U.S. electricity 2030: Work harder, but also smarter
- By Vincent Minier & Vincent Petit
- 21 Apr 2026
- 8 min read
After 15 years of nearly flat demand, U.S. electricity demand growth is back. It is now projected across forecasts to rise from 700 to 1,100 terawatt-hours by 2030 — a 20% to 25% increase compared with today’s levels. This massive and sudden shift is tied not only to significant investment in AI factories, which are expected to account for 40% to 50% of total growth, but also to the modernization of energy systems across all sectors of the economy, including industrial reshoring, EV penetration, and building heating system upgrades.
Yet supply is chasing demand. There is not enough room on the grid for all that demand, and the grid cannot be developed fast enough. The risk is price inflation and reliability issues, with the DOE estimating that average loss-of-load hours could increase 100-fold by 2030 if nothing is done.
The solution is known: flexibility. There is considerable potential for the grid to operate more efficiently and, at its current size, to connect more load. Duke University has shown that 70 to 200 gigawatts of data centers could be plugged onto the grid if they implemented flexibility provisions. Schneider Electric and the Energy Transitions Commission have explored the potential for creating additional headroom in the existing building stock and have shown, through both macro- and micro-modeling approaches, that 20% to 40% of peak-power headroom and up to 35% energy headroom could be provided at the building level. Such headroom would come at no cost to the grid operator, and payback periods for building occupants would be less than 10 years for 85% of the use cases studied.
A conservative adoption scenario focused on commercial buildings could provide at least 20 gigawatts of additional headroom across the U.S. in the coming years.
- 700 - 1,100 TWhProjected U.S. electricity demand growth by 2030, a 20-25% increase
- 20% - 40%Peak power headroom achievable at building level through flexibility solutions
- 100-foldPotential increase in loss-of-load hours by 2030 without action (DoE estimate)
After nearly 15 years of stagnation, U.S. electricity demand is back to growth. Since 2021, electricity demand has grown by an average of 2% annually. It accelerated recently, with 3% growth in 2024 and 2.8% in 2025.
Data centers have been a key driver of this growth. Data center demand tripled between 2014 and 2023 to reach about 170 terawatt-hours, or about 4% of U.S. electricity demand. Artificial intelligence is one of the main reasons for the recent acceleration in the sector.
There are other factors why U.S. electricity demand is growing again. Building electrification, industrial development, and EVs all played a role in the recent acceleration, suggesting that this return to growth is structural – meaning here to stay.
By 2030, we estimate that AI and data center electricity demand could reach 480 terawatt-hours, although uncertainty suggests a range of scenarios between 350 to 720 TWh, globally consistent with other studies. This increase in demand would represent only 40% to 50% of total growth. The increasing adoption of EVs, even if often projected to be modest by 2030, also adds to demand.
Buildings continue to electrify, with heat pump sales now 30% higher than gas boiler sales, a trend visible since 2022. The reindustrialization of the U.S. economy and the large investments announced across all sectors of industry also add to electricity demand. Overall, we estimate total electricity demand growth of 700 to 1,100 terawatt-hours by 2030 across our scenarios. This represents a 20% to 25% increase in demand in just a few years and corresponds to a load increase of 100 to 160 gigawatts.
This return to growth is good news. After a decade and a half of stagnant demand, it is expected to support a new wave of modernization and expansion for the U.S. economy. AI and its broad applications, economically-compelling and environmentally-friendly heating systems, better cars, and new industries will all contribute to higher living standards, more jobs, and higher GDP.
The issue is that we may have forgotten how to supply that growth. For decades this was not a problem, until demand flattened about 15 years ago (Figure 1). But returning to growth is now a challenge. The Department of Energy has estimated that, without radical change in the provision of supply resources in the coming years, the U.S. could face a 100-fold increase in average loss-of-load hours by 2030. This figure is closer to 35-fold if currently scheduled capacity retirements are put on hold, but it would still represent a major reliability challenge for the U.S. grid.
This issue first has to do with lagging supply. While the typical duration for a connection request did not exceed two years in the 2000s, it has reached a median of five years today. The result is congestion, with capacity in the queue exceeding the entire U.S. power fleet today. Only 13% of capacity that submitted interconnection requests between 2000 and 2019 had reached commercial operation by 2024, while 77% had been withdrawn and 10% remained active).
