Dominion Virginia: Data Centers Drive the Grid Buildout, Households Are Billed Its Largest Share

The chart shows three representative Dominion customer profiles. Residential (Schedule 1, 1,000 kWh/mo) climbed roughly 38%; mid-commercial (GS-2, 30 MWh per month, 75 kW peak) climbed roughly 36%; large-load (GS-4, 8 GWh per month, 15 MW peak) climbed roughly 34%. The divide is in what each class pays in absolute terms (residential pays 17.6 cents per kWh all-in against the large-load profile's 8.5 cents, nearly twice the rate, on the complete-bill basis) and what share of the buildout it is allocated.

A few events shape the trajectories. A mid-2022 fuel-price spike (the Russia-Ukraine pass-through) lifts all three lines roughly in parallel, then partly resets. Virginia's HB 1770 restored biennial rate reviews in 2023 after the prior rate-freeze era: a regulatory regime change that enabled the later rider and base-rate buildout rather than a visible step on the line (the recurring June bumps are the seasonal generation reset, not a one-time event). The single largest durable build in the series is the mid-2025 rider ramp, with offshore wind, RPS, storage, SMR, and Rider T1 transmission all stepping up between June and September 2025. And an early-2026 base-rate revision (Dominion's 2025 biennial review, PUR-2025-00058) drives the visible final step upward. Those last two events are why nearly three-quarters of the residential increase landed in the final year of the window.

The buildout now entering rates responds to load growth without precedent in this system, and it is being financed under allocation rules set decades ago.

A larger structural change is also underway: in May 2026 Dominion agreed to be acquired by NextEra Energy in a roughly $67 billion all-stock merger, with applications to the Virginia SCC and federal regulators expected in the third quarter of 2026. As part of the deal, the companies have proposed $2.25 billion in customer bill credits across Virginia and the Carolinas, spread over two years after close, with about $1.8 billion (roughly $10 a month per residential customer) of it in Virginia. The announced terms propose bill credits, not changes to class cost allocation, so the framework this case study traces stays in place as filed.

Dominion's load-growth signal is now visible at multiple layers of the system. The Dominion zone's peak load reached 25.2 GW (25,168 MW) on February 9, 2026, up 43% from the zone's 2019-20 winter peak of 17.5 GW, a new high above the 24.7 GW set in January 2025 (Grid Flexibility analysis of PJM DOM-zone hourly load; the zone now peaks in winter, not summer).# Dominion's 2024 Integrated Resource Plan (PUR-2024-00184) and its 2025 Update (PUR-2025-00184) indicate roughly 9 GW of new data-center peak demand over ten years (more than a third above today's peak), with total electricity usage rising more than 100% over a 15-year horizon. The volume of large-load requests in process has grown faster still: from 8 GW in 2021, to 23 GW in July 2024, to 47 GW in July 2025, to roughly 70 GW by February 2026 per Dominion's application in PUR-2026-00011 (as of Dec 31, 2025).#

This article treats those figures as different demand states, not interchangeable totals: forecast peak demand (planning), queue requests (pipeline), and contracted/energized load (realization).

Per Dominion's Q4 2025 earnings disclosures, the contracted stages below are reported; the request-only balance and the ~70 GW total come from Dominion's SCC filing in PUR-2026-00011 (as of Dec 31, 2025). Each request is at one current stage; the rows don't overlap.

Separately, per Dominion's SCC filing, approximately 25 GW of the 70 GW total have been assigned projected connection dates through December 31, 2031; the remaining 45 GW are still under study to determine connection dates. Dominion also reports receiving roughly ten new large-load DP requests per month, representing 2-3 GW of additional requested load each month. The standards apply to requests serving data-center customers in the DOM Zone at approximately 100 MW or more. As of Dominion's Q1 2026 earnings (reported May 1, 2026), contracted capacity across the binding, conditional, and engineering-study stages had risen to roughly 51 GW as of March 31, 2026, about 2.5 GW above the December figures in the table.

