WASHINGTON — For most of modern American politics, the electricity bill was background noise: a monthly charge that rose slowly, drew little attention, and rarely decided an election. That has changed in the run-up to the November 3, 2026 midterms, when voters will elect all 435 members of the House of Representatives and 35 senators and decide which party controls Congress. According to data from the United States Energy Information Administration cited by Reuters on October 31, household electricity prices have risen more than 10 percent since 2024, when President Donald Trump was elected alongside Republican majorities in both chambers of Congress, and that increase has outpaced the rise in prices for other goods and services. In several battleground states that are also hot spots for new data centers, including Ohio, Michigan and Pennsylvania, bills have climbed faster still, in some cases by three times the rate of inflation. The result is that a subject once left to utility commissions has become a closing argument in the fight for Congress: who should pay for the enormous new demand that artificial intelligence and cloud computing place on the power grid, households or the companies building the data centers?
Congress has already had one clear chance to answer that question, and it produced one of the most lopsided votes of the year followed by a Senate failure. The House of Representatives passed the Ratepayer Protection Act, designated H.R. 9340, by 417 votes to 3 on September 16, 2026. Two weeks later, on September 30, the Senate fell short of the 60 votes needed to advance the measure, with 57 senators in favour and 43 against, and the motion to proceed was withdrawn the same day. The bill is therefore not law, and nothing in it will lower any household bill on its own. But the split between the two chambers, and the argument over why the Senate stopped a bill the House had approved almost unanimously, now frames what the next Congress may do about power prices, and it gives voters a concrete test of how each party says it would handle the cost of the artificial intelligence boom.
The stakes are not abstract. Utilities are on track to match last year’s record of about 31 billion dollars in rate increase requests to state regulators, according to a survey by the nonprofit group PowerLines cited by Reuters, after last year’s total had already doubled the previous record. Investor-owned utilities are expected to carry out nearly 1.3 trillion dollars in capital spending over the next five years, according to an analysis by S and P Global’s Regulatory Research Associates. Credit rating agency Fitch downgraded the North American utility sector to deteriorating from neutral in June 2026, citing political pressure in the midterms to limit electricity bills and utility profits. An Associated Press-NORC poll in October found that electricity bills were a major source of stress for 36 percent of American adults. When a bill arrives each month, voters do not need a policy briefing to feel the change, and candidates in competitive House and Senate races know it.
How data center power bills became a 2026 midterm issue
The story begins with demand. A report prepared for the United States Department of Energy by Lawrence Berkeley National Laboratory found that data centers used about 176 terawatt-hours of electricity in 2023, roughly 4.4 percent of all electricity used in the country, up from 58 terawatt-hours in 2014, and projected that the share could reach between 6.7 percent and 12 percent by 2028 depending mainly on how fast artificial intelligence computing grows. The International Energy Agency has estimated that a typical artificial intelligence-focused data center uses as much electricity as 100,000 households, and that the largest facilities now under construction will use many times more. The agency projects that data center electricity demand will more than double by 2030. Those are national averages, but the pressure is felt locally first: a single very large facility can change the load forecast for a whole region, and the grid upgrades built to serve it are paid for through a system of rates that state regulators approve.
That is where household bills enter the picture. A residential electricity bill pays for three broad things: the power itself, the high-voltage transmission lines that move it across a region, and the local distribution network that delivers it to the home. When a utility builds new generation, transmission or distribution capacity to serve a very large new customer, regulators must decide how the cost is recovered. If the cost is spread across all customers, households help pay for infrastructure built largely for one facility. If it is charged to the facility, the facility bears the cost of its own arrival. Supporters of the House bill argue that the second outcome is simple fairness. Critics, including utility advocates, argue that freezing rates or cutting the returns utilities are allowed to earn will not meaningfully reduce household bills and could discourage the long-term investment the grid needs. The president of the Edison Electric Institute, the trade organisation for investor-owned utilities, was quoted by Reuters as saying that the grid requires consistent investment every year and does not operate on the political calendar, a point that captures the industry’s warning that short-term political fixes can have long-term costs.
The regional picture makes the argument sharper. In much of the eastern United States, a grid operator called PJM Interconnection, which covers 13 states and the District of Columbia, runs auctions that pay power plants to be available years in advance. Reporting on those auctions has described record prices driven in large part by forecast demand growth from data centers, with the auction for the year starting in June 2027 clearing at 333.44 dollars per megawatt-day and still falling about 6,600 megawatts short of its reliability target. PJM itself has said that nearly all of the forecast growth came from data centers. For households in that region, the auction is not a distant market mechanism: it feeds into the supply portion of bills, and it helps explain why a national debate about data centers can feel, in Pennsylvania or Ohio or Michigan, like a local grievance.
