U.S. electricity use is headed for a record in the Energy Information Administration’s September forecast, with data-center construction and manufacturing helping drive the increase. The agency projects electricity sales of 4,135 billion kilowatthours in 2026, almost 2% above 2025, followed by another increase in 2027.
That national number does not automatically translate into one household’s next bill. Retail rates are set through a patchwork of utilities, fuel costs, state regulators, contracts and local grid investments. It does mean that a large new source of commercial demand is arriving in regions where power systems must add generation, wires and substations, and those investments are eventually paid for somewhere.
EIA’s record forecast is about electricity sales, not a single giant machine
The September Short-Term Energy Outlook says U.S. electricity consumption reaches record levels in its forecast. EIA expects commercial-sector sales to rise 3.3% in 2026 and says that sector accounts for 63% of the forecast increase. Industrial sales add another source of growth. The forecast therefore describes a broad demand shift, even though data centers are one of its most visible pieces.
Data centers use electricity to run servers, cooling equipment, networking gear and backup systems. A facility’s name can make it sound like a quiet office building, but its load can be large and steady. EIA’s longer-term analysis separates server consumption from other commercial computing and projects continued growth across the commercial building stock. That separation matters because a data center can change how planners think about a local system’s peak demand and its need for round-the-clock capacity.
The national forecast also contains an important caution: it is a forecast, not a rate order. EIA’s September 9 release projects record generation of 4,368 billion kilowatthours in 2026, but it does not declare that every customer will pay more. Generation and sales are national measures; a customer’s bill is a local retail outcome.
Texas shows why location matters as much as the national total
The West South Central region is the largest contributor to electricity-sales growth in EIA’s forecast, and EIA specifically cites data-center development and manufacturing. Texas has become a useful illustration of the strain and the uncertainty. The state paused connections for some new data-center proposals while it audited the projects, yet the region still accounts for a large share of projected growth.
EIA’s weekly ERCOT analysis recorded average hourly load of 74.5 gigawatts for the week ending August 22, 2026, a record in that series. Hot weather was a central factor, which is a reminder that data centers are not the only driver. Air conditioning, industrial growth and population all matter. A useful account of grid pressure has to keep those forces together rather than assigning every high-load day to AI servers.
Location also changes the available response. One area may have transmission that can bring new generation to demand centers; another may need years of upgrades. Some utilities can add solar, storage, gas generation, demand-response contracts or large customers willing to shift use at peak times. The cost and timing of those choices vary considerably, which is why a national record forecast cannot answer a particular customer’s rate question by itself.
Why an expanding grid can affect bills without guaranteeing an increase
Electric bills commonly combine energy charges with the cost of maintaining and expanding the delivery system. When a utility seeks to build infrastructure, state regulators decide how costs are recovered and how different customer classes are treated. A large new customer can contribute substantial revenue and help pay for shared assets; it can also require upgrades that would not otherwise be needed. The result depends on the tariffs and agreements in that utility’s territory.
EIA’s Annual Energy Outlook narrative describes data-center servers as a major factor in long-run demand scenarios after years of relatively flat U.S. electricity use. It also makes clear that scenarios differ: assumed server efficiency, the pace of installations and the type of buildings hosting equipment change the outcome. That is a better frame than treating a forecast as an inevitability.
The immediate fact is still substantial. EIA expects record U.S. electricity consumption in 2026 and identifies data centers as part of the commercial-demand growth behind it. Whether a particular bill follows depends on a more local chain: the utility’s demand, its generation mix, transmission plans, regulatory decisions and the terms offered to new large users. Those are the documents worth watching when a national power forecast reaches a household budget.
For that reason, a customer concerned about rates should distinguish a national forecast from a pending local proceeding. A utility commission filing, a proposed transmission project or a published tariff will normally say more about near-term costs than a national headline will. EIA’s forecast establishes the scale of the demand shift; local public records show how a particular system proposes to meet it and who regulators expect to bear the cost.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
More from Morning Overview