A note on forecasts. For external context, PJM's 2026 Load Forecast Report projects roughly 32 GW of total PJM-wide load growth through 2030, of which ~94% (30 GW) is data-center-driven; Dominion captures the largest geographic share of that growth. Realization rates from comparable markets reinforce that queues attrit: ERCOT applies a ~50% realization rate to its large-load queue, and SPP reports ~39% conversion of requests to signed service agreements. Virginia is also moving to scrutinize the forecasts directly: 2026's HB 892 directs the SCC to review how Dominion calculates load growth and to limit the ratepayer risk if it overshoots. The Data Centers page covers the broader forecasting discussion.

Supply position. Dominion's bill pressure reflects both what is getting built and when it can reliably reach DOM-zone load. On the procurement side, PJM's capacity auctions show the Dominion zone's exposure: it cleared at a steep local premium in the 2025/2026 Base Residual Auction, and although that specific zonal constraint later eased, region-wide prices have since climbed to the auction cap (detailed below).

Dominion's full announced generation program totals roughly 11.3 GW of nameplate (7.4 GW intermittent plus 3.9 GW of new gas) but only about 6.6 GW of firm, accredited capability, set against the ~9 GW of incremental data-center peak over ten years and the ~70 GW request queue. The rest of this section is that program in two halves, intermittent then firm.

Most of the pipeline is intermittent, and intermittent nameplate is not firm capacity. Applying PJM's published class accreditation ratings to the DOM-zone pipeline (ELCC, Effective Load Carrying Capability, is the share of a resource's nameplate that reliably counts toward meeting peak), roughly 7.4 GW of nameplate solar, wind, and storage accredits to only about 2.65 GW of firm capacity, a blended ~36% ELCC.

On the firm side, Dominion is advancing new dispatchable gas, but it arrives years out. The 944 MW Chesterfield Energy Reliability Center peaker was approved in 2025 (and is being challenged by environmental groups), with completion around 2029. In May 2026 Dominion announced a proposed 3 GW Cumberland Energy Center combined-cycle plant targeted for 2033 to 2034 (SCC filing expected in 2027). The 3.9 GW of new gas supplies the firm capacity the intermittent pipeline lacks, but neither plant closes the near-term gap while the queue is filling now, with CERC due around 2029 and Cumberland in 2033 to 2034. (The 7.4 GW intermittent figure counts only the solar, wind, and storage in Dominion's EIA-reported planned additions; the gas plants run through separate proceedings and are not in it.)

Dominion's own modeling isolates how much of this buildout is for data centers. In the SCC-directed sensitivity (2024 IRP Supplement, Figure 3.1), re-running the resource plan with no data-center load growth cuts planned wind by 98% (3,460 to 60 MW), eliminates all new nuclear (1,340 MW), cuts storage by 44% (4,100 to 2,250 MW), and cuts new gas by 57% (5,934 to 2,580 MW); only solar holds. That puts the data-center-driven increment at about 37% of the buildout, and an independent state model (JLARC Report 598) puts the data-center share of new generation through 2040 at 37 to 61%. Dropping data-center load also cuts the plan's net present value by about $22.4 billion (roughly 21%).

Almost none of this is replacing retiring plants. Nationally, firm capacity is retiring faster than reliable replacement comes online; Dominion is the exception, with no generation retirements planned before 2045 and none scheduled in federal data. Its buildout is additive load growth, not replacement.

Resource detail behind the accreditation figures: offshore wind carries most of the intermittent pipeline's firm total (2.64 GW nameplate at a 69% rating, about 1.82 GW firm); 3.94 GW of solar contributes about 434 MW firm (an 11% rating); 731 MW of four-hour storage adds about 366 MW. These figures separate nameplate from accredited capability and do not collapse planned, queued, and in-service resources into one realized number. Dominion's filings do not publish a complete resource-level ELCC table for each planned addition; the firm-capacity figures apply PJM's published 2026/2027 Base Residual Auction class accreditation ratings to Dominion's pipeline by resource type, the best available proxy where a Dominion-specific table is not public. Class ratings are an approximation, not a unit-by-unit ELCC study.

Four pieces of public evidence connect the grid buildout to data-center load in Dominion's service territory.

