Extreme-heat grid performance is increasingly separating solar-and-storage regions from capacity-constrained markets as data center load accelerates.

Extreme heat is impacting grid performance differently across regions, highlighting a structural divide between areas with robust solar-and-storage infrastructure and those facing capacity constraints. As data center demand accelerates, markets like PJM Interconnection, the Southwest Power Pool (SPP), and the Electric Reliability Council of Texas (ERCOT) are grappling with heat-driven demand spikes and the challenge of maintaining price stability. This shift is prompting urgent procurement discussions centered on solar-plus-battery penetration.

The operational contrast became stark during recent summer heat events. According to E&E News, PJM’s early-July heat wave and Texas’ mid-July heat wave presented the same fundamental problem: record air-conditioning load. Yet the outcomes diverged sharply. PJM saw prices spike above $2,000 per megawatt-hour around its July 2 peak, while ERCOT hit a July 22 demand record without emergency actions and with real-time prices around $29/MWh. For data center operators and the utilities that serve them, that spread defines the current landscape. The AI infrastructure boom is arriving in a climate regime that is already producing more frequent, sharper summer peaks. The practical question is no longer whether grids can “handle” large loads in the abstract, but which markets can deliver predictable capacity and predictable pricing during the same late-day hours when both people and servers are pulling hard.

E&E News, reporting by Benjamin Storrow, described three U.S. regions where extreme heat pushed demand to historical highs: PJM (mid-Atlantic), SPP (Great Plains), and ERCOT (Texas). It also noted a key wrinkle for large-load planners: PJM and Texas sit at the center of the country’s data center buildout, and data centers are a small but growing portion of load in SPP. GridStatus.io analyst Abby Lestina told E&E News that large loads are coming online and that heat is making the impact easier to see. That framing matters for enterprise operators because it ties two planning timelines together: multi-year interconnection and construction schedules for new campuses, and the immediate reality that heat-driven peaks can show up any summer.

If a region is using demand response to avoid breaking records, that demand response is already part of the “power availability” spec for any new 50 to 200 MW load. PJM said that demand would have broken a 20-year-old record if not for conservation programs that reduced load, according to E&E News. For operators writing requirements for colocation expansions or self-built campuses, that’s a signal that “availability” increasingly includes curtailment pathways: contractual demand response, utility-managed conservation programs, or on-site load shedding that can be triggered on short notice.

The clean-energy narrative around heat events often collapses into a simplistic debate about whether solar can perform after dark. E&E News reported that a U.S. Department of Energy social media post argued that solar did not deliver power after dark during PJM’s peak, prompting pushback from clean-energy advocates and coverage elsewhere. But the more useful operational comparison in E&E News’ reporting is intra-day timing and controllability. When ERCOT demand peaked around 6 p.m. on July 22, solar was generating roughly a third of the grid’s power, according to E&E News. The story also credited Texas’ large solar and battery buildout as the key difference versus PJM and SPP, which it described as having less solar and storage.

For buyers, “solar share at peak” is an actionable metric because it points to what should be validated in a feasibility study: the local net-load curve, the late-afternoon ramp rate, and the amount of storage that can be dispatched through the top 10 system peaks. If the peak is shifting later in the evening as heat lingers, solar alone changes energy economics but may not change capacity risk unless storage is in the stack.

Broader industry milestones sometimes obscure localized grid realities. A National Wildlife Federation Facebook post pointed to a separate, upbeat milestone: globally, solar and wind have together produced more electricity than coal for the first time. Even as the post did not provide the underlying dataset, the claim reflects a widely discussed direction of travel in the energy transition. What the U.S. heat-week experience adds, per E&E News, is that the enterprise consequence shows up at the interconnection and tariff level. A global generation milestone doesn’t tell a hyperscaler whether a specific utility territory can firm up 100 MW of new load without forcing years of network upgrades, or whether a market will expose that load to scarcity pricing during the hottest 20 hours of the year. The clean-energy transition is increasingly judged by peak-hour performance and price volatility, because that’s what large-load operators pay for.

E&E News also reported that Dominion Energy executives told investors that nine of the utility’s top 10 power-consumption days have occurred this year, with eight of its highest summer peaks occurring in just over two months. Virginia hosts the world’s largest concentration of data centers, and Dominion’s territory is a bellwether for what happens when the largest load cluster in the country meets sharper weather extremes. Southwest Power Pool’s experience points to a broader footprint issue. E&E News cited preliminary U.S. Energy Information Administration data showing SPP demand reached nearly 58 gigawatts on July 27, the highest level recorded in EIA’s preliminary power data. SPP also had to implement emergency measures across parts of the Great Plains, E&E News reported. That matters for operators considering “second-tier” markets for cost reasons: lower land and fiber costs don’t automatically translate into lower power-system risk.

Ultimately, grid resilience during extreme heat has evolved from an abstract engineering challenge into a direct determinant of commercial viability for AI infrastructure. Markets that integrate solar, storage, and flexible demand response are demonstrating the predictability required to support accelerating data center growth, while capacity-constrained regions face mounting pressure to modernize procurement standards and grid architecture.