Quick Answer
AI data centers are contributing to rising US electricity bills by drawing power from shared grids while utility rules spread infrastructure costs across all customers. New York paused large data center permits in July 2026, and New Jersey and California enacted ratepayer-protection laws. A bipartisan federal bill passed the House 417-3 in September 2026 and awaits the Senate.
Key Takeaways
- US data centers consumed roughly 4.4% of national electricity in 2023, and Lawrence Berkeley National Laboratory projects that share could reach between 6.7% and 12% by 2028.
- New York Governor Kathy Hochul signed Executive Order No. 62 on July 14, 2026, imposing the country’s first statewide moratorium on state environmental permits for data centers of 50 MW or more.
- New Jersey Governor Mikie Sherrill signed S731/A796, the Data Center Fair Share bill, on July 7, 2026, creating a separate rate class so large data centers pay for their own energy and grid infrastructure rather than shifting those costs to residential ratepayers.
- The bipartisan Ratepayer Protection Act (H.R. 9340) passed the full House 417-3 on September 16, 2026, after a 52-0 committee vote, and now awaits action in the Senate.
- Voluntary industry pledges from Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI exist but are not legally enforceable under current federal tariff rules.
How are AI data centers actually driving up residential electricity bills?
AI data centers raise residential electricity bills primarily through a cost-allocation mechanism that spreads infrastructure expenses across all grid customers rather than assigning them to the large power user that created the demand. Average electricity customers are paying for the costs of new transmission that utilities are building for data centers, according to a Union of Concerned Scientists report. The mechanism works because most existing utility tariff rules were not written with hyperscale facilities in mind, so the grid upgrades needed to connect a data center drawing hundreds of megawatts get bundled into rates that every customer pays.
Since 2020, residential electricity prices in the US have risen by more than 36%, from 12.76 cents per kilowatt-hour to 17.44 cents per kilowatt-hour in February 2026, according to the US Energy Information Administration. The agency’s October 2026 Short-Term Energy Outlook puts the average residential price at 17.30 cents per kilowatt-hour in 2025 and forecasts 18.21 cents in 2026 and 18.68 cents in 2027. Data centers are not the sole cause of that increase, which also reflects aging infrastructure investment and broader energy market conditions. That said, energy costs drove most of the recent increase in PJM wholesale power costs, but capacity costs also rose, and that matters because capacity is one area where rapid growth from large electricity users, including data centers, can affect costs if rules do not assign those costs fairly.
PJM’s 13-state region, covering the mid-Atlantic and Midwest, hosts the country’s largest data center density and saw capacity market prices rise 174% for the 2025-26 delivery year, with those costs flowing to ratepayers. For consumers in Virginia, the impact is already concrete: data centers accounted for roughly 40% of the state’s total electricity consumption in 2024, per Bloomberg analysis, and Dominion Energy’s first base-rate increase since 1992 adds approximately $8.51 per month to a typical household’s bill.
How large is the data center energy footprint, and how fast is it growing?
US data center annual energy use in 2023 was approximately 176 terawatt-hours, roughly 4.4% of US annual electricity consumption that year, according to a report by Lawrence Berkeley National Laboratory. That baseline figure is already significant, but the trajectory is the more pressing concern. Lawrence Berkeley National Laboratory predicts that data center demand will grow from 176 TWh in 2023 to between 325 and 580 TWh, or between 6.7% and 12.0% of total US electricity consumption, by 2028. The range is wide because actual AI workload growth and energy-efficiency gains remain uncertain, but even the low end of the projection represents a near-doubling of current consumption.
Power-hungry data centers that provide computing power for artificial intelligence and cryptocurrency are pushing US electricity consumption to successive record highs. In its October 2026 Short-Term Energy Outlook, the US Energy Information Administration estimates total consumption at 4,195 billion kilowatt-hours in 2025 and forecasts 4,288 billion kWh in 2026 and 4,356 billion kWh in 2027. EIA expects demand from the commercial sector, which includes data centers, to grow 2.8% in 2027, while residential consumption stays mostly flat. Researchers note that it is difficult to fully separate how much of that new demand comes specifically from AI workloads versus other data center operations, but the overall trajectory is documented by federal sources.
