HN Debrief

County with 37 Data Centers Asks Schools to 'Conserve Electricity'

  • Infrastructure
  • Energy
  • AI
  • Regulation
  • Economics

The article reports that Henrico County, near Richmond, sent an email asking schools and other county facilities to conserve electricity after power costs rose, and frames that request against a local boom of 37 data centers with more on the way. The obvious read is that residents and public services are being told to tighten belts so AI and cloud operators can keep expanding.

If you build or buy large power-intensive infrastructure, assume the fight will center on who pays for generation and transmission upgrades, not just on total demand. Watch for utilities and regulators to ringfence data center costs with special rate classes, minimum contracts, curtailment rules, or mandatory behind-the-meter power before public backlash hardens into outright moratoria.

Discussion mood

Mostly hostile to the article’s framing and to the current policy setup. People disliked the optics of schools being asked to conserve while data centers expand, but the dominant mood was frustration that the piece implied a simple causation story without proving it and ignored the bigger mix of PJM market design, transmission constraints, and Virginia energy policy.

Key insights

  1. 01

    PJM market rules distort the picture

    PJM’s forward capacity auction means rate spikes can come from a bad three-year forecast and a flawed market design long before a transformer melts in your neighborhood. That changes the diagnosis from "data centers raised prices" to "large new loads are colliding with a pricing system that is already brittle," which is a much more useful place to intervene.

    If your company depends on grid power in PJM territory, track capacity-market rules as closely as local load growth. Regulatory design can move your power costs faster than your own consumption does.

  2. 02

    Transmission is the hidden cost center

    Even when a region has enough generation on paper, moving multiple gigawatts to a dense data center cluster can require multibillion-dollar transmission projects like long-distance high-voltage direct current lines. The hard part is not just making electricity. It is deciding whether everyone else should bankroll the wires for a few hyperscale customers.

    Ask where the electrons will travel, not just where they will be generated. In siting and infrastructure planning, transmission exposure is now a first-order business risk.

      Attribution:
    • FireBeyond #1
    • nonplus #1
    • Havoc #1
  3. 03

    Virginia rate hikes predate a simple AI story

    The Lawrence Berkeley National Laboratory report cited here says Virginia’s recent price increases were driven largely by the Virginia Clean Economy Act transition and higher natural gas prices, with load growth offsetting some fixed costs rather than dominating them. That undercuts any claim that one county’s data centers neatly explain a statewide rate jump.

    Be careful using statewide bill increases as proof of a local demand story. Separate policy-driven utility spending from load-driven network expansion before making strategy or policy bets.

      Attribution:
    • jonas21 #1 #2
  4. 04

    Flexible load matters more than raw megawatts

    Industrial users like foundries often sign contracts that let utilities dial usage up or down in response to grid conditions. Commenters working around large loads said data centers may do some of this, often with behind-the-meter generation, but AI-heavy facilities can also behave more like steady baseload than an interruptible customer. That difference determines whether a big user helps stabilize the grid or just parks a giant constant draw on it.

    When evaluating a power-hungry facility, ask about curtailment, demand response, and on-site generation before asking about average consumption. Grid friendliness now depends on controllability as much as scale.

      Attribution:
    • LeifCarrotson #1
    • infecto #1
    • esseph #1
  5. 05

    Some cost-shifting fixes are already starting

    Virginia has already moved toward a dedicated data center rate class with long-term minimum contracts, which is exactly the kind of ringfencing many people say should have existed from the start. That signals the politics are shifting from broad outrage to technical tariff design, which is where the real allocation of costs gets decided.

    Watch utility commissions, not just legislatures. Special tariffs and minimum-load contracts are where communities will try to claw back leverage over hyperscale users.

      Attribution:
    • jeffbee #1
    • nonethewiser #1

Against the grain

  1. 01

    Data centers could help absorb renewable oversupply

    A few commenters pushed the opposite frame and said large compute loads may eventually complement a renewable-heavy grid by soaking up excess solar and providing a controllable demand sink. That only works if operators are willing to modulate usage, but it points to a future where compute is treated less like a parasite and more like dispatchable demand.

    Do not assume all large compute load is structurally bad for the grid. There may be real value in pairing flexible workloads with oversupplied renewable regions.

      Attribution:
    • legitster #1
  2. 02

    Load growth can lower average rates

    One useful objection was that new demand can spread fixed utility costs across a larger base, so banning data centers does not automatically protect households from higher bills. In some systems, the extra customer can improve economics even while making peak-capacity planning harder.

    Avoid simplistic anti-growth assumptions in utility policy. The question is who bears marginal upgrade costs, not whether any new load is inherently harmful.

      Attribution:
    • Kamq #1
  3. 03

    The headline oversold the schools angle

    Several people noted that the conservation email went to county government facilities generally, with schools highlighted because that made a stronger headline. That does not change the underlying power issue, but it does make the article feel more like framing than reporting.

    Treat stories about infrastructure stress as much for their narrative packaging as for their facts. The most emotional example in a headline is often not the cleanest description of the policy.

      Attribution:
    • jabroni_salad #1
    • skywhopper #1

In plain english

baseload
A steady, continuous level of electricity demand that does not vary much over time.
behind-the-meter generation
Power generation located on a customer’s site and used directly there rather than fully relying on the shared grid.
Dominion Energy
The main electric utility serving much of Virginia.
forward capacity auction
A market where power suppliers are paid in advance to guarantee they will have enough electricity available during future peak demand periods.
gigawatt
A unit of power equal to one billion watts, often used to describe the scale of power plants or very large electricity loads.
hyperscale
Very large cloud or data center operations run at massive scale by major technology companies.
Lawrence Berkeley National Laboratory
A United States national research lab that publishes technical and policy analysis, including studies on electricity markets and rates.
PJM
PJM Interconnection, the regional grid operator and electricity market that coordinates power across a large part of the eastern United States.
tariff
The formal rate schedule and rules a utility uses to charge customers.
Virginia Clean Economy Act
A 2020 Virginia law that requires the state’s major utility to move toward carbon-free electricity on a set timeline.

Reference links

Energy pricing and policy analysis

Grid cost allocation and market structure

Utility tariffs and rates

Data center energy sourcing and local impacts

Political and media context