Nor is the pressure limited to one region or one party’s territory. Seats on nine state public utility commissions are up for election this year, covering nearly all the states where voters directly elect utility regulators, including the growing data center hubs of Arizona and Georgia. Governors will be chosen in 36 states, and governors appoint utility regulators in states where commissioners are not directly elected. The utility issue also featured prominently in state elections last year, when candidates who promised to hold down bills found a receptive audience. That history matters for Congress because it shows the issue is not a single campaign’s invention: voters have already rewarded politicians who talk about power prices, and the 2026 congressional campaigns are following that signal.
What the Ratepayer Protection Act would actually do
The Ratepayer Protection Act was introduced in the House on June 18, 2026, by Representative Gabe Evans, a Republican from Colorado, with bipartisan support including Representative Kathy Castor, a Democrat from Florida. The House Energy and Commerce Committee ordered it reported by a 52 to 0 vote on July 21, and the full House passed it, as amended, on September 16 by 417 to 3 under suspension of the rules, a procedure normally reserved for measures expected to command broad support and requiring a two-thirds majority. The bill would amend the Public Utility Regulatory Policies Act of 1978 to create a federal standard for large-load customers. A large-load customer is defined as a nonresidential electricity consumer that requests or enters into an agreement for one or more facilities with peak electricity demand of at least 100 megawatts at a single site or campus. Data centers are the example most often cited, but the definition is written by size, not by industry.
Under the standard, state utility regulators and nonregulated utilities would be required to consider, within a set timetable, a rule under which a utility’s rate for a large-load customer must be designed to recover from that customer the full, incremental cost of any generation, transmission or distribution upgrade needed to serve the load. The utility would also have to require the large-load customer to provide financial assurances or contributions to cover those upgrade costs before the utility makes them, and the cost recovery would apply even if the customer later ends its agreement or stops buying electricity. The Congressional Research Service summary of the bill identifies data centers as an example of the kind of customer the measure is aimed at, and the Congressional Budget Office has estimated that the measure would add no federal direct spending or deficits.
That last detail is also the source of the fiercest criticism. The bill would require states to consider a standard; it would not itself impose a nationwide rate. A state could go through the consideration process and decline to adopt it, and no household bill would fall immediately as a result. Senate Democratic Leader Chuck Schumer called the measure a fraud, in plain terms, and argued that its optional character meant it imposed no real requirement at all, saying in debate that the Republican approach made the standard optional, which meant zero, while Democrats wanted it mandatory. Democrats also pointed to the political context: the bill’s Senate sponsor was seeking re-election, and some Democrats argued that passing the measure would hand that campaign a talking point. Republicans answered that the bill was the only available vehicle before the election to force regulators to confront the question, and that blocking it showed Democrats were unwilling to accept a bipartisan House product on affordability.
Democrats have offered a competing approach. Senator Martin Heinrich, a Democrat from New Mexico, sponsored legislation that would direct the Federal Energy Regulatory Commission to issue rules for large electricity customers requiring them to pay for the grid upgrades their demand makes necessary. The difference between the two approaches is therefore not whether large customers should pay, but how forcefully federal law should make them do so: a federal standard that states must consider, or a federal rule that the national regulator must impose. That distinction is technical, but it is also the whole argument. A consider-the-standard law leaves the final decision with state regulators, who already control most retail rates. A mandatory federal rule would move the decision to Washington. Voters choosing a new Congress on November 3 will, in effect, choose which of those philosophies has the votes to proceed in 2027.
Why the Senate vote failed despite a 417-3 House vote
The Senate’s 57 to 43 vote on September 30 was a cloture vote on the motion to proceed, not a vote on final passage, and it fell seven votes short of the 60 needed to overcome a filibuster. All 43 opponents were Democrats and independents voting together, while supporters included Republicans and, by the arithmetic of the tally, some Democrats who were willing to advance the measure. The bill was never debated on its merits in the Senate, and no amendment was adopted. After the vote, the motion to proceed was withdrawn, and both chambers left Washington for the campaign season, the House having already departed in mid-September. Any data center cost bill in the next Congress will therefore have to be reintroduced, and its fate will depend on the majorities voters elect on November 3 and on whether the new Senate can find 60 votes for either the consider-a-standard model or a mandatory rule.
The politics of the vote were unusually open. Senate Republicans had reserved one of the chamber’s final votes before the recess for the data center bill precisely because they believed it would show voters they were acting on energy costs. Senate Democrats, for their part, calculated that blocking a measure they described as toothless was better than passing a bill they believed would let supporters claim the problem solved while households kept paying. Both calculations may be partly right, which is why the episode is a useful guide to the next Congress rather than a simple story of obstruction. A bill that passes the House 417 to 3 and dies in the Senate 57 to 43 has not been rejected by Congress as a whole; it has been caught between two different ideas of what counts as action, and between a House that could pass it under suspension and a Senate that requires a supermajority to do almost anything.
