A University of Michigan study reveals a new layer of complexity for keeping our lights on and homes comfortable. The research, led by postdoctoral fellow Mai Shi, tackles a critical question for utilities and policymakers: what will push the power grid to its absolute limits in the coming decades? The answer lies at the intersection of two powerful forces—our changing climate and the technology we choose to heat and cool our buildings. The findings suggest that for many regions, the choices we make about technology inside our walls could strain the grid more than the weather outside them.
“We need to plan for extremes,” Shi told me. Her team’s focus wasn’t on average daily energy use. It was on those few critical hours each year when demand soars, when every air conditioner or heater clicks on at once. A grid built for moderate conditions that fails under that peak stress is a grid that has failed entirely. To forecast these future peaks, the researchers employed machine learning models, analyzing data from 183 U.S. counties. They ran thousands of simulations combining four climate futures with 18 different pathways for how building technologies might spread. This approach lets us see the isolated and combined effects of a warming planet and a shifting technological landscape.
Consider the technology piece first, specifically the rapid rise of heat pumps. These electric systems are far more efficient than traditional air conditioners for cooling. For heating, they can be several times more efficient than a standard electric furnace. This efficiency is a win for lowering overall annual energy consumption and carbon emissions. But there’s a catch for grid planners. A heat pump moves heat rather than creating it from scratch, like burning gas. This process becomes less efficient as outdoor temperatures plunge. During a deep winter freeze, a heat pump must work much harder, drawing a significant and sustained surge of electricity. A gas furnace, in contrast, ignites with a relatively consistent fuel requirement regardless of the cold.
The modeling bears this out. When the team isolated the effect of technology diffusion—imagining a future with widespread heat pump and insulation adoption but no additional climate change—the results were dramatic for colder regions. Extreme peak electricity demand in the Northeast and Midwest could skyrocket by 60% to 104% by 2060, driven almost entirely by heating needs. In more temperate zones, the increase was closer to 15%. In already-hot regions like Texas, peak demand actually decreased slightly, as efficient heat pumps for cooling eased summer strain more than winter heating added to it.
Now, layer in climate change alone. A world that is hotter on average, with more intense and frequent heat waves, generally means more air conditioning use. The models projected that by 2060, climate-driven increases in cooling demand would push extreme peaks up by 3% to 9% across most of the country. The notable exception was the Pacific Northwest, where heating currently drives winter peaks. There, a warmer future actually decreased the extreme peak demand by about 3%.
The real world, of course, will not present these factors in isolation. They will combine, and the interplay is complex. For cold regions, this interplay offers a curious silver lining. Climate change means winters are somewhat milder. This warming improves the operating efficiency of heat pumps during those critical cold snaps, slightly tempering the enormous peak demand surge driven by their adoption. In hot regions, the story flips. Rising temperatures drive peak cooling demand higher, but the diffusion of efficient heat pump technology acts as a partial buffer, reducing the peak increase compared to a climate-change-only scenario.
This complexity creates substantial uncertainty for anyone planning our energy infrastructure, which has a lead time of decades. “Tech diffusion may have a larger impact than climate change on peak demand in a lot of regions,” Shi noted. This is a crucial insight. While a single utility has limited power to slow global warming, it has direct influence, through incentives and partnerships, over which technologies gain traction in its service area. A utility in Michigan, for instance, might look at these forecasts and prioritize programs that pair heat pumps with supplemental heating systems or targeted grid upgrades in areas with older housing stock.
Michael Craig, the study’s senior author and an associate professor at U-M, emphasized that climate risk remains fundamental. “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,” he said. The study, published in the Proceedings of the National Academy of Sciences, provides a new framework to weigh these dual uncertainties. It moves the conversation beyond simply building more renewable generation—which addresses the source of electrons—to managing the timing and intensity of demand itself.
The ultimate takeaway is that grid reliability is no longer just an energy problem. It is a deeply integrated climate and technology challenge. Preparing for the future means running the numbers on how a winter polar vortex in Minnesota will stress a grid serving millions of high-efficiency heat pumps, and how a prolonged heat dome in Arizona will test a fleet of next-generation coolers. The decisions we make today about the appliances in our basements and attics will resonate for decades, shaping how resilient our communities are when the weather turns most extreme.
- Changing climate impacts on energy demand
- Technological advancements in heating and cooling
- Effects of heat pumps on electricity demand
- Machine learning models for forecasting peaks
- Regional differences in peak demand increases
- Complex interplay between technology and climate
| Region | Peak Demand Increase by 2060 |
|---|---|
| Northeast | 60% to 104% |
| Midwest | 60% to 104% |
| Temperate Zones | ~15% |
| Texas | Decreased slightly |
| Pacific Northwest | Decreased by ~3% |
