This is an extract from a recent report “Annual Energy Outlook 2026” published by The U.S. Energy Information Administration (EIA). This extract focuses on Electricity.
Data centers bolster electricity demand growth
After 15 years of nearly flat U.S. electricity consumption, demand has increased by 2.1% per year, on average, over the last five years. Electricity consumption will continue growing through 2050 at a rate of 0.9% to 1.6%, with data center server energy use a major factor. Energy use in commercial buildings, home to data center activity, grows more rapidly than in the residential or industrial sectors in all modeled cases. AI servers will increasingly skew more energy intensive, the installed stock of AI servers grows exponentially through at least 2040, and computational efficiency will increase over time.
In the High Electricity Demand case, the exponential growth in AI servers will continue through 2050, and the rate of improvement in server computational efficiency will reflect historical trends. These assumptions lead data center server energy use alone to grow to 818 billion kilowatthours in 2050 in the High Electricity Demand case. Server electricity consumption in 2050 is more than 16 times that in 2020. In 2050, the High Electricity Demand case shows 84% more data center server electricity use than the Counterfactual Baseline case.
Average annual electricity consumption growth through 2050 in the High Electricity Demand (1.6%) and High Economic Growth (1.5%) cases are comparable. In all cases, electricity use is highest in the commercial sector. By case design, commercial buildings alone account for the incremental electricity growth in the High Electricity Demand case—largely to meet additional data center server and space cooling demand. The entire economy would need to grow at the rate projected in the High Economic Growth case to match the data center-related electricity growth in the High Electricity Demand case. Data center server electricity use grows fastest in the South Atlantic and the West South Central census divisions, home to Virginia and Texas, respectively.
Timing of electricity demand affects power sector costs
Although electric vehicle (EV) and data center server demand show significant growth across most of the cases, they account for a relatively small share of total load projections in 2050. Overall electricity demand grows between 25% and 50% by 2050, with EV and data center server demand accounting for between 50% and 80% of that growth. However, EV and data center electricity demand only accounts for between 10% and 25% of total demand in 2050. Despite the relatively low share of total demand, EV and data center server demand growth influences the load factor of the system, which is the ratio of the average system load to peak. Changes in the load factor can affect how system costs, such as reliability and distribution, are incorporated into customer rates. In the Alternative Transportation case, where demand from EVs is much lower, a greater portion of demand takes place during the peak afternoon and early evening hours.
As a result, distribution costs in this case constitute a higher share of total electricity costs than in the Counterfactual Baseline case because transmission and distribution investment is tied to growth in peak demand but shared over the smaller total sales. Conversely, in the High Electricity Demand case, increased electricity use from data center servers occurs consistently throughout the day, along with additional growth in space cooling demand that is more pronounced midday. In this case, average generation costs increase compared with the Counterfactual Baseline to recover costs of building additional generating capacity along with increased operating costs to meet the higher demand levels. The projections include EV charging in either residential or commercial sectors to reflect the customer charging location. Load for data center servers is assumed to be relatively constant throughout the day.
The electricity generation mix depends on resource availability and market conditions
Total generation grows between 25% and 50% through 2050 across all cases, with the most growth resulting from the cases that assume high economic growth and high electricity demand in the commercial sector. To meet this increasing demand for electricity, total installed electric generating capacity increases between 50% and 90% across modeled cases. The mix of new electricity generation sources that meets this demand varies considerably depending on the assumptions within each side case. Natural gas, solar, and wind generation increasingly meet U.S. power demand across all cases examined here.
The combined generation share of these technologies rises from about 60% in 2025 to around 80% in most cases by 2050; in the Counterfactual Baseline case, natural gas accounts for about 40%, wind for 20%, and solar for 20% in 2050. Although the absolute amount of natural gas generation increases, its share of electricity generation generally stays flat in most cases. This is in contrast to coal; in 2025, coal’s share is 16%, but it declines by 2050 in all cases. In cases where regulations curbing greenhouse gas emissions from power plants remain in place, coal’s share declines to less than 1% by 2050; in cases where these regulations are not in place, its share declines to about 5% over the same period.
Nuclear generation remains relatively stable in most scenarios, except for the Low Oil and Gas Supply case, in which the high cost of natural gas significantly enhances nuclear power’s economic competitiveness. However, the share of nuclear generation declines in all cases considered here, from 17% in 2025 to between 12% and 15% by 2050. The assumptions regarding natural gas resources and prices play a crucial role in shaping the generation mix by altering the relative economic competitiveness of other resources. Consistently a strong correlation between natural gas prices and electricity prices is observed, particularly in High Oil and Gas Supply and Low Oil and Gas Supply cases. This correlation underscores the critical role natural gas plays in setting the dispatch margin on the electricity grid.