The second issue is the lack of firm supply. Of the 209 gigawatts of capacity planned to come online by 2030, only 22 gigawatts are considered firm, meaning eligible for dispatch during peak hours. The vast majority of new capacity, mainly wind and solar, is not considered firm power.
Thirdly: the rate of power grid buildout. The U.S. built just 888 miles of 345-kilovolt-and-higher transmission in 2024, against a need for 5,000 miles a year. Meeting load growth requires a sixfold acceleration.
Finally, we have uncertainty itself. There may be systematic flaws in future consumption estimates, and utility forecasts may overstate future demand. Data center demand, one of the key drivers of growth in the short term, is also highly uncertain, as demonstrated in our scenarios and other research. This uncertainty weakens the business case for highly capital-intensive capacity and grid investments.
Whether supply or demand falls short, the consequence of a mismatch is electricity price inflation. Prices have already started to rise, and this is just the beginning. The consequences will be profound, not only for meeting rising demand but also across the broader U.S. economy, because electricity prices directly affect the cost of goods and services.
We need to react and organize, because the U.S. is now entering a period of structural electricity demand growth. Forecasts may have flaws, but there is broad consensus on the long-term trend: all indicators strongly associate modernization of the U.S. economy with expanding electricity demand. After all, nothing truly innovative runs on anything else.
All of these issues require a response. It is beyond the scope of this article to address each of them in detail. What we want to highlight here is the potential to use the U.S. power system more efficiently so it can cope with this increase in demand in the short term and at low cost, helping address the inflation and reliability risks described earlier.
As a starting point, we must recognize that power systems are structurally underused. They are built to meet peak demand. Figure 2 shows this in a simple building archetype load profile. The average load is 38 kilowatts, but it varies throughout the day, from 25 kilowatts up to three times that average, at about 100 kilowatts. Grid power capacity is designed to supply that peak load. When demand is not at its peak, the entire system runs below capacity.
When all buildings, data centers and factories across the U.S. are added together, each with its own load profile, the result is structural underuse of the power system. This is a nuanced and complex topic, and situations vary significantly across time and place. Yet a Duke University study found that the U.S. power system may be operating at only 53% of its capacity.
That level of underuse is not a problem so much as an opportunity. If load can be reduced during peak periods or shifted to other times, then the same system can accommodate new load at no cost. Duke University calls this “headroom” in the power system, created by flexibility solutions.
For a long time, shifting loads was largely not achievable at scale because of the inelasticity of electricity demand. Some loads, such as water heaters, have traditionally shown some flexibility, and new AI factories are typically more “amendable to load flexibility”. Some assets have also long been equipped with local generation, typically diesel generators. This is notably the case for data centers. While primarily deployed for reliability, they can play an important role in shifting load demand. Because they bring additional supply behind the meter, they can relieve the power system from supplying all or part of that load at certain times. The power system then supplies only a net load, after accounting for what distributed resources provide directly. Duke University researchers studied the potential for flexibility provisions in data centers. They found that the U.S. power system could accommodate 76 to 215 gigawatts of new AI load if such flexibility provisions were implemented.
Yet the bulk of loads on the power system, especially in buildings and factories, remains inelastic and difficult to shift. But the emergence of distributed energy resources, including modern rooftop solar systems and batteries, along with the development of digital controls, has fundamentally changed that. Distributed energy resources produce electricity locally, and batteries can store excess supply and use it to shift the hourly profile of net load as needed. All this is possible because real-time controls can now be deployed at scale through modern digital technologies.
When properly designed, these flexibility solutions can provide multiple benefits at an unprecedented scale:
They help reduce peak levels at the asset
Whether a data center, building or factory. They create peak-power headroom, allowing new loads to be added because more peak power is available.
They help shift net-load demand to hours when demand is low relative to supply
Helping flatten the load curve and increasing overall use of grid assets. They create energy headroom, allowing new loads to be added because new margins become available to the power system.
They contribute to investment deferral
Because new margins are created on the power system, less investment is required in new capacity. And as existing assets further electrify, such as EV charging within buildings, these solutions can minimize the extra demand drawn from the power system and reduce investment needs.
The Energy Transitions Commission has estimated that demand-side flexibility solutions could provide up to 20% of building electricity demand, 50% of passenger EV demand and 45% of industrial electricity demand by 2050 (Figure 3).
If properly activated, such levels of flexibility would not only help accommodate rapid electricity demand growth but also substantially reduce infrastructure costs.
There is a large payoff in activating this flexibility at scale, and buildings, including industrial buildings, are a prime target.
