TL;DR
- 63.3% Virginia residential-to-industrial spread in December 2025, the narrowest of the three geographies we measured (EIA data through December 2025).
- 137.9% Texas residential-to-industrial spread in December 2025, the widest of the three.
- 9.2% Virginia commercial-to-industrial spread in December 2025, the narrowest cross-class gap anywhere in the set.
- 41.5% Henry Hub natural gas year-over-year move as of December 2025, carried as a fuel-cost control.
There is a tidy story making the rounds: data centers pull cheap industrial power while households pay for the grid. We wanted to know whether the retail price data actually looks like that. It does not, at least not in the way the story assumes.
Does the densest data-center market really have the biggest rate gap?
No. Virginia, the densest data-center market in the world, has the narrowest residential-to-industrial spread of the three geographies at 63.3%. Texas sits at 137.9% and the national figure is 102.1%. If the cheap-industrial-power framing were driving rate structure, you would expect the data-center capital to show the widest gap. It shows the smallest.
The state with the most data centers has the narrowest rate-class spread. That is the opposite of what the popular framing predicts.
Our read: this is descriptive, and we are careful to keep it that way. Nothing in these numbers isolates a data-center effect. But it is a useful reality check on a claim that gets repeated as if it were settled.
How do Virginia industrial electricity rates compare inside the state?
Inside Virginia, the classes have nearly converged. Commercial power ran 10.21 cents per kWh against 9.35 cents for industrial, a spread of just 9.2%. That is the narrowest cross-class gap we measured in any geography. Virginia industrial also runs above the national industrial rate of 8.53 cents per kWh, so this is not a story of unusually cheap large-load power.
You can pull the underlying industrial series straight from the U.S. Energy Information Administration if you want to see it yourself. The rate classes in Virginia are stacked close together, which is not what you would sketch if you assumed households were quietly subsidizing server farms.
Which class moved fastest over the year?
Virginia commercial did. It rose to 10.21 cents per kWh, up 15.2% from a year earlier, the fastest year-over-year class move in the whole dataset. Virginia industrial rose 5.8% over the same window. We note the divergence and stop there, because attributing a single-year move to any one cause would be exactly the overreach this whole exercise is meant to avoid.
Retail price increases reflect a stack of factors: general inflation, fuel costs, utility capital cycles, and potentially incremental load. Pulling one thread out and calling it the cause is the mistake. So we describe, we do not diagnose.
How do we keep fuel moves from masquerading as allocation shifts?
We carry Henry Hub natural gas as a control. As of December 2025 it sat at $4.26 per MMBtu, up 41.5% year over year. The logic is simple: a move that lifts every rate class together is a fuel story, while a change in the relative position between classes is an allocation story. This report is only about the second kind. When gas swings 41.5% in a year, it will push residential, commercial and industrial rates in the same direction, and that shared push is not what we are trying to read.
That distinction matters because it is easy to see a rising industrial rate and reach for a data-center narrative when the honest answer is that the fuel input moved for everyone.
What about the national baseline?
Nationally the classic pattern holds. Commercial power costs 59.8% more than industrial power, at 13.63 cents per kWh against 8.53 cents. Residential runs 102.1% above industrial. Industrial customers pay less per kWh than households and businesses at the national level, which is the ordinary shape of utility rate design and predates the current AI data center electricity conversation by decades.
What is interesting is not that industrial power is cheaper nationally. That has always been true. What is interesting is that in Virginia, the place you would most expect the gap to blow out, the classes have moved closer together, tracking with a convergence rather than a widening.
So what should you actually take from this?
Our read, clearly labeled as opinion: the AI data center electricity debate deserves better than a slogan. The rate-class spread data is consistent with a story where Virginia’s classes are converging, not diverging, and that is worth sitting with before anyone builds policy on the cheap-industrial-power assumption. We are describing the shape of the numbers, not assigning blame, and the shape does not match the headline.
We also want to be honest about the limits. This compares one regulated market, Virginia, against one deregulated retail market, Texas. Their rate designs differ for reasons that have nothing to do with data centers. That caveat is load-bearing, not decorative.
EIA data through December 2025.
The full report walks through all three geographies, quantifies each spread, and shows the Henry Hub control in context. Read the electricity rate-class spread report.