The Data Center Question Idaho Hasn’t Settled: Who Pays for the Power?

Start with the bill people can feel.

Idaho Power’s 2025 general rate case produced new Idaho rates effective Jan. 1, 2026. The approved settlement increased retail revenue by $110 million annually, an average increase of 7.48 percent, according to Idaho Public Utilities Commission records and Idaho Power’s customer summary.

For an average residential customer using 900 kilowatt-hours per month, Idaho Power estimated the increase at about $12.13 per month. Residential rates rose 9.74 percent. Small general service customers rose 9.78 percent. Irrigation rose 8.49 percent.

That does not mean data centers caused this specific rate hike. PUC staff said the increase in the 2025 rate case was driven by baseline load growth and inflationary pressure, not new large-load growth.

But the timing matters. Ordinary customers are already seeing higher bills while the state is still deciding how future large-load costs should be separated, assigned, or blended.

Idaho Power is planning for a much bigger system.

Idaho Power’s 2025 Integrated Resource Plan says the company expects “unprecedented demand growth.” Its peak load is expected to grow by about 1,700 megawatts over the 20-year forecast period, with nearly 1,000 megawatts of that growth arriving in the next five years.

For a household, a megawatt is an abstract unit. So put it this way: using the U.S. Energy Information Administration’s average household electricity figure, a 100-megawatt facility running flat all year can use roughly as much electricity as more than 80,000 average U.S. homes buy in a year.

Not every large Idaho project is a 100-megawatt data center. Not every megawatt in Idaho Power’s forecast comes from tech. Population growth, heat waves, irrigation, industry, electrification, and ordinary business growth all matter.

Still, Idaho Power’s planning documents single out a category called “additional firm load.” That category includes large customers such as Brisbie/Meta, Idaho National Laboratory, Lamb Weston, Micron, Simplot, and other committed large-load customers. Idaho Power projects that category rising from 257 average megawatts in 2026 to 875 average megawatts in 2030.

That is the heart of the issue. A handful of very large customers can add load faster than normal homes and shops ever could.

The state’s biggest tech wins are also power events.

Micron’s Boise expansion is a semiconductor story, a jobs story, and a national supply-chain story. It is also an electric-system story. Idaho Power filed a special-contract application for Micron Idaho Semiconductor Manufacturing, a Micron subsidiary, in December 2024. The PUC approved the contract, with modifications, on May 8, 2026.

The contract applies when Micron’s aggregate power requirement exceeds 20,000 kilowatts, or June 1, 2026, whichever comes first. The Commission’s order approved specific demand and energy charges and told Idaho Power to track incremental costs tied to serving Micron and future large loads.

Meta’s Kuna data center is already part of the same story. The Idaho Department of Commerce described the project as an $800 million, 960,000-square-foot data center expected to support about 100 operational jobs and more than 1,200 jobs at peak construction. Idaho Power’s 2025 planning materials say Meta is supporting 100 percent of operations through new renewable resources connected to Idaho Power’s system.

That sounds reassuring, but it does not close the whole question. A renewable-energy arrangement can cover energy over time. It does not automatically answer every question about capacity at peak hours, backup supply, balancing, transmission, substations, delay risk, or what happens if a customer uses less power than expected after the utility has built around it.

“Paying for power” is not one thing.

When people hear that a big customer has a special contract, it is easy to assume the problem is solved. Sometimes it may be. But the phrase “paying for power” hides several different costs.

There is the electricity itself. There is the local equipment needed to connect a facility. There are transmission lines that move power from where it is generated to where it is used. There is generation capacity, which means having enough supply available during the hottest or tightest hours of the year. There is reserve capacity for reliability. There is the risk that a customer asks for a huge amount of capacity, then ramps up slowly, downsizes, delays, or leaves.

That last part is where ordinary customers can get exposed. A utility can build or contract for assets because a large customer says it needs power. If the customer later does not use enough power to cover the costs, the wires and plants do not disappear. Someone still has to pay for them.

This is why regulators in other states are moving toward minimum bills, take-or-pay contracts, collateral, exit fees, and separate rate classes for large loads. The basic idea is not complicated: if one customer causes the system to be built bigger, that customer should carry the risk created by that decision.

Idaho lawmakers tried to draw a harder line.

In 2025, House Bill 395 tried to make that principle explicit. As amended, the bill defined a new large load as 30 megawatts or more added in any 12-month period. It would have barred a public utility from serving that load from existing resources, or from resources already planned for general ratepayers within 10 years, unless the large customer paid the costs.