What has New York done, and why does it matter nationally?
New York became the first state in the US to impose a statewide moratorium on hyperscale data center permits. Governor Kathy Hochul signed Executive Order No. 62 on July 14, 2026, imposing the country’s first statewide moratorium on hyperscale data centers, stating that “as data center development threatens to hike up utility bills, deplete our natural resources, and create uncertainty for New Yorkers, it’s my responsibility to take action and lead,” and the order will pause state permitting for new large data centers while directing state regulators to create standards addressing environmental impacts, energy demand, water usage, and other factors. According to legal summaries from Davis Polk and other firms, the order holds state environmental approvals for data centers using 50 megawatts or more for up to one year, or until a statewide environmental review and new standards are complete, and it exempts facilities used mainly for manufacturing, research, education, or medical care. Projects that already hold their permits are not affected.
The New York State Legislature had passed the Responsible Data Center Development Act (S10642/A11560), which would impose a one-year moratorium on state permits for new large data centers with peak demand of 20 megawatts or more, with the bill clearing the Senate 44-16 and the Assembly 102-39 on June 4, 2026. Hochul issued the narrower executive order instead of signing that bill, so the act’s 20-megawatt threshold is not in force. The act also carries cost-allocation provisions: utilities would be required to create separate service classifications for large data centers, assigning infrastructure upgrades, administrative and operational costs, and commodity-price increases attributable to large data centers to that class rather than to general ratepayers.
The national significance is that New York’s action creates a replicable model. Within 30 days of the order, by August 13, 2026, the Empire State Development agency opened the Community Investment Framework for public comment, and municipalities and developers may submit input while the framework is still being shaped. Other states watching this process may adopt or adapt similar moratorium-plus-cost-allocation structures.
What has New Jersey done to protect ratepayers?
New Jersey enacted what the Natural Resources Defense Council described as trailblazing ratepayer protection legislation. Governor Mikie Sherrill signed S731/A796 on July 7, 2026, alongside two other utility bills, and the governor’s office says the law creates a new ratepayer class and rate structure for data centers and requires them to cut back during grid strain before residential customers are affected. The governor also announced a $25 bill credit for all 3.6 million New Jersey ratepayers, with an additional $150 payment for lower- to moderate-income households. The New Jersey legislature passed S731/A796, a bill that will ensure electricity remains affordable, reliable, and increasingly clean as data centers come online, with the bill requiring that large data centers meet new criteria designed to protect ratepayers, strengthen grid reliability, and accelerate construction of new clean energy resources.
The mechanics of the New Jersey law address the cost-shift problem directly. Earlier versions of the bill applied to data centers drawing at least 100 megawatts, but the version that passed lets the Board of Public Utilities define a large data center for each utility, with the size threshold set no higher than 50 megawatts, according to DatacenterDynamics and the New Jersey Senate Democrats. Utilities must also obtain financial guarantees that new large data center customers will take at least 85% of the service they request for at least 10 years, which is intended to stop stranded infrastructure costs from boosting residential ratepayers’ bills. The law also includes a grid reliability mechanism: the Board of Public Utilities must monitor policies across PJM states, and New Jersey’s requirements activate only when a majority of PJM states adopt similar mandates, reflecting the legislature’s judgment that unilateral action could disadvantage New Jersey by driving investment elsewhere while still imposing regional grid impacts.
The context behind the legislation is important for New Jersey residents in particular. New Jersey residential electric bills rose roughly 20% in 2025, with AI-related data centers accounting for approximately 70% of new state power demand, and as part of the 13-state PJM market, New Jersey ratepayers face capacity cost increases from data center growth regionwide. That regional interdependency is precisely why the coordination trigger was written into the bill.