Much of the forward $7.6 billion transmission buildout is triggered by data-center load (per the IRP inventory above, including projects that serve only data centers), yet transmission cost is allocated by coincident-peak demand. Because those are different bases, residential is allocated to pay the majority share even for projects it did not trigger. Dominion's 2024 IRP testimony and JLARC Report 598 (Table 4-1) put residential at 53 to 55% of transmission-related cost allocation on that forward plan (53% under the independent consultant E3, 55% under Dominion's own rate schedules). The same pattern is already visible in current rates: of Dominion's present $1.343 billion annual transmission revenue requirement (Rider T1 plus the base transmission component, PUR-2025-00076), residential is assigned about $648 million, roughly 48% of the cost against about 35% of the energy, the highest per-kWh transmission rate of any class. The 53-to-55% is the allocation of the forward buildout; the 48% is residential's share of what is in rates today.

When Dominion proposes new infrastructure (transmission lines, substations, generation, distribution), the SCC's Class Cost-of-Service Study (CCOSS) divides the cost among customer classes. Costs are typically separated into three categories and allocated differently:

• Capacity costs (generation, transmission, distribution infrastructure): allocated by each class's contribution to the system peaks that drive the investment
• Energy and fuel costs: allocated by each class's share of total kWh consumption
• Customer service costs: allocated by customer count

The first allocation, capacity by coincident peak, is the one that matters most here. Dominion's system peaks are typically summer afternoons (driven by residential air-conditioning) and winter mornings (driven by residential heating). Residential customers' contribution to those peak hours is large; their share of the peak determines their share of the demand-driven cost. The result, from the JLARC rate-schedule analysis (Report 598, Table 4-1) cited above, is the share each class is allocated: residential is allocated 40% of generation and 53% of transmission costs; GS-4 (most data centers) 26% and 18%; the remaining commercial and industrial classes split the balance (34% of generation and 29% of transmission). The Data Centers page section on who drives the peak walks through how PJM and Dominion measure these peak contributions across customer types, with a supporting chart.

Coincidence with those peak hours differs by class. Commercial load peaks on weekday afternoons (partial overlap). Data centers run at a high, steady load factor, giving them lower coincidence with the short peak hours that drive transmission allocation, even as their constant load raises the baseline and drives the capacity buildout. A transmission upgrade triggered by a specific large load is not billed to that customer; it is recovered from all classes by peak usage. The Union of Concerned Scientists estimated that roughly $4.4 billion of 2024 PJM transmission cost driven by data centers was socialized across all ratepayers this way (the net incidence is disputed), and E3, in a 2026 review funded by the Data Center Coalition (the data-center industry's trade group) but independently authored, concedes the mechanism: connection costs “were not designed to directly attribute to the specific customers that triggered them” and so are “paid by all ratepayers.”

The 2026 General Assembly moved on this directly, but not decisively. Among roughly 15 data-center bills, HB 1393 / SB 253 (signed May 2026, effective July 2026) directs the SCC to ensure that high-load customers, including data centers, are not subsidized by other ratepayers. But the legislature accepted Governor Spanberger's amendments striking the bill's explicit cost-shift mechanism, so the enacted law directs the SCC rather than mandating that data centers bear their own capacity and distribution costs. The stronger measure, HB 503, which would have required that the cost of serving 100 MW-plus data centers be recovered only from those data centers, failed; and SB 339, which would have ordered an SCC proceeding to determine whether other customers are subsidizing data centers and to revise the allocation formula, was carried over to 2027. The statutory direction now points toward cost causation, but the binding allocation question this section describes remains open.

A capacity-recovery framework built around stable load growth, with a class-allocation structure set decades ago, is now being applied to the fastest load growth in this system's record.

On a typical Dominion residential bill in June 2026, capital and fixed charges (base rates, grid-expansion riders, clean-energy mandates) account for roughly $139/mo (79% of the bill), while fuel and energy costs account for roughly $30/mo (17%). The commercial and large-load profiles lean the same way, with fixed infrastructure outweighing the commodity, though the exact split differs by class. This is where flexibility has leverage, though only on part of the bill: cutting kilowatt-hours only touches the energy share, while cutting a customer's contribution to system peak hours reduces the demand-driven capacity Dominion has to build (generation and transmission), though not the policy-mandated clean-energy riders, which a peak reduction does not shrink.

