In the 2024 through 2025 delivery year, PJM's capacity auction cleared at 28.92 USD per megawatt-day. The 2025 through 2026 auction increased drastically to 269.92 USD, and the 2026 through 2027 auction cleared at 329.17 USD, the Federal Energy Regulatory Commission (FERC) approved price cap. The December 2025 auction for 2027 through 2028 reached the updated cap of 333.44 USD, while clearing short of PJM's reliability requirement for the first time in the capacity market's history. Capacity costs are covered, more or less directly, by the bills of the 67 million people PJM serves across 13 states and DC, and a massive price increase in just two years occurs accordingly. PJM's independent market monitor estimated that data centers were responsible for 63% of the price increase in the 2025/2026 auction, meaning 9.3 billion USD in added costs recovered from customers. NRDC estimates the average family in PJM territory faces about 70 USD per month in increases by 2028. This isn’t limited to PJM and its service areas only: utilities nationwide requested more than 29 billion USD in electricity rate increases in the first half of 2025, double the first half of 2024.
The question underneath all of this is cost allocation: whether the large customers requiring new utility infrastructure end up paying for it, or whether residential and small commercial customers subsidize it by footing the majority of the bill. State legislatures and utility commissions have been answering this, and according to the Edison Electric Institute, as of May 2026, 23 states have approved at least one large-load tariff, with another seven pending. The principle is the same everywhere (cost-causers e.g. heavy customers pay), the mechanisms vary, and three states show the range.
Oregon built the dedicated rate class through the POWER Act, which passed in June 2025. This created a separate class for data centers and crypto facilities at 20 MW and above, as well as requiring rates to include the full cost of serving them. The Oregon Public Utilities Commission (PUC) approved Portland General Electric's (PGE) implementing tariff, Schedule 96, on May 7, 2026, which takes effect on June 10. What this does is require industrial customers in the class to pay 100% of the distribution upgrades their projects require, minimum demand charges of 90% of contracted capacity whether they use it or not, contracts running 10 to 30 years depending on size, and a 1 cent/kWh surcharge above 100 MW that funds low-income energy burden programs. According to PGE's own estimates, this should result in data center electricity rates up 29%, with a corresponding decrease in residential (1.3%) and small business (3.7%) rates.
Virginia, the state with the largest concentration of data centers, is attempting 2 regulatory pathways at once. The State Corporation Commission approved Dominion's GS-5 rate class in November 2025, effective January 2027. This will stipulate 14 year minimum contracts for loads of 25 MW and above, payment for 85% of contracted transmission demand and 60% of generation demand regardless of actual use, and collateral of 1.5 million USD per MW. The legislature did try to go further with SB 253 and HB 1393, which would have put PJM capacity auction costs onto the heavy infrastructure class outright, with the SCC estimating roughly 5.52 USD per month in residential savings and a 15.8% increase for data centers. Governor Spanberger amended this in April, replacing it with a directive that the SCC take “all steps necessary” to keep residential customers from paying the generation/distribution costs of serving data centers, and signed the amended bills in May. The cost shift in Virginia now depends on how aggressively regulators use that discretion... which is worth keeping in mind when reading headlines about what the law does.
Moving to Pennsylvania, regulators acted at the statewide level. On April 30, 2026, the Pennsylvania PUC voted 5-0 to adopt a model large-load tariff for the state's electric distribution companies to incorporate into their own filings, applying to customers above 50 MW individually (or 100 MW in aggregate). Interconnection upgrade costs are thus recovered directly from the large-load customer. In addition, deposits/collateral must be enough to fully cover upgrade costs, each utility must publish its large-load interconnection queue, and contracts must include load ramping and early-termination provisions.
There is a federal angle to all of this as well: the White House issued the Ratepayer Protection Pledge in March 2026, with seven big-tech signatories (Google, Microsoft, Meta, Amazon, Oracle, OpenAI, xAI) pledging to “build, bring, or buy” the energy for their data centers and pay its full cost. However, the pledge is voluntary and carries no enforcement mechanism. Ari Peskoe, who directs Harvard's Electricity Law Initiative, argues it “does nothing to help consumers”, since electricity rates are set by state commissions, and the White House sets none of them. I would add to this that if your organization is keeping track of cost allocation risk, the binding decisions are in the rate case dockets, not voluntary pledges.
Importantly, one risk that gets less interest is stranded cost. NV Energy has received roughly 22,000 MW of data center interest inquiries against a system peak of about 9,000 MW, and has signed agreements for about 6 GW. However, much of that interest will never be built (developers flood several markets with applications and pick the best deal), but the infrastructure planned against it is real, and without protections, ratepayers have to cover costs for assets stranded by speculative load. The long contract terms, termination penalties, take-or-pay minimums, and full collateral in the Oregon, Virginia, and Pennsylvania frameworks all exist for this reason. PJM's board took the incentive route in January 2026: an expedited interconnection track for large loads that bring their own new generation, and curtailment ahead of pre-emergency demand response for any load growth that doesn't.
Nevada is one of the more interesting cases: NV Energy's Greenlink transmission project was estimated at 2.5 billion USD in 2021 and reached 4.2 billion USD by federal approval in 2024, with around 4,000 MW of the lines' 5,000 MW capacity promised to data center operators. Residential customers in Southern Nevada are already paying a 4 USD per month rider for Greenlink West construction, approved in September 2025, and regulators have also approved a daily demand charge based on each day's highest 15-minute usage period, now postponed to January 2027 and being challenged in court by the state attorney general. The utility's 2026 resource plan projects 47% more electricity demand than it anticipated two years ago, primarily for data centers in the north. The backlash was predictable: the Reno Planning Commission voted 4-2 in February 2025 to recommend a moratorium on new data center permits, though the City Council declined to adopt it. I published an analysis of this in OGEL Energy Law Journal earlier this year, examining how IRP process design in the American West creates regulatory risk for clean energy deployment, and the Nevada case shows that cost allocation is where that risk is most acute. As I told ConsumerAffairs, the cost allocation question, whether data center operators are paying their fair share of the grid buildout they're driving, or whether residential and small commercial customers are subsidizing it, is one that utility commissions are only starting to discuss.
In sum, cost-causation is becoming the regulatory default for deciding who pays for infrastructure builds, and planning assumptions about data center power costs formed in 2023 are wrong in 23 states and counting. The Oregon model (e.g. separate rate class, full cost reflection, long contracts, exit penalties) is spreading fast, so the large-load tariff proceeding in your own service territory deserves more attention than federal announcements. If your strategy is dependent on data center load that is still speculative or unconfirmed, the new collateral and deposit requirements will materially impact the risk profile, for both sides. The next events to watch are Pennsylvania's revised final order, as well as Virginia's SCC implementation of the amended law.
Arif Gasilov leads Gasilov Group's natural resources and built environment practice. His work covers energy regulatory analysis (utility rate cases, cost allocation, grid capacity), water governance, and built environment assessment including urban heat and shade infrastructure. He is LEED accredited with a background in environmental science.