Where does the federal Ratepayer Protection Act stand, and what would it do?
Congressman Gabe Evans and Congresswoman Kathy Castor introduced the bipartisan Ratepayer Protection Act to ensure American families are not left footing the bill for the grid upgrades and increased energy generation required to develop new data centers, with the act proposing standards for states and their Public Utility Commissions when connecting large-load customers such as data centers and hyperscalers to the grid. The House Energy and Commerce Committee approved H.R. 9340, sponsored by Reps. Gabe Evans (R-Colo.) and Kathy Castor (D-Fla.), in a 52-0 vote on July 21, 2026, with the proposal requiring states to consider adopting a federal standard directing data centers to pay the full cost of new generation and transmission upgrades needed to serve them.
The full House passed the Ratepayer Protection Act 417-3 on September 16, 2026, under a suspension of the rules that requires a two-thirds majority, according to the National Association of Counties. The bill would establish federal standards that state regulators must consider, but are not required to adopt, when connecting new data centers. The committee text applied to large-load customers with a demand of 100 megawatts or more, and later amendments narrowed its scope to data centers. Sen. Jon Husted (R-Ohio) sponsors a Senate version, but on September 17, 2026, Sen. Martin Heinrich (D-N.M.) objected to Husted’s request to pass it by unanimous consent, saying the bill falls short, and Sen. Bernie Moreno (R-Ohio) then objected to Heinrich’s own GRID Savings Act. The bill therefore still needs Senate passage and the President’s signature before it could become law, and its passage is not guaranteed.
On the industry side, a voluntary commitment already exists but carries no legal weight. The Ratepayer Protection Pledge is a voluntary, nonbinding commitment signed on March 4, 2026, by Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI to build, bring, or buy the electricity their data centers need and pay for all required grid infrastructure upgrades, rather than passing those costs to ratepayers. Consumer advocates and state regulators have noted that voluntary pledges cannot override existing tariff structures that legally require cost socialization across all customers. Understanding the limits of such pledges is important context as readers track progress on the OpenAI and broader AI industry’s financial obligations to the public grid.
Which states and regions face the steepest risk to household bills?
| State / Region | Key Risk Factor | Legislative Status (as of Oct. 14, 2026) |
|---|---|---|
| Virginia | Data centers ~40% of state electricity use; Dominion rate increase adds ~$8.51/month to typical bill | 15 bills enacted this session; court invalidated some project approvals |
| New Jersey | Residential bills up ~20% in 2025; data centers ~70% of new state power demand | S731/A796 passed June 30, 2026; signed into law July 7, 2026 |
| New York | Residential rate up 12.1% year-over-year as of May 2026 (EIA data) | Executive Order No. 62 (July 14, 2026) pauses state permits for data centers of 50 MW or more |
| PJM Region (13 states) | Capacity market prices rose 174% for 2025-26 delivery year | Multi-state tariff coordination under discussion; federal bill passed the House Sept. 16, 2026, pending in Senate |
| California | SB 886 directs the CPUC to set data center tariffs that keep grid costs from shifting to other customers | Signed Sept. 21, 2026, in a seven-bill package; CPUC tariffs due by Jan. 1, 2028 |
Maryland, Washington D.C. and the broader PJM territory have the steepest documented residential bill increases, and California and the Pacific Northwest are next in projected impact. Consumers in states not currently covered by ratepayer-protection legislation remain most exposed to the cost-shift mechanism described above. One practical indicator worth monitoring is your state’s public utility commission docket page, where proposed rate cases are published before they take effect.
What can individual consumers do right now?
Individual consumers have limited direct influence over how utilities set rates, but several practical steps can reduce exposure and increase accountability. First, check whether your state has a public utility commission that accepts consumer comments on pending rate cases. Rate cases are the formal proceedings where utilities propose increases, and public comment is part of the legal record. Your state PUC website will list open dockets and comment deadlines.
