Power grid sits at the center of this dementia and brain health question.

America’s power grids are facing an unprecedented squeeze as electricity demand surges during heat waves, straining infrastructure built for a different era. When temperatures soar above 100°F—as they did in Houston, Dallas, and San Antonio during recent summers—millions of air conditioners kick in simultaneously, pushing demand to historic records that utilities struggle to meet. On July 18, 2023, Texas’s ERCOT system hit 82,579 MWh during the 6:00 p.m. hour, shattering the previous record of 79,830 MWh set just one year earlier. This article examines why power grids are struggling under peak demand, what regions are most vulnerable, how data centers are reshaping demand patterns, and what the grid’s future looks like as extreme weather becomes more frequent.

For households with older adults or those managing chronic conditions, understanding these grid pressures matters—power failures can have serious health consequences. The core problem is simple but growing more severe: demand now exceeds what infrastructure can reliably supply during summer heat waves. ERCOT broke electricity demand records on four consecutive days in June 2023 alone, with hourly demand exceeding 80,000 MWh each day from June 26 through June 29. Across the country, the Mid-Atlantic’s PJM system (serving 67 million people) experienced a similar crisis during summer 2025, when actual peak load hit 161,000 MW in June—exceeding forecasters’ predictions by 7,000 MW. These aren’t close calls; they’re warnings that the grid is running near maximum capacity during peak heat.

Why Does Heat Wave Demand Peak at Dangerous Levels?
Regional Vulnerability and the Coming Shortage
Data Centers as the Hidden Driver of Demand Growth
How Utilities Are Attempting to Expand Capacity
What Happens When Supply Doesn’t Meet Demand?
Renewable Energy’s Growing But Incomplete Role
What the Grid Looks Like in 2026 and Beyond
Conclusion

Why Does Heat Wave Demand Peak at Dangerous Levels?

During heat waves, electricity demand concentrates during evening hours when people return home and turn on air conditioning while outdoor temperatures remain dangerously high. Air conditioning uses far more electricity than most household appliances—a single unit can draw 3,500-5,000 watts continuously—and when millions run simultaneously across a region during peak evening hours, the demand surge tests every component of the power system. The Texas heat wave of June-July 2023 demonstrated this vividly: as temperatures climbed past 100°F across major cities, ERCOT operators faced back-to-back records not because of equipment failure, but because customer demand simply overwhelmed available supply during the critical 4:00 p.m. to 10:00 p.m.

window when cooling demand peaks while the sun’s contribution to solar generation drops. What makes this particularly vulnerable is timing. Peak demand arrives exactly when large generating plants may be offline for maintenance, when older coal plants run less efficiently in extreme heat, and when renewable resources like solar generation drops as evening approaches. Some power plants are designed to run 24/7, while others are dispatched only during peak periods—and during heat waves, utilities run everything they have, competing with each other across regions for power supply. This creates a domino effect: if Texas’s grid lacks capacity, it cannot purchase power from neighboring regions because those regions are also at maximum capacity.

Why Does Heat Wave Demand Peak at Dangerous Levels?

Regional Vulnerability and the Coming Shortage

The Mid-Atlantic region offers a cautionary snapshot of what happens when growth outpaces infrastructure. PJM’s 2025 summer saw actual peak demand exceed predictions, with June 23 and June 24 setting the 3rd- and 4th-highest peak demand days in PJM’s entire history. More concerning: PJM’s capacity forecasts predict that starting summer 2026, the region will have only enough power generation to maintain grid reliability—meaning virtually no buffer for unexpected outages, equipment failures, or more severe heat waves than anticipated. A capacity shortage in the PJM region could arrive as early as the 2026/2027 Delivery Year, which begins June 1, 2026. This isn’t speculation; it’s documented in PJM’s Long-Term Load Forecast report.

Texas faces a different but related challenge. ERCOT demand is forecast to rise 14% in 2026 alone as large data centers and cryptocurrency mining facilities come online, pushing total demand to 425 TWh annually. To meet this growth, ERCOT has added approximately 23 GW of new generation between 2024-2025, with another 9 GW slated for early 2026. However, this growth is uneven: solar generation is expanding rapidly (forecast to grow 92% between 2024-2026), but wind and natural gas additions may not keep pace during peak evening hours when demand is highest and solar generation drops to zero. The risk isn’t that generators cannot be built—it’s that they won’t be built fast enough, or won’t be the right mix for evening peak demand.

Data Centers as the Hidden Driver of Demand Growth

Few consumers realize that data centers—sprawling facilities housing thousands of servers—now represent the fastest-growing driver of electricity demand in the U.S. These facilities run 24/7, consuming power equivalent to entire cities, and they’re increasingly concentrated in ERCOT and PJM regions where land is available and power historically has been abundant and cheap. In PJM’s December 2024 capacity auction, data centers accounted for 40% of the region’s $16.4 billion in total capacity costs—an extraordinary shift in just a few years. This matters for grid stability because data centers operate on predictable 24/7 schedules, unlike air conditioning demand which concentrates during specific evening hours; adding baseline demand from data centers doesn’t relieve peak-hour pressure, it compounds it.

The mismatch between demand and supply is stark. PJM forecasts that data center additions will average 5-7 GW annually through 2032, but the region is only adding 2-3 GW of new power supply annually. This growing gap means that even if utilities meet current targets, they will fall further behind demand growth each year. The problem is most acute in the Mid-Atlantic where land and cooling water availability make data centers attractive; in Texas, similar issues loom as data centers gravitate toward ERCOT’s growing pool of cheap solar power. Unlike a heat wave, which is temporary, data center demand is permanent—once a facility powers on, it draws electricity every single hour for years.

