In this article:
- How AI data centers can affect household electric bills
- Why new grid costs can end up in household electric rates
- What policies could make large energy users pay their fair share
Data centers that support artificial intelligence, or AI, use enormous amounts of electricity, and demand for those facilities is growing quickly. Because they’re connected to the same power grid that supplies our homes, that extra demand affects the whole system.
When demand for electricity increases, utilities and grid operators must ensure there’s enough power available at all times. That can mean building new power plants, upgrading transmission lines, and paying to keep extra electricity capacity available when demand is high.
Right now, in most cases, those costs don’t just go to the companies running the data centers. Instead, they’re spread across everyone who uses the grid. That’s because electricity rates have not been uniformly adjusted to require data center companies to pay for all the costs they impose on the grid. So even if you’re not using AI directly, you may still see higher monthly bills because you’re helping pay for the expanded infrastructure and reliability needed to support that growth.
One way to think about this phenomenon is like a growing town where a large new facility moves in and needs a lot of power. The local infrastructure must improve and expand, and the cost of those upgrades is shared across the community, without a careful redesign of rates and fees.
What “fair share” means for AI data centers
When we say data centers should pay their “fair share,” we mean they would cover the costs of the new infrastructure and services their electricity use requires, instead of those costs being spread across all customers.
Today, when demand rises due to energy-intensive facilities like data centers, grid operators such as PJM Interconnection plan for new power plants or upgrades, and electric utilities in PJM recover those costs from everyone on the system. That means households can end up helping pay for investments driven by large new users.
Under a “fair share” approach, more of those costs would be assigned directly to the data centers that are creating the demand.
For consumers, the effect would likely be gradual. New rules would not necessarily lower bills right away, but they could help limit future increases by keeping more data center-related infrastructure costs from being passed on to households.
The costs of new generation capacity for data centers are only a small part of our electricity bills, but Pennsylvania Governor Josh Shapiro’s office, which has researched PJM capacity in depth, believes that reining in these ballooning costs caused by data centers could make a meaningful difference in our individual bills.
In other cases, average electricity customers are paying for the costs of new transmission that utilities are building for data centers, as a Union of Concerned Scientists report shows. Policymakers and regulators could help prevent those costs from being passed on to consumers by requiring data centers to pay directly for the transmission upgrades needed to meet their electricity demand, rather than spreading those costs across all customers.
What do these cost categories mean?
Wholesale power costs are not the same as a household electric bill, but they are one part of the larger electricity system that can influence what customers eventually pay. These costs are often grouped into several categories:
- Energy: The cost of producing or buying the electricity used by homes, businesses, and other customers.
- Capacity: Payments to make sure enough power is available during times of high demand, even if that power is not needed every hour.
- Transmission: Costs tied to moving electricity across the high-voltage power grid from where it is generated to where it is needed.
- Other: Smaller costs related to operating the power system and maintaining reliability.
Policies that could help protect household bills
First, requiring data centers to pay directly for the new infrastructure they need would help ensure those costs aren’t passed along to everyone else. Without that kind of policy, those expenses can push household bills higher.
Second, states should consider requiring data centers to commit to minimum contract terms with utilities. States should also consider requiring that data centers pay “exit fees” to utilities if those data centers close or move out of the utility’s service territory before the contract period. This ensures that a utility’s ratepayers aren’t stuck paying for infrastructure investments that the utility made for the data center before the data center closed or moved.
Finally, increasing transparency around the deals utilities make with large energy users would allow regulators and the public to better understand who is paying for what. When those agreements are clear and open to review, it’s easier to prevent costs from being shifted onto residential customers.
Together, these changes help make sure that the companies driving new demand are responsible for the costs they create, which can help keep household electricity bills from rising as quickly.
Related research
The policies highlighted here are part of a broader set of recommendations in Nine Ways to Address the Energy Impacts of AI Data Centers, a Vanderbilt Policy Accelerator report co-authored by Hannah Wiseman and Matthew McHale.
For a deeper legal analysis of how large electricity loads, including data centers, are affecting grid access, rates, and governance, read Wiseman and McHale’s related paper, Governing the Energy Bottleneck.
How these policies could work in a community
Imagine a utility serving a region (say, one-third of a state) where several companies are proposing to build new AI data centers. Those facilities would use a large amount of electricity, and the utility needs to decide how much new infrastructure to build and who should pay for it.
One tool is more careful forecasting of new data center demand. If the utility closely reviews those proposals, it might find that some projects are still early in development, may not ultimately be built, or are being considered in multiple locations. By avoiding overestimating demand, the utility can avoid building more power plants or transmission lines than are needed. That helps keep overall costs and rate increases lower for everyone.
Another tool is a large-load tariff, a pricing structure designed specifically for very large energy users, such as data centers. With this approach, the utility can require those customers to cover more of the costs tied to their electricity use, rather than spreading those costs across all households.
Together, these policies can make a difference. More accurate planning reduces unnecessary spending, and targeted pricing ensures that large new users pay for the costs they create. For a local community, that can mean smaller rate increases over time and fewer households at risk of shutoffs due to rising bills.
Why the next few years matter
One reason for optimism is that many states are starting to adopt special pricing structures for very large energy users, such as data centers, that include the structures mentioned above—large-load tariffs, minimum contract terms, and exit fee requirements. These are designed to ensure that companies driving major increases in electricity demand pay more directly for the infrastructure and services they require and don’t saddle customers with legacy costs. Some states are even requiring data center companies to pay annual fees. These fees go into state funds to support energy conservation and help customers who struggle to pay electricity bills.
At the same time, there are still real concerns. One is that the rapid growth in AI could outpace expectations or even shift direction. If utilities build significant new infrastructure for data centers that later close, scale back, or relocate, those costs could still be spread across the remaining customers, who could be left paying for abandoned infrastructure that is difficult to repurpose for other uses.
Some of the stronger policies try to address this risk by requiring data centers to provide financial guarantees or pay exit fees if they leave earlier than expected. But not all states have adopted these protections yet.
So, while there is meaningful progress, the key challenge for policymakers is making sure these safeguards are in place before large new costs are locked into the system.
Hannah Wiseman is a professor of law, a professor and Wilson Faculty Fellow in the College of Earth and Mineral Sciences, and an Institute of Energy and the Environment faculty member. Her teaching and research focus on energy law, environmental law, land use regulation, oil and gas law, and administrative law, including how regulation and governance can respond to changing energy systems and expanding demands on infrastructure.
