Can America's Grid Handle the Data Center Boom?

For most of the last two decades, American electricity demand went essentially nowhere. Efficiency gains in lighting, appliances and industrial equipment cancelled out the growth from population and new construction, and utilities planned around a grid that was not going to need to get much bigger.

That period is over.

Federal forecasters now expect record electricity consumption in consecutive years, the Department of Energy says the country has a pressing need for additional transmission capacity, and a new category of customer is showing up at utilities asking for more power than a mid sized city.

The question is no longer whether the United States can generate enough electricity. It is whether the infrastructure to move that electricity can be built fast enough to keep up with where the demand is landing.

The numbers

According to the U.S. Energy Information Administration, electricity demand grew by an average of roughly 0.1% per year between 2005 and 2019. Between 2020 and 2025 that rate jumped to about 1.7% annually.

The agency's September 2026 Short-Term Energy Outlook, released on September 9, projects U.S. electricity sales of 4,135 billion kilowatt-hours in 2026 and 4,211 billion kilowatt-hours in 2027. The 2026 figure represents an increase of almost 2% over 2025 and would be a record.

EIA attributes the growth primarily to data center development and increased manufacturing activity. Commercial sector electricity sales alone are forecast to rise 3.3% in 2026 and 2.7% in 2027, accounting for 63% and 56% of the total increase in those years.

In other words, after fifteen years of a flat line, the curve has turned upward, and a single category of customer is responsible for much of it.

Why data centers are a different kind of customer

A hyperscale data center does not connect to the grid the way a new subdivision or big box store does. Individual campuses can require hundreds of megawatts, which places them closer to heavy industry than to commercial real estate.

That changes how utilities have to plan. A retail development can usually be absorbed by existing distribution infrastructure. A 600 megawatt computing campus generally cannot, and serving it may require new substations, new transmission lines, new transformers and in some cases dedicated generation.

Each of those carries a multi-year permitting and construction timeline, which is the part that does not compress no matter how quickly the demand arrives.

What the Department of Energy is saying

The DOE Office of Electricity released a draft of the 2026 National Transmission Needs Study on July 9, 2026, opening a 60-day public comment period. The study is required under the Federal Power Act and is intended to identify current and anticipated capacity constraints on the national transmission system.

Its conclusion is direct. The draft finds a pressing need for additional transmission infrastructure driven by load growth from data centers, expanding domestic manufacturing, large industrial loads and a growing economy.

DOE's Assistant Secretary for the Office of Electricity has characterized the pace as faster than anything seen in decades, citing data centers, manufacturing growth and new forms of industry emerging almost by the month.

The study identifies Virginia and Texas as having the highest current data center electricity demand, with both projected to see some of the largest increases through 2030, alongside Arizona and Oregon. It also notes that transmission bottlenecks are already limiting delivery of electricity to Dominion Energy's rapidly growing data center load in Virginia.

That last detail is the one worth sitting with, because it describes a constraint that has already arrived somewhere rather than one that might arrive eventually.

Idaho is at the leading edge of this

Meta has already built its data center campus in Kuna, and additional facilities have been proposed around the Treasure Valley since.

The largest of them would dwarf it. Rhea Data, a subsidiary of Nevada-based infrastructure developer Arevia Power, has proposed a data center on federal Bureau of Land Management land south of Boise, near the Blacks Creek exit, with a project corridor extending from Ada County into Elmore County.

The application describes roughly 3,700 acres, approximately four million square feet of buildings and 600 megawatts of data processing capacity. Plans also include an on-site gas-fired plant with around 450 megawatts of generating capacity, intended as redundancy and for peak loading.

By area and capacity it would be roughly four times the size of Meta's Kuna facility, and it would be the first data center in Idaho proposed on public lands.

Rhea Data submitted its initial application to BLM in the fall of 2025. After the agency requested revisions for completeness, the company resubmitted in the spring of 2026. The proposal remains under review, details are preliminary, and Boise Mayor Lauren McLean has publicly criticized the push as shortsighted.

Whether or not that specific project advances, the scale of it indicates what states outside the traditional data center corridors are now being asked to accommodate.

A note on what this is not

Idaho Power has separately warned customers in parts of southwest Idaho that outages could last longer than normal during hot, dry and windy conditions. Those two stories are unrelated and should not be combined.

The outage warning concerns wildfire prevention. During elevated fire danger, Idaho Power applies what it calls enhanced protection settings to lines in high-risk areas, making protective equipment more sensitive and preventing lines from automatically re-energizing after a fault. The reasoning is that repeatedly energizing a line obstructed by a branch or debris can throw sparks into dry fuel.

The tradeoff is restoration time. A line that trips under those settings stays down until crews confirm it is safe, which can require physical patrols by truck, ATV, drone, helicopter or on foot, turning a momentary interruption into an outage lasting hours. A Public Safety Power Shutoff, in which lines are de-energized deliberately, is the more severe version of the same approach.

Idaho Power has stated explicitly that these measures are not driven by population growth or rising electricity demand. Attributing them to AI or data centers would be inaccurate.

What the warning does illustrate is how many competing pressures a utility already manages, and a new class of industrial customer is being added on top of all of them.

The question of who pays

When a technology company requests hundreds of megawatts, somebody has to fund the generation, transmission and distribution infrastructure required to serve it.

In principle those costs are assigned to the large customer creating the need. In practice, regulators across the country are examining how data center expansion affects electricity rates, infrastructure planning and existing ratepayers, and the answers vary by state and by utility.

The framing of data centers as simply good or bad does not survive contact with the details. They are the physical infrastructure behind cloud computing, streaming, online banking, business software and artificial intelligence, and they bring construction investment and tax revenue with them. They also tend to employ comparatively few people once operational relative to their footprint and power draw.

The more useful questions are narrower. How much new load can a given system absorb. How quickly can the supporting infrastructure be built. Who pays for those upgrades. And whether communities understand the scale of what they are approving at the point of approval rather than afterward.

The open question

The federal data points in one direction. Demand is rising after a long flat period, data centers are a primary driver, and the Department of Energy has concluded that the transmission system needs significant expansion to keep pace.

Generation can be added relatively quickly. Transmission cannot, and that mismatch is the structural problem underneath the whole conversation.

Idaho is not the largest market in this story, and it is not where the constraints have bitten first. But a 600 megawatt proposal on public land outside Boise, in a state that was not previously part of this conversation, suggests the boom is spreading well beyond Virginia and Texas.

Whether the grid can absorb it is a question that gets answered one interconnection request, one transmission line and one regulatory proceeding at a time. The recent news out of Idaho is the tip of it.

Figures cited are from the U.S. Energy Information Administration September 2026 Short-Term Energy Outlook and the U.S. Department of Energy draft 2026 National Transmission Needs Study. Details of the Rhea Data proposal are as reported at the time of writing and the application remains under BLM review, so confirm current status before relying on specifics.