Electricity bills are up, and data centers are getting the blame. But in the data so far, the states with the fastest-growing power use have mostly seen rates rise slower than elsewhere, not faster. Rising demand is supposed to push rates up, not down, but there are good reasons to think rising demand has held electric rates down in some areas. The question worth asking is whether we can count on that to last.

Regulated electric rates are essentially one big division problem. Regulators put an electric utility’s annual cost of providing service in the numerator of a fraction and the total expected sales for the year in the denominator, then do the arithmetic. When a data center joins the system, it can boost the denominator quite a bit. Whether rates rise or fall depends on whether total costs grow faster or slower than consumption in the denominator.

In other words, when the costs of serving that additional load are below average, adding a data center can bring down the costs for other customers. Naturally, developers of data centers sought out places to build where they could get low-cost power supplies. The result was that the areas adding the most demand were the ones that saw rates fall.

Unfortunately, those good times may be coming to an end.

In recent years, data centers could locate in rural areas of North Dakota, New Mexico, and Nebraska and get cheap land and cheap energy. All three states have seen electric rates stay stable or decline over the last six years (once adjusted for inflation) as consumption of electricity shot up. But the spare capacity on the electric grid is getting filled up, meaning newer data centers can no longer be served at below-average costs.

The cost of the materials used to build the grid, like wires, transformers, and switchgear, is rising fast too. That increases costs for electric consumers even in parts of the country that haven’t seen much data center construction. The arithmetic has not changed: growing demand can still lower rates when new customers use spare capacity. What has changed is the underlying condition. In many places, the spare capacity is gone.

PJM—the regional grid for the mid-Atlantic states stretching as far west as Chicago—shows the other side. Electricity demand is growing there, too, but with spare capacity gone, new supply and grid upgrades must be built to serve it. Now getting one more megawatt of power delivered costs more than the average, not less. Growing demand pushes the numerator up faster than the denominator. Rates have followed. Data centers have contributed to those increases, but they aren’t the whole story. It’s the same division problem that North Dakota, New Mexico, and Nebraska faced, only with the spare capacity already used up.

But all is not lost for the residential consumers who want to see their electric rates kept under control. Growing demand can still leave rates stable for other customers, but only if the new large commercial and industrial power consumers cover all of their costs. Remember that division problem. If new large customers pay their own way, they add to total sales without adding to costs that land on everyone else.

One increasingly common idea is part of the White House’s ratepayer protection pledge: make large new data centers build their own power plants or buy power from new generators rather than leaning on the existing system. That helps, but it only covers part of the problem. Even a data center that generates all its own power will still rely on the poles, wires, and transformers that keep it connected to the grid, and these have been the costs that are growing fastest. Special utility rates for the largest new industrial consumers can cover all new costs, but only if they’re designed carefully.

Growing demand can still help hold rates down. But only if big customers like data centers, which drive the growth, pay for both the power and the wires they’re using.

Low-Energy Fridays: Every Friday we take a complicated energy policy idea and bring it to the 101 level.