Electricity capacity decisions are affected by policy and regulations
Natural gas prices and technology costs affect the generation mix because of tight cost competition between natural gas and renewables for new power plant construction. Wind capacity additions are very sensitive to natural gas price changes, with over five times more additions in the Low Oil and Gas Supply case than in the High Oil and Gas Supply case. Solar additions, meanwhile, vary by a factor of two across the cases and are less sensitive to natural gas prices in part because of their tendency to suppress peak mid-day electricity prices.
Displacing generation from low-efficiency simple turbines to higher efficiency combined-cycle units reduces the revenue available to new solar units, which in turn makes solar less likely to be competitive under a wider range of natural gas prices. In cases with more generation from wind and solar resources, greater total capacity is generally needed due to the seasonal and diurnal limitations of wind and solar and the necessity for dispatchable resources during times of low or no wind and solar output. Dispatchable resources typically include coal-fired generation, natural gas-fired generation, oil-fired generation, and nuclear power.
Between 100 GW and 125 GW of coal capacity retirements is projected by 2050—about 65% to 80% of the coal fleet—across all our cases, with the exception of the Alternative Electricity cases. If regulations that aim to curb greenhouse gas emissions from power plants are no longer in effect, cumulative coal retirements decrease to about 70 GW—just over half of the retirements is projected in most other cases. In cases in which these regulations remain in place, about 25 GW to 40 GW of coal plants convert to natural gas and coal co-firing (included in fossil steam) before they ultimately retire by 2038. Natural gas capacity additions are projected to reach about 430 GW in the Counterfactual Baseline case and about 365 GW in the Alternative Electricity case.
Renewables additions vary by region
Renewable power capacity will increase in all regions of the United States in all cases, although regional resource availability results in a varying renewable resource mix. Growth in energy storage capacity, primarily batteries, is largely correlated with growth in solar capacity, as diurnal storage systems are a good fit for the daily and seasonal variability of solar output, which is often offset by a few hours from system peak load periods. Additional storage is often co-located and jointly operated with solar plants. In 2050, about 2.5 GW to 25 GW of batteries, representing 2% to 10% of installed storage capacity across the cases, are located within a solar-hybrid plant. This co-located storage capacity is accounted for within the capacity of the hybrid plant, which is reported in the figure above as solar photovoltaic capacity.
Across all cases, renewable power capacity in the Mid-Continent region increases between 75 GW and 300 GW throughout the projection, with renewables capacity projected to increase by 160 GW in the Counterfactual Baseline case. Most of the increases come from wind generation, which grows between 20 GW and 170 GW relative to 2025, with the Counterfactual Baseline case accounting for 85 GW of wind capacity growth. The Mid-Continent region has a lot of untapped wind potential and low cost wind resources, which drives the expansion of wind capacity. Wind is also more economical in that region because of the above-average price of natural gas in the central portion of the country, which ranges from 2% to 6% above the national average after 2035.
Expansion of solar capacity through 2050 remains between 100% and 235% across most cases. The Southeast, the region with the most significant solar growth, has a two-fold increase in solar capacity in the High Oil and Gas Supply case when natural gas prices are below $3 per million British thermal units (MMBtu) in real 2025 dollars, or when the costs of renewable technologies are high. This increase is closer to five-to-sevenfold in the Low Oil and Gas Supply case when natural gas prices are above $10/MMBtu, or when the costs of renewable technologies are low.
Future U.S. coal demand largely depends on electric sector policies, exports may grow
Future coal use in the United States in the projections is heavily dependent on policy assumptions. Electricity sector regulations finalized in 2024 target carbon dioxide emissions from coal, oil, and natural gas plants and would require existing steam coal plants without carbon capture to convert to natural gas or retire by 2038. With those policies in place, coal use in the electric power sector, which is currently 75% of coal disposition in the United States, decreases from 388 million short tons (MMst) in 2025 to nearly nothing by 2050. In the two cases which ran without those regulations in place—Alternative Electricity and Combination— coal use for electricity generation decreases from 388 MMst in 2025 to about 150 MMst in 2050, and accounts for about 50% of U.S. supply instead of less than 1% in the other cases.
Power sector operators’ retirement plans reported to EIA as of October 2025 account for about half of the decrease in generation, with the rest coming from assumed retirements based on the cost projections in NEMS. Coal consumption for heat and power in the industrial sector decreases by about 5 MMst, or 30% of industrial use by 2050, across all cases examined here. This decrease occurs because coal is partially replaced by natural gas and electricity. Natural gas replaces coal as a boiler fuel driven by lower operations and maintenance costs, greater thermal efficiency, and more expedient fuel accessibility. With a drop-off in domestic coal consumption, U.S. coal produced for exports increases throughout the projection period in all examined cases by about 20% from the 96 MMst exported in 2025 to 115 MMst in 2050, assuming no restrictions on global coal trade.
Access the report here