Flexibility programs operational or filed in Dominion's territory today, with their reach across customer classes:

• Smart Thermostat Rewards (residential, active): Dominion's bring-your-own-thermostat demand-response program. Customers who enroll a compatible smart thermostat receive a $25 enrollment incentive and $25/year for participating; in exchange, Dominion can dispatch the thermostat (typically by raising the set-point a few degrees) during peak events. Dominion's own event log shows 17 events totaling 50 hours of dispatchable load reduction in 2025. (The legacy switch-based Smart Cooling Rewards program was retired after the 2022 cycling season.)
• Voluntary time-of-use rates (residential): Schedule 1G (the Off-Peak Plan) and Schedule 1EV (EV charging) are SCC-approved tariffs available now, requiring an advanced meter. Schedule 1G was approved as an experimental TOU rate in SCC Case PUR-2019-00214 (May 2020 order), launched in early 2021, and expanded from a 10,000 to 20,000 customer cap in 2023. The 2022 DNV evaluation found 9.4% peak load reduction in summer and 2.9% in winter among the roughly 7,700 retained enrollees in that year's sample. Newer enrollment figures are not publicly disclosed.
• Statutory Virtual Power Plant Pilot (all classes, ramping): Va. Code § 56-585.1:16 (the 2025 Community Energy Act) directs Dominion to operate a VPP pilot of up to 450 MW of aggregated distributed energy resources across residential and C&I customers, with aggregator participation permitted. Dominion filed the pilot tariff at the SCC in December 2025; ramp through August 2026; pilot concludes July 1, 2028. (This sits within the FERC Order 2222 framework that allows distributed resources into PJM capacity and energy markets.)
• Statutory large-load demand-flexibility filing (large C&I, pending): HB 284 / SB 371 (2026, signed) directs Dominion and Appalachian Power to file voluntary demand-flexibility programs for customers 25 MW and above by January 2027 (load-shifting, dynamic voltage, storage, energy credits), with costs not shifted to other customers; co-ops follow by 2029. It is the first statutory hook for large-load flexibility in Virginia.
• Large-customer rate structures (commercial and industrial): Schedule MBR (the Market-Based Rate Pilot that passed PJM wholesale rates through to large C&I customers) is no longer accepting new participants. The new GS-5 rate class effective January 1, 2027 introduces minimum demand commitments for customers above 25 MW. BYOG and BYOC arrangements (described in the aside above) are the national pattern for new hyperscale loads; no publicly-confirmed Dominion BYOC tariff structure exists beyond GS-5 today.
• Class-cost-of-service reform that updates the allocation factors deciding who pays for what within the existing capacity-recovery system. GS-5 is one example; deeper reforms (decoupling, performance-based ratemaking) have been adopted by other states but not yet by Virginia.

The metering is deployed; the rate design is not. Dominion's fleet is 99% advanced-metered (2024, EIA-861), so the interval data needed to price peak hours exists. But standard residential service (Schedule 1, about 99% of residential customers) is billed on a fixed customer charge plus seasonally-blocked volumetric energy charges, with no demand ($/kW) charge (the only $/kW component on Schedule 1 is the net-metering standby charge for on-site generation above 15 kW). Seasonally-blocked means the rate varies by season and usage tier, not by time of day, so it carries no peak-hour price signal. Time-of-use is voluntary opt-in only and, as the program list above shows, reaches under 1% of residential customers, leaving the class no individual price signal to shift the peak that drives its allocated cost. Net metering exists (NEM 2.0 was approved in May 2026, though critics say it resembles NEM 1.0), but net metering compensates solar export; it is not demand flexibility.