At the Schneider Electric Research Institute, we explored this potential in recent research complementing the above effort with additional modeling. The question we asked was:
How much headroom can be created by deploying flexibility at the building level?
The value of this work is that it confirms the data above by offering a complementary assessment at a granular level, specifically the building level, across different building archetypes and geographies, subject to different weather patterns and electricity tariff structures.
In this report, we considered five building archetypes — a hospital, large office, school, small hotel and strip mall — across 11 geographies, including details for three U.S. regions. To make the analysis more realistic, we also accounted for the growing adoption of EVs and considered the effect of EV charging, which adds to existing peak demand and worsens headroom potential.
For each case, we modeled the load profile every hour of the day, every day of the year, both for our baseline and after flexibility provisions were deployed. We presented results on load duration curves. We measured both peak-power and energy headroom. Peak headroom is the difference between peak levels (x = 0) at baseline and after flexibility deployment, and energy headroom is the difference between the integrals of the baseline duration curve and the flexible one. We also separately analyzed the effect in terms of investment deferral.
Figure 4 — Illustrative load duration curve
Figure 5 summarizes the results of our modeling exercise. We found that, in the U.S., buildings consistently provide 20% to 40% of peak-power headroom, and up to 60% in the best cases, along with energy headroom of up to 35%.
Figure 5 — Headroom potential at the building level
Figure 6 shows additional results across different U.S. and Canadian regions and different building archetypes. More details are available in our recent report.
Figure 6 — Headroom potential at the building level. Left: peak-power headroom. Right: energy headroom.
These granular results confirm the broader estimates from the Energy Transitions Commission. They are also consistent with earlier studies. A massive opportunity exists to create both peak-power headroom and energy headroom in the U.S. power system.
Additionally, this comes at no cost to the utility. The investment is borne by the building occupant, and in our modeling, the occupant optimizes for cost, not flexibility. The headroom therefore emerges as an additional benefit, contingent on retail electricity price structures. Some of our previous work has shown that such investment typically generates payback periods of less than five years for 25% of use cases and less than 10 years for 70% of use cases, making the transition economically compelling and a key resource for easing electricity price inflation. The nature of this solution design also suggests that the flexibility potential evaluated here is conservative.
Deploying flexibility solutions is therefore both technically and economically sound and, in buildings, likely one of the best options available.
In the U.S., buildings account for more than 70% of final electricity demand and a similar share of peak demand. Commercial buildings are more aggregated, and they also benefit from a strong economic rationale and fewer investment barriers. Assuming an ambitious adoption rate of 20% to 25% among commercial buildings could quickly yield about 20 gigawatts of additional peak headroom for the system. This figure remains conservative compared with other studies that estimate full potential, including the residential sector, at about 60 to 200 gigawatts.
Our estimate of 20GW remains voluntarily conservative for several reasons. First, it focuses only on commercial buildings, which represent a much lower share of the peak than the residential sector. Since the data discussed above reflects commercial building potential, we have not integrated any further analysis on the potential within the residential sector. Moreover, since the commercial sector is less affected by access to capital than the residential sector, it is reasonable to focus on it first for rapid headroom creation. Second, we have kept rates of adoption conservative once again in light of the time criticality of the discussion. Third, our estimates of full potential do not include additional energy efficiency measures (which can reduce the peak) and on the contrary integrate the growing constraint on peak from the deployment of electric vehicles within building premises.
Related
Dive DeeperFor a deeper look, download the report "Grid Relief from Smart Buildings" to see how 65 building archetypes across 13 regions were modeled, and what the findings reveal about unlocking grid capacity through smart charging, rooftop solar, and stationary storage.
Related
Dive DeeperFor a deeper look, read the report "Power Systems Transformation: Delivering Competitive, Resilient Electricity in High-Renewable Systems" in partnership with the Energy Transitions Commission.
_BUILDER.IORectangle.webp)
Latest from Schneider Electric Research Institute
Powering sustainable AI in the United States
Beyond grid dependency: Technical and economic case for new energy systems in green steel
Copper dynamics and horizons: Building sector mitigation and electric vehicle-driven growth
United States water infrastructure: From fixing leaks to regenerating the system
Need help?
Product Selector
Quickly and easily find the right products and accessories for your applications.
Get a Quote
Start your sales inquiry online and an expert will connect with you.
Where to buy?
Easily find the nearest Schneider Electric distributor in your location.
Help Center
Find support resources for all your needs, in one place.