Outside Congress, some of the largest technology companies have tried to blunt the argument with a voluntary commitment. In March 2026, seven artificial intelligence and cloud companies, Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI, signed a Ratepayer Protection Pledge at the White House, committing among other things to build, bring online or buy the electricity their data centers require, to pay for transmission upgrades, and to pay agreed rates whether or not electricity is actually used. Supporters of the pledge present it as evidence that the industry is willing to pay its own way without new legislation. Critics note that a pledge is not a statute: it binds only its signatories, it is not enforced by regulators in the same way as a rate, and it does not cover every large-load customer. The existence of the pledge, like the House vote, shows that the political pressure is real; neither, on its own, settles who pays.
What voters in battleground states are actually paying
The national averages hide wide local variation, and that variation is why the issue lands differently in different congressional districts. The Energy Information Administration estimates that the average residential electricity price was 17.3 cents per kilowatt-hour in 2025 and will average 18.2 cents in 2026, an increase of about 5 percent, with a smaller rise to 18.7 cents forecast for 2027. Reuters reported that household prices have risen more than 10 percent since 2024, and that in Ohio residential bills have grown by more than 20 percent over the last two years. Those figures measure prices, not total bills, which also depend on how much electricity a household uses, on weather, and on fixed charges. But the direction is consistent across the measures: power is getting more expensive faster than most other things households buy, and the increase is steepest where new data center demand is concentrated.
Local ballots are already reflecting that pressure. Reporting in late October counted at least 39 local measures on data centers in eight states on this year’s ballots, more than half of them in Ohio, most concerned with whether and where facilities can be built rather than with electricity rates directly. Those measures can address noise, water use, land use and tax deals, but they have limited power over prices, which are set by state regulators and, in regions like PJM, by regional markets. The distinction matters for voters who hope that a local vote on a data center will lower their bill: it may decide whether a facility is built, but it will not by itself rewrite the rate that pays for the grid. That is why the congressional debate, however technical, is the one that could change the cost-sharing rule itself, and why the failure of the Senate vote leaves the question open for the next Congress.
There is also a counter-argument that deserves a fair hearing, and it comes from the utilities and from some economists rather than from either party’s campaign. Grid investment is long-lived: transmission lines and substations built today will serve households for decades, and spreading their cost over time and across customers is how regulated utilities have traditionally financed them. If large customers are charged the full incremental cost up front, supporters say, households are protected. If that charge is set too high or too rigidly, utilities warn, some projects may not be built, or investment may slow, and the grid may be less reliable when demand keeps growing. Neither side has published a definitive accounting of how much of the recent bill increase is attributable to data centers as opposed to other forces: rising natural gas prices, which set the price of much of the country’s electricity, the cost of replacing ageing poles, wires and substations, hardening the grid against storms and wildfires, and meeting demand from new factories. Data centers are one cause among several, and any honest account of the bill must say so.
What the next Congress will have to decide
When the new Congress takes office in January 2027, the data center question will still be waiting, and it will arrive alongside a crowded calendar that includes funding the government and the annual defence policy bill. The next House and Senate will have to decide whether to reintroduce the consider-a-standard model, to pursue a mandatory federal rule through the Federal Energy Regulatory Commission, to leave the matter to state regulators and voluntary pledges, or to combine those approaches. They will also have to decide whether to pair any cost-sharing rule with measures on permitting, transmission planning and grid reliability, since a rule about who pays for upgrades does not by itself build the upgrades faster. Those choices will be shaped by the majorities elected on November 3, by the size of those majorities, and by whether 60 senators can be found for any version of the bill.
For voters, the practical question is simpler than the legislative mechanics. The November 3 election will not itself lower a power bill, and no candidate can promise otherwise with credibility. What it will do is decide which party writes the next version of the rule, and whether that rule is a nudge to state regulators or a mandate from Washington. The House has already shown that a consider-a-standard bill can command near-unanimous support. The Senate has shown that near-unanimity in one chamber is not enough. Until a new Congress bridges that gap, households in data center hot spots will keep paying bills that reflect a grid built for a demand surge they did not ask for, and candidates will keep arguing about who should pay for it. That argument, more than any single vote, is why electricity prices have become a midterm issue for the first time in living memory.
Sources
USN reviewed reporting from multiple outlets for this story. Read more from Tribune, Broadband Breakfast and IEN.