The capex bias is a structural headwind for flexibility. A regulated utility earns a guaranteed return on capital investments but typically does not earn a return on operational solutions like demand flexibility. This is the Averch-Johnson effect documented in regulatory economics since the 1960s: utilities have an incentive to prefer building infrastructure over operating it differently, even when flexibility would be cheaper for customers. Dominion compounds this: it is vertically integrated, owning generation, transmission, and distribution and earning its allowed return (an ROE of 9.8% as approved in the 2025 biennial review, Case No. PUR-2025-00058, up from the prior-period 9.7%) on the capital base of all three, so each megawatt of flexibility that avoids a build shrinks the rate base it earns on. The regulatory reforms that address this directly are decoupling (separating utility revenue from sales volume), performance-based ratemaking (paying utilities for outcomes rather than for capital deployed), and shared-savings mechanisms (utilities keep a portion of the savings flexibility programs deliver). Several states have implemented forms of these reforms; Virginia has adopted neither comprehensive performance-based ratemaking nor revenue decoupling.

A new supply-side mandate, recovered the same way. In 2026 Virginia enacted HB 895 / SB 448 (effective July 1, 2026), directing Dominion to build roughly 16 GW of short-duration and 3.5 GW of long-duration energy storage by 2045. Storage is a flexibility-enabling resource, but the statute procures it as utility-owned capital recovered through a non-bypassable charge on all retail customers, the same socialization mechanism described in Who Pays and Why. And under PJM's marginal ELCC, four-hour storage is accredited well below its nameplate, an accreditation that falls further as more storage is added (a capacity-market construct, not a limit on what a battery can physically deliver), so the nameplate target far exceeds the firm capacity it earns.

Could flexibility fill the gap? Capacity is procured for rare hours, not for the year: Dominion's entire 2025 thermostat season, noted above, came to 50 dispatched hours, and those are the hours the buildout is sized to serve. At pilot scale, the 450 MW VPP cap, if its dispatch proves out, is roughly half the size of the 944 MW Chesterfield peaker, the only new firm generation approved in this study's window. On the data-center side, the Data Centers page covers the evidence: the Duke Nicholas Institute estimates that accepting roughly 0.5% average annual curtailment (about 177 hours per year) would let existing US grids absorb most projected data-center growth without new capacity, and HB 284 now gives Virginia a statutory vehicle for those programs. None of this eliminates building under 5.4%-per-year peak growth. Flexibility does shape how much gets built and when, and whether customers who shift their use are paid for it.

Set the opportunity against what is deployed. The zone's roughly 25 GW peak sits about 9 to 10 GW above its ~15 GW average load, capacity that exists to serve the top hours of the year, with the reserve margin held on top of that (Grid Flexibility, from the peak and average-load figures above). Flexibility competes for the top of that gap. What is operational or planned on the demand side today reaches a small fraction of it: the 450 MW VPP pilot cap is under 2% of the 25 GW peak and about 5% of the peak-to-average gap, and Smart Thermostat Rewards dispatched 50 hours in 2025. The per-customer effect is measured but barely scaled: Dominion's own evaluation found a 9.4% summer peak reduction, among only the roughly 7,700 time-of-use enrollees, while the peak-blind default rate still covers about 99% of residential customers, the class the peak is built for.

The grid Dominion is operating today is more constrained than it was four years ago. PJM's own forecast captures the shift: the 10-year summer peak growth rate for the Dominion zone rose from 2.2% in PJM's 2022 Load Forecast Report to 5.4% in the 2026 report (winter growth rose similarly, from 2.6% to 5.1%). The planning horizon for the projects that would relieve those constraints stretches into the 2030s, and demand has been growing faster than the planning cycle. Among the available tools, demand flexibility operates on both timescales: it can be deployed in months rather than years (immediate years, shaving the peak that drives the buildout now, and it builds toward a grid where customers who move their consumption are paid for it.

The grid must be sized for its coincident peak, and that peak is residential-driven. Dominion's own filings put the data-center load factor at about 92 to 93% (high and steady, not perfectly flat), while non-data-center load runs at 56-58% and the residential class, the peakiest, at about 36 to 41%. As data centers grow they pull the overall system load factor up, from 63% in 2024 toward 72% by 2039, but they add baseload, not peak: the sharp peaks that size the local grid still come from the low-load-factor residential class. That makes households the class with the most peak to shave, even as the system as a whole looks flatter. On average the grid uses only a little over half the capacity it must keep available once the required reserve margin is counted.

