Bill Would Put More Energy Costs on Data Centers — Virginia Mercury - 2026-02-10

Original source

Summary

The single cleanest regulatory number in the wiki for the residential cross-subsidy question. Virginia SB 253 (Sen. Louise Lucas, D-Portsmouth) was amended on Feb 9, 2026 to let Virginia’s State Corporation Commission (SCC) decide whether large-load customers — rather than residential ones — should cover the cost of distributing power to data centers and the cost of Dominion Energy’s capacity-auction purchases (the out-of-state power Dominion buys to meet data-center-driven demand). If the SCC approves, those costs shift onto Dominion’s GS5 rate class (mostly data centers) for new and existing data centers through 2033. The SCC’s own letter projects the shift cuts a typical residential bill by 3.4% (~$5.52/month), raises the data-center rate by ~15.8%, and saves local governments ~$8.3 million in 2027. Dominion fully supports the bill — the rare case where the incumbent utility and the cost-shift-onto-its-biggest-customers align, because soaring PJM Interconnection capacity prices are themselves data-center-driven and Dominion would rather have a legislated allocation than a ratepayer revolt.

This is the retail/state-legislative face of AI Cost Incidence: where the PJM Interconnection Q1 2026 data quantifies the cost at the wholesale-capacity layer and the Michigan/DTE Energy sources show the proportional-socialization mechanism, SB 253 is a named statutory attempt to reverse the incidence — to move the AI buildout’s grid cost off households and onto the load that caused it. It is a concrete instance of Ratepayer Protection legislation with a regulator-scored dollar figure attached.

Key Points

• The mechanism. SB 253 (amended) authorizes the SCC to determine whether it is “in the public’s interest” for large-load customers to cover (a) the cost of distributing power to data centers and (b) Dominion’s capacity-auction costs — and to assign those costs to the GS5 rate class instead of spreading them across all classes.
• The scored numbers (SCC letter). Typical residential rate −3.4%, ≈ −$5.52/month; data-center (GS5) rate +15.8%; ~$8.3M saved for local governments in 2027. Shift is time-limited (“the next few years”) because capacity-auction prices “continue to break records.”
• Reaches existing data centers, not just new ones — through 2033. This is the distinguishing feature: most ratepayer-protection mechanisms only bind new load; SB 253 can re-rate the installed base.
• GS5 rate class lineage. Created in Dominion’s last rate case: 14-year contracts for customers >25 MW, with minimum demand charges of 85% transmission / 85% distribution / 60% generation. SB 253 builds on this class rather than inventing a new one.
• Dominion supports (lawyer Joe Reid): connecting data centers costs “hundreds of millions of dollars”; the shift shrinks every other class’s “share of the pie” and delivers “immediate rate relief.”
• Industry pushback (Data Center Coalition, Nicole Riley): says the industry “is committed to paying for the power it uses” but wants the SCC required to quantify how much of the capacity cost is actually attributable to data centers — arguing the drafted language “precludes any analysis of that.”
• Opt-out asymmetry. High-load industrial/manufacturing customers get a one-time chance to exit GS5; data centers cannot opt out. The bill targets the load class by design.
• Origin story. SB 253 began as a low-income weatherization program extension (through 2038) before the data-center cost-shift amendment redefined it — a clean example of a vehicle bill repurposed mid-session.
• Status (as of source date): advanced out of Senate Labor and Commerce; headed to Senate Finance, which Lucas chairs.

Newsletter Angles

• The incidence-reversal story. Almost every AI-cost-incidence source documents cost flowing toward households (PJM capacity socialization, DTE proportional allocation). SB 253 is the first wiki source showing a legislated attempt to flip the arrow — and a regulator put a number on it ($5.52/mo). That contrast is the piece: “we can measure exactly what de-socializing the AI grid bill is worth, because Virginia scored it.”
• The utility-supports-it twist. Dominion backing a bill that raises its largest customers’ rates looks paradoxical until you see the mechanism: Dominion is indifferent to who pays as long as the capacity bill gets paid, and it would rather legislators assign the burden than face residential backlash. Names the real Dominion incentive (revenue certainty), not “good corporate citizen.”
• Existing-vs-new is the whole ballgame. The 2033 reach over existing data centers is what makes this different from tax-abatement-era ratepayer protections. Worth contrasting with Michigan’s new-load-only tariffs.
• Ohio/Issue #7 tie-in. Operation Metro Surge runs through the same cost-incidence logic; Virginia is the precedent state (data-center alley) for what Ohio is now arguing about.

Entities Mentioned

• Virginia SB 253 — the bill itself (subject of the source); SCC cost-shift mechanism, GS5 targeting, 2033 reach.
• Louise Lucas — D-Portsmouth, sponsor; chairs Senate Finance (the bill’s next stop).
• Dominion Energy — Virginia’s largest utility; created the GS5 class; fully supports SB 253.
• State Corporation Commission — Virginia regulator; scored the $5.52/15.8%/$8.3M figures; would hold approval authority.
• Data Center Coalition — industry trade group; conditional opposition (wants causation analysis required).
• PJM Interconnection — the capacity market whose record-breaking auction prices are the underlying cost driver.
• Southern Environmental Law Center — called Dominion’s prior rate case “a good first step”; per editor’s note, takes no position on SB 253 itself.

Concepts Mentioned

• AI Cost Incidence — SB 253 is the statutory-reversal instance; retail/legislative layer.
• Ratepayer Protection — the policy category; SB 253 is a scored example reaching existing load.

Quotes

“There are more than 200 energy bills this session, and as far as I’m aware, this is the only proposal to actually reduce rates in the near term.” — Sen. Louise Lucas

“They’re going to pay a little bit more for the next few years, so that the rest of the customers’ share of the pie will get smaller and there will be immediate rate relief for the residential customers…” — Joe Reid, lawyer for Dominion

“We … request clarification that the SCC should investigate as part of the proceeding how much the capacity costs are due to data centers. We are concerned that the language as drafted precludes any analysis of that.” — Nicole Riley, Data Center Coalition

Notes

• Capture method: Claude-in-Chrome (Brave) on 2026-06-06. WebFetch returned HTTP 403 (Cloudflare bot-block, not a paywall) — hence the required raw/ clip per the schema’s ingest_method: chrome rule. Virginia Mercury is States Newsroom, CC BY-NC-ND 4.0.
• Resolves the 2026-06-06 TO INGEST flag logged earlier the same day (surfaced while verifying the $5.52 figure cited in the Andy Masley dialogue).
• Figures are SCC projections, not realized. The bill grants the SCC authority to make the shift; it does not itself mandate the numbers. Track whether SB 253 passed the 2026 session and whether the SCC acted — a clear follow-up for an incidence piece. As of the Feb 10 source it had only cleared one committee.
• Sponsor-power detail worth keeping: Lucas chairs Senate Finance, the committee the bill heads to next — relevant to its odds.
• Possible contradiction to watch: the Data Center Coalition’s “committed to paying for the power it uses” framing vs. the SCC’s finding that residential customers are currently over-absorbing distribution/capacity cost. Not a hard ⚠️ contradiction yet — they’re arguing about causation attribution, not the direction of the subsidy.