The bill named the categories plainly: cost of capital, generation, transmission, distribution, and customer-related costs. It passed the Idaho House 62-8 on March 26, 2025, then went to Senate State Affairs. The bill does not appear to have become law.

That makes HB 395 important even in failure. It shows that a broad House majority saw the cost-shift risk as real enough to legislate. But because the bill stalled, the live rules are still being built through PUC cases, special contracts, tariffs, and future rate proceedings.

The PUC is now working through the machinery.

On March 31, 2026, Idaho Power opened a PUC docket asking regulators to evaluate class cost-of-service methodology and cost-of-service considerations for new large-load customers. Intervenors include the City of Boise, Idaho Irrigation Pumpers Association, Industrial Customers of Idaho Power, Clean Energy Opportunities for Idaho, Bayer, and federal executive agencies.

That sounds dry because it is dry. But this is where the money gets assigned.

A cost-of-service case decides how the utility’s costs are split among customer classes. Residential customers, small businesses, irrigators, industrial customers, and special-contract customers do not all use the grid the same way. If the formula is wrong, one class can quietly subsidize another.

In the Micron order, the PUC said protecting customers from cost shifting due to new large-load entrants is one of the key parts of fair, just, and reasonable rates. It also said Idaho Power faces near-term capacity deficits requiring large investments, and that strong protections are required.

That is a careful regulator’s way of saying: this could get expensive if the accounting is wrong.

Other states show what Idaho is trying to avoid.

Virginia is the warning label. Northern Virginia became the center of the data-center world, and state analysts found that unconstrained power demand could double within 10 years, with data centers as the main driver. Virginia’s Joint Legislative Audit and Review Commission estimated a typical Dominion residential customer could see generation and transmission costs rise by $14 to $37 per month by 2040, in real dollars, because of statewide demand growth.

Ohio offers another example. AEP Ohio’s data-center tariff applies to new data centers or expansions of at least 25 megawatts. The tariff generally requires large data centers to pay a minimum tied to 85 percent of subscribed capacity for up to 12 years, whether or not they actually use all of it.

Georgia regulators created a rule for new customers above 100 megawatts, allowing Georgia Power to use terms beyond standard tariffs and charge large customers for site-specific and upstream generation, transmission, and distribution costs as construction proceeds.

These examples do not map perfectly onto Idaho. Idaho Power is not Dominion. Boise is not Northern Virginia. But the mechanism is the same: large, fast, steady loads can force utilities to build sooner and bigger than they otherwise would.

Idaho should want growth without pretending it is free.

There is a lazy version of this debate where one side says “tech jobs” and the other side says “rate hikes,” and nobody has to do the harder accounting.

Idaho should want serious industry. Micron matters. INL matters. Data centers can bring tax base, construction work, and long-term infrastructure. Cheap, reliable power is one of Idaho’s advantages, and using that advantage to attract real projects is not foolish.

But cheap power is only an advantage if the public can trust the bill. If a company is large enough to require custom planning, it is large enough to carry custom obligations. If a project needs new power plants, new lines, new substations, or reserve capacity, the public deserves to know which costs are direct-assigned, which are shared, which are confidential, and what protection exists if the project changes.

The phrase to watch is not “clean energy” or “economic development.” The phrase to watch is “hold other customers harmless.” It sounds simple. In practice, it has to survive rate cases, construction delays, forecast misses, confidential contracts, and the next hot summer evening when everyone needs power at once.

What ordinary Idaho customers should watch next

• The large-load cost-of-service docket: this is where Idaho Power and the PUC work out how future large-load costs should be assigned.
• Special contracts: Micron, Meta/Brisbie, and any future 30-megawatt-plus customer should be watched for demand charges, minimum bills, exit provisions, and direct-assignment language.
• Transmission milestones: Boardman-to-Hemingway and SWIP-N matter because delays can tighten summer capacity and raise market exposure.
• New gas and resource approvals: Idaho Power has asked regulators to approve new natural-gas plants for 2029 and 2030 capacity needs.
• Future rate cases: watch whether “growth” costs are tied to ordinary population growth, large-load growth, or both.
• Tax incentives: if Idaho grants incentives for data centers, energy and water conditions should be enforceable, not decorative.

The data-center question is not whether Idaho should join the technology economy. It already has. The question is whether Idaho can grow without asking households, irrigators, and small shops to silently insure the largest electric bets in the state.