Data Centers as the Hidden Driver of Demand Growth

How Utilities Are Attempting to Expand Capacity

Utilities cannot simply build power plants overnight; construction timelines stretch years, and permitting adds months or years more. Solar generation offers the fastest deployment path: utility-scale solar farms can be built in 1-2 years in favorable conditions, which is why ERCOT’s solar generation is projected to roughly double between 2024-2026. However, solar has a critical limitation—it produces zero electricity during evening peak demand hours. This forces utilities to maintain natural gas plants or battery storage to cover the evening peak, adding cost and complexity.

Natural gas expansion faces similar challenges but different constraints. Building a new natural gas plant takes 3-5 years of planning and construction, and environmental regulations now make this slower and more expensive than historically. Expanding transmission lines—the wires that carry electricity from generators to users—often takes even longer because it requires easements, environmental reviews, and community approval. The irony is that utilities could often solve localized grid stress more quickly by building distribution-level battery storage or upgrading transformers, but investment incentives and regulatory frameworks don’t always reward this approach. The result: utilities are expanding generation and transmission, but often not fast enough to match demand growth, particularly the unexpected surge from data centers.

What Happens When Supply Doesn’t Meet Demand?

When electricity demand exceeds available supply, grid operators invoke load-shedding procedures—deliberately cutting power to some users to prevent a complete blackout affecting everyone. These “rolling blackouts” are temporary outages imposed on rotating neighborhoods or customer classes, intended to reduce total demand enough that the remaining customers can stay powered. However, rolling blackouts are crude tools: they cannot distinguish between someone using air conditioning for comfort versus someone running medical equipment, dialysis machines, or refrigeration for medications. For older adults with health conditions, power interruptions can be dangerous.

Alternatively, grid operators can pay large industrial users to reduce consumption immediately—essentially paying factories or data centers to shut down or reduce operations temporarily. This works in the short term but is expensive and can only be repeated so many times before it becomes economically unsustainable. A third option is purchasing emergency power from neighboring regions at premium prices, which works during normal heat waves but fails if the entire eastern or western U.S. is experiencing heat simultaneously, as happened in summer 2023 when western states also hit record demand. The long-term solution requires new generation, transmission, and often storage, all of which take years to build and billions of dollars to finance.

What Happens When Supply Doesn't Meet Demand?

Renewable Energy’s Growing But Incomplete Role

Non-fossil fuels (wind, solar, and nuclear) proved surprisingly valuable during ERCOT’s June-July 2023 heat wave, contributing as much as 55% of total generation on June 28-29 as evening wind generation kicked in just as solar dropped to zero. This demonstrates that renewable energy can contribute significantly to peak demand if the mix includes evening-friendly sources like wind. However, renewable contribution to peak demand is highly variable: on days with calm evening winds and hot daytime weather, wind provides little help during evening peak hours, forcing reliance on natural gas or battery storage.

Texas’s aggressive solar expansion (projected to increase 92% by 2026) will help during morning and midday peaks but provides minimal evening relief. Smart charging of electric vehicle fleets could theoretically shift demand away from evening peaks if millions of cars charged during midday when solar generation is abundant, but the infrastructure for this—residential and commercial chargers, grid integration—remains limited. Residential rooftop solar is growing, but most customers lack battery storage, so excess midday solar generation they produce doesn’t help them during evening hours when they also run air conditioning.

What the Grid Looks Like in 2026 and Beyond

The trajectory is clear: electricity demand will continue rising faster than new generation comes online, tightening the margin between supply and reliability. In Texas, ERCOT expects to be resource-adequate through 2026 as the 32 GW of generation additions are completed, but 2027 onward becomes uncertain if data center additions accelerate beyond forecasts. In the Mid-Atlantic, PJM is explicitly warning that capacity reserves will become scarce starting June 1, 2026, with potential shortages thereafter.

Both regions are investing heavily—new transmission lines, additional solar, new natural gas plants, and battery storage facilities—but none of this occurs fast enough to fully close the demand-supply gap. Longer term, the grid’s trajectory depends on policy decisions made today: whether governments mandate demand-side management (widespread smart thermostats that reduce peak-hour cooling), whether data center growth is concentrated in high-capacity regions or distributed to understressed areas, and whether investments in transmission and storage keep pace with demand growth. Infrastructure built in the 1990s and 2000s was designed for demand patterns that no longer exist. Modernizing the grid takes decades and trillions of dollars, and delay only increases both the cost and the risk of supply failures during future heat waves.

Conclusion

Power grids across Texas and the Mid-Atlantic are struggling under peak demand during heat waves, with record-breaking hourly usage becoming commonplace rather than exceptional. The core issue is that utilities are adding generation capacity faster than ever before, but growth is still insufficient because demand is also accelerating faster than historical norms. Data centers, which run continuously and concentrate in energy-rich regions, have become a hidden major driver of demand growth; in some areas, data centers now account for 40% or more of new power demand.

The immediate outlook through 2026 remains manageable in most regions, but warnings are mounting: PJM explicitly forecasts capacity shortages starting summer 2026, and ERCOT’s margin of safety is shrinking as data center megafacilities continue connecting to the grid. For households with older adults or chronically ill residents who depend on continuous power for medical equipment or for maintaining safe temperatures during extreme heat, understanding these grid pressures matters. Building resilience—backup power sources, improved home insulation, and community awareness of outage risks—becomes increasingly important as peak demand events grow more frequent and severe.