Load factors are from Dominion's response to PEC Interrogatory No. 01-16 (PUR-2025-00058): the system load factor rises from 63% in 2024 toward 72% by 2039 with data-center growth, data-center load runs about 92-93% (projected; measured operating values are nearer 79-82%), and non-data-center load 56-58%. The residential class load factor (about 36% in the 2023 biennial review, 40.6% in the 2025 review) is from Dominion class cost-of-service exhibits (SCC Staff witness Glenn Watkins' Schedule GAW-2, PUR-2023-00101; Company witness Robert Miller's Exhibit 25, PUR-2025-00058). The capacity-utilization figure is a Grid Flexibility calculation from DOM-zone load and PJM's reserve requirement.

The next 12-18 months will determine whether current bill pressure stabilizes or intensifies. The milestones, in chronological order:

Every Virginia SCC proceeding cited in this case study, what it decides, and where it stands. This tracks only the dockets that appear in the analysis above, not all Dominion proceedings; each docket number links to the SCC's official case record.

Sources

Every dollar figure in this case study was reconstructed from Virginia State Corporation Commission tariff filings, with additional context from PJM Interconnection reports, Federal Energy Regulatory Commission orders, the Joint Legislative Audit and Review Commission's 2024 data-center study, and Piedmont Environmental Council expert testimony. Each Dominion rate component (base rates, the fuel factor, and the 15 active rate adjustment clauses) was traced to a specific SCC docket: for every rider, the approved cents-per-kilowatt-hour value, effective period, legal authority, and underlying revenue requirement come from the filing record. Original PDF filings are preserved with cryptographic checksums so that any future change to a primary source can be detected.

Calculation

Bills were reconstructed for three representative customer profiles month by month from June 2022 through June 2026: residential at 1,000 kWh per month (Schedule 1); mid-commercial at 30 MWh per month with a 75 kW peak (Schedule GS-2, mid-market); large-load at 8 GWh per month with a 15 MW peak (Schedule GS-4, transmission voltage). Each monthly bill sums every approved per-kWh tariff component (base generation, transmission, distribution, customer charge, fuel factor, and all active rider charges). State and local taxes and fee assessments are an unmodeled line item, which produces a small gap between the reconstructed tariff stack and the customer's total bill. The headline class-growth figures (residential +38%, mid-commercial +36%, large-load +34%) and the 17.6 versus 8.5 cents-per-kWh comparison are reported on this complete-bill basis (base, fuel, and all riders), not a single rider or the narrower tariff-stack subset. The window starts in June 2022, just after the one-time 2022 commodity (Russia-Ukraine) fuel spike, to isolate the buildout-era trajectory from that spike.

One scope note. The “clean energy mandates” bucket appears small at the start of the window because most of the explicit clean-energy riders were authorized under Virginia's 2020 Clean Economy Act and ramped onto bills beginning in late 2021. Pre-VCEA renewable compliance costs sat inside base generation rates and were progressively reclassified into the explicit clean-energy bucket as the new riders came online. The chart reflects what appeared on a customer's bill statement; the magnitude of post-2022 growth still primarily reflects real new buildout costs (offshore wind, utility-scale solar, small-modular-reactor pre-development).

Validation

The reconstruction was cross-validated against two independent anchors: (1) EIA Form 861 annual realized revenue per customer class for Dominion, 2018-2023; and (2) the SCC Voluntary Electric Utility Regulation annual typical-bill reports. Both anchors agreed with the tariff-stack reconstruction within the modeled scope. The 2023 SCC typical-bill figure was $129.01/mo for residential 1,000 kWh (a 2023 anchor, not a current bill); the tariff-stack reconstruction (the narrower subset, excluding taxes and fees) averaged $117.86/mo over that same 2023 window, with the difference consistent with the unmodeled taxes and fee assessments noted above. The current complete-bill headline ($175.98/mo in June 2026, up from about $127/mo in June 2022) sits above the 2023 anchors because of post-2023 rider growth. The full source-document inventory, extraction code, per-month component dataset, and chart-rendering scripts are available on request pending publication of the project repository.