New research led by the University of Michigan provides utilities and policymakers with forecasts that can help them better prepare for future extremes in peak energy demand in the United States.
"We focused on the extreme peak demands because we want to keep our power systems reliable," said Mai Shi , a postdoctoral research fellow at the U-M School for Environment and Sustainability, or SEAS . "If the system works for moderate conditions but fails in extremes, it doesn't work. That's not how we build the power grid. We need to plan for extremes."
In the new study, supported by the U.S. National Science Foundation, Shi and colleagues modeled how those extremes will evolve between 2020 and 2060 with an eye to heating and cooling. Heating and cooling are among the top consumers of electricity for both residential and commercial buildings. The team analyzed how these would shape extremes in peak power demand against the backdrop of two important changes.
The first is climate change. Not only is the world's average global temperature climbing, but heat waves are growing more intense and frequent, meaning cooling buildings will require more energy.
The second change is that technology that helps heat and cool buildings more efficiently will become more common, diffusing to more buildings over time. For the technology component, the team considered the growing use of insulation and heat pumps.
Heat pumps are electrified heating and cooling systems that can offer substantial efficiency gains compared with air conditioners, for example. The rub is that, being fully electric, heat pumps also place a greater burden on the grid compared with conventional systems that use fossil fuels, like gas furnaces. This is also compounded by the fact that the efficiency of heat pumps drops the colder it gets. Thus, increased heat pump adoption would also mean higher electricity demands, especially in regions with cold winters.
Knowing that, it's not surprising that the team found that the impacts on extreme peak demands varied across the country and its regional climates. Still, one general trend did emerge.
"Tech diffusion may have a larger impact than climate change on peak demand in a lot of regions," said Shi, who is also a member of U-M's Center for Sustainable Systems .
To perform the study, published in the journal Proceedings of the National Academy of Sciences , the team used machine learning algorithms to model energy demand. The researchers considered four climate change scenarios and 18 different technology diffusion pathways in 183 counties across the U.S. This allowed them to investigate how peak demand could evolve without technology diffusion, without climate change and a multitude of scenarios where their impacts compound.
When only climate change is considered, extreme peak demand is projected to increase by 3%-9% over the next four decades everywhere except the Northwest. Outside the Northwest, the peaks are driven by cooling demands, but in the Northwest, they are driven by heating and this scenario shows a 3% decrease in peak demand extremes.
When only technology diffusion is considered, the increasing adoption of heat pumps drives up extreme peak demands in cold regions like the Northeast and Midwest by 60%-104% due to heating requirements. The increases are more modest—closer to 15%—in more temperate regions, including parts of the Southwest and Southeast. In hot regions, like Texas and California, peak demand extremes decrease.
The picture becomes more complex when climate change and tech diffusion are considered together. For example, rising temperatures can improve heat pump efficiency during winters in cold regions. Although warming also increases extreme peak demands in hot regions, tech diffusion does mitigate the impact compared to the climate-change-only scenario.
There is, thus, a substantial degree of uncertainty in the results. But how the study addresses that uncertainty will also be valuable for researchers, professionals and policymakers working on the power grid, the team said.
While there is only so much an individual region can do on its own to limit climate change, this and studies like these can help regions adjust incentives and regulations to steer tech diffusion to a preferred path, Shi said. But they also underscore that climate change cannot be ignored.
"Climate change still matters a lot," said senior author Michael Craig , associate professor at SEAS and Industrial and Operations Engineering . "If I'm thinking about my power system and what I should worry about, climate change is still an important source of risk and uncertainty."
The research team also included Ziqi Wei of the Massachusetts Institute of Technology and Flavio Lehner of Cornell University. Additional support for the research was provided by the U-M Graham Sustainability Institute.
Proceedings of the National Academy of Sciences
Disentangling the evolving effects of climate change and technology diffusion on building electricity demand
20-Oct-2026