roughly 14 percent toward 35 to 50 percent.

However, this growth does not automatically translate to higher consumer electricity bills. The Institute for Energy Research found that the greater electricity demand from data centers actually allowed utilities' high fixed costs to be spread across a larger sales base, resulting in smaller rate increases than would otherwise occur.

What Policy Solutions Could Protect Ratepayers?

The John Locke Foundation's report on energy poverty in North Carolina recommends several policy changes to restore affordability and reliability as top priorities in state electricity policy. The most significant recommendation specifically addresses data centers and large-load users:

• Consumer-Regulated Electricity (CRE) Legislation: This approach would allow data centers to supply, bring, or buy their own power without impacting residential ratepayers. The legislation would enable large-load users to procure electricity off the grid, preventing utilities from charging household customers for new power plants built solely to serve data center demand.
• Restore Affordability as a Priority: State leaders must put affordability and reliability ahead of ideological energy mandates, similar to how Florida and Louisiana residents benefit from policies that balance environmental goals with practical energy needs.
• Diversify Energy Sources: Rather than mandating rapid transitions away from reliable baseload power sources like natural gas and nuclear, states should maintain a balanced energy portfolio that can meet growing demand without destabilizing rates.

The CRE approach has bipartisan backing. Both Governor Josh Stein and President Donald Trump have promoted similar legislation, and Duke Energy, North Carolina's largest utility, recently signed onto the Trump administration's "Ratepayer Protection Pledge," which calls on hyperscalers and AI companies to "build, bring, or buy" their own power.

What Does This Mean for Data Center Growth?

The research suggests that data centers themselves are not inherently problematic for electricity affordability. The problem arises when states simultaneously pursue aggressive renewable mandates while retiring reliable power plants, then expect utilities to build new capacity to serve data center demand. Under current regulatory frameworks, utilities can pass the cost of new generation facilities to all ratepayers, including low-income households already struggling with energy poverty.

By allowing data centers to procure their own power through CRE legislation, states can accommodate the infrastructure needs of the AI era without burdening residential customers. This approach separates the question of whether data centers should exist in a state from the question of who pays for the power infrastructure they require. The evidence suggests that with proper policy design, both data center growth and affordable electricity for households are achievable goals.