Second, audit your own electricity usage to reduce your bill regardless of rate decisions. The US Department of Energy’s Energy Saver resource covers practical steps including smart thermostat settings, appliance efficiency ratings, and time-of-use rate plans that let you shift usage to off-peak hours when grid demand, and often pricing, is lower. In states with competitive electricity markets, switching suppliers through your state’s regulated comparison portal can also lower your per-kilowatt-hour rate, though transmission and delivery charges set by the utility remain fixed regardless of supplier choice.
Third, contact your state legislators if you believe data center cost-shifting is affecting your bill. The legislative activity described in this article moved quickly in 2026 partly because constituent pressure created political urgency. Because data and privacy concerns often intersect with infrastructure decisions, being informed about both the energy and policy dimensions helps you engage more effectively. Consumers in the PJM region in particular should follow their state’s position on multi-state tariff coordination, since a single state acting alone produces limited protection when grid costs are set regionally. The PJM capacity market reports are publicly available and updated each delivery year, making them a reliable primary source for tracking how regional grid costs are changing.
FAQ
Will AI data center laws actually lower my electricity bill?
The new state laws are designed to stop further bill increases rather than reduce current bills. New Jersey’s S731/A796 and New York’s moratorium order both aim to prevent future infrastructure costs from being socialized across all ratepayers, but they do not reverse rate increases that utilities have already received approval to charge. The practical benefit is protection against additional data-center-driven cost shifts going forward, not a rollback of existing rates.
Does using AI tools like ChatGPT raise my electricity bill directly?
Using AI tools does not add a charge to your personal electricity bill in any direct or itemized way. The cost-shift mechanism works at the grid level, where utilities spread the infrastructure expenses of large data centers across all customers through base rates and transmission charges. Your individual usage of AI services contributes to aggregate data center demand, but the resulting bill impact, if any, arrives as part of a general rate increase rather than as a line item tied to your AI usage.
Which states have the strongest ratepayer protections against data centers as of August 2026?
New Jersey and New York had the most advanced ratepayer-protection frameworks in place as of August 2026, and California joined them when Governor Gavin Newsom signed SB 886 and six related data center bills on September 21, 2026. New Jersey’s S731/A796, signed July 7, 2026, creates a separate rate class for large data centers (a threshold of no more than 50 MW) and requires guarantees that they take at least 85% of the service they request for 10 years. New York’s Executive Order No. 62 pauses state permits for data centers of 50 MW or more and directs regulators to build new standards. At least 18 states have introduced bills creating special rate classes for large energy users, with some requiring data centers to fund infrastructure improvements and demonstrate benefits to ratepayers.
What is the Ratepayer Protection Pledge, and is it enforceable?
The Ratepayer Protection Pledge is a voluntary, nonbinding commitment signed on March 4, 2026, by several major AI and technology companies including Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI. The signatories committed to covering the full cost of new electric generation resources needed to meet their energy demands, but the pledge does not have legal enforcement mechanisms. Existing federal tariff rules still require grid costs to be spread across all customers by default, which is why state legislation and the federal Ratepayer Protection Act are necessary for the commitment to have practical effect.
Could data center moratoriums slow AI development or affect the services I use?
State moratoriums pause new permit issuance for large facilities but do not shut down existing data centers or interrupt services running on already-built infrastructure. New York’s moratorium applies to state environmental permits for facilities of 50 MW or more and lasts up to one year while regulators develop standards. More than 300 data center-related bills have been filed across more than 30 states in 2026, at least 12 states have considered moratorium legislation, and more than 100 local moratoria have been adopted nationwide, with an estimated $64 billion in projects blocked or delayed by community opposition. The longer-term effect on AI service availability depends on whether companies can redirect planned capacity to states with less restrictive regulatory environments, which several industry groups have warned is a likely outcome. Users relying on AI tools should monitor service announcements from their specific providers rather than drawing broad conclusions from any single state’s action.
