Record Demand in PJM and ERCOT, July 2026: What Drove It and What It Did to Prices

Noreva, an energy market data and analytics provider (formerly Karbone Research), tracks how demand records translate into price outcomes across the full hourly curve, beyond the single peak-to-peak comparison. PJM and ERCOT both broke all-time demand records in July 2026, three weeks apart, and the two grids answered with close to opposite price responses. PJM ran nine hours above $500/MWh, peaking at $1,868.67/MWh; ERCOT’s record hour cleared at $47.75/MWh and did not reach its evening high of $208.85/MWh until four hours later. Same summer, same category of record, two different price stories – and the gap traces to what each grid still had left to dispatch once demand peaked.

This piece walks through both record days: what drove the demand, whether the price response matched expectations, and how far batteries and the still-present solar carried ERCOT’s evening. For PJM’s longer-run load trajectory and the auctions it feeds, see Noreva’s companion piece on the 2026 PJM load forecast.

What Happened on the Record Days in PJM and ERCOT?

PJM’s all-time peak came first. On July 2, 2026, PJM served an estimated 168,158 MW between 5 and 6 p.m., breaking the prior record of 165,563 MW set in August 2006 (PJM Inside Lines, July 10, 2026). Noreva’s hourly grid dataset for the same window shows average demand of 162,648 MW at hour ending 18:00 (HE18) – lower than PJM’s figure because PJM’s number is an instantaneous metered peak, while Noreva’s series reports the hourly average, which always sits below the peak instant inside it (Noreva hourly grid data, July 2026). Both describe the same event.

Three weeks later, ERCOT set its own all-time record: 91,075 MW at HE18 on July 22, 2026, which the U.S. Energy Information Administration reported as “exceeding 91 GW,” 6 percent above the prior record of 85.5 GW set August 10, 2023 (EIA, Today in Energy, August 3, 2026). ERCOT’s demand also ran ahead of the dataset’s day-ahead forecast in 13 of the day’s 24 hours, by as much as 2,350 MW at HE18 itself (Noreva hourly grid data, July 2026) – a modest miss for a 91 GW system, but a miss nonetheless.

The table below lines up both peak hours on the metrics that shaped price: gas and solar output, batteries, and what each looked like three hours later.

July 2026 Record Days: PJM vs. ERCOT at the Demand Peak (Noreva hourly grid data, July 2026, unless noted)

What Drove Record Demand in Both Grids?

PJM’s own account ties its record to an extreme, prolonged heat event across the RTO. The RTO posted its hottest-ever average temperature, 97 degrees, on July 2, with much of the Mid-Atlantic in the low 100s (PJM Inside Lines, July 10, 2026). Forced generation outages ran 18,100 to 19,400 MW from July 2 through 4, well above the roughly 12,800 MW average PJM has seen on its top ten summer peak days over the prior three years (PJM Inside Lines, July 10, 2026). PJM leaned on DOE 202(c) emergency orders, permit relief for about 3.25 GW of generation, and demand response – roughly 6,113 MW on July 2 – to serve the load reliably (PJM Inside Lines, July 10, 2026). The peak landed just under PJM’s own 90/10 seasonal forecast of 169,128 MW: a low-probability case the system was already sized for. Structural, data-center-driven load growth is the multi-year backdrop; Noreva’s PJM load forecast coverage covers that trajectory in full.

ERCOT’s record arrived the same way: a heat wave pushing demand past prior highs in a region where both consumption and generation keep growing (EIA, Today in Energy, August 3, 2026). The same EIA analysis notes a regional record followed five days later – the Southwest Power Pool’s own all-time peak of 57.9 GW on July 27, 2026 – and concludes that heat-driven records are likely to continue in growing-load regions. July 2026 was a multi-grid heat event, and each grid’s resource mix determined how it showed up in price.

Was the Price Impact as Expected, Worse Than Expected, or Better Than Expected?

In PJM, the price response matched what a thermal-heavy, capacity-constrained system does near its planning ceiling. Real-time RTO prices climbed from late morning – $656/MWh at HE12, $882/MWh at HE17 – then broke into scarcity territory as demand peaked: $1,621/MWh at HE18, a day maximum of $1,868.67/MWh at HE19, staying above $1,200/MWh through HE21 (Noreva hourly grid data, July 2026). Nine of 24 hours cleared above $500/MWh, and the congestion component of price stayed small throughout, peaking at just $2.88/MWh at HE18 – a system-wide scarcity signal, not a localized transmission problem. The pattern did not scale cleanly with the peak megawatt figure across the wider heat event: July 1’s lower peak (161,965 MW) averaged $191/MWh with 2 hours above $500, while July 3’s still lower peak (155,028 MW) averaged $360/MWh with 5 hours above $500 (Noreva hourly grid data, July 2026). Price severity tracked heat persistence and forced-outage levels as much as the peak MW itself: July 3 cleared higher on average than July 1 despite a lower peak, in a grid where gas sets the marginal price across most PJM peak hours (Noreva, PJM LMP Merchant Curve coverage).

ERCOT’s response looked different, and arguably milder than a bare “record demand” headline suggests. The record hour, HE18, cleared at just $47.75/MWh – close to the $25-38/MWh midday range – even as demand hit an all-time high (Noreva hourly grid data, July 2026). Price only moved once the sun went down: $95.38/MWh at HE19, $136/MWh at HE20, $152/MWh at HE21, and a day maximum of $208.85/MWh at HE22, four hours after the demand peak, with demand already down to 81,593 MW. No hour across the three-day window touched $500/MWh. That lag fits how ERCOT’s scarcity mechanism works: the Operating Reserve Demand Curve adds price as reserves compress, and reserves in ERCOT compress in the evening ramp, once solar has faded, well after the demand peak itself (Noreva, ERCOT price spikes coverage).

How Much Did Batteries and Behind-the-Meter Generation Limit ERCOT’s Price Response?

Batteries explain a meaningful share of why ERCOT’s evening ramp stayed under $209/MWh instead of tracking toward PJM-style scarcity. Solar fell from a midday peak of 33,484 MW at HE13 to just 1,001 MW at HE21, a drop of 32,483 MW in eight hours (Noreva hourly grid data, July 2026). ERCOT’s battery fleet, which had charged as much as 10,002 MW at HE10 and was still net charging 1,394 MW at HE13, reversed into discharge as solar rolled off: 748 MW at HE19, 7,084 MW at HE20, and a peak of 11,316 MW at HE21, tapering to 7,236 MW at HE22, 1,879 MW at HE23 and 37 MW at HE24 (Noreva hourly grid data, July 2026). Gas also climbed through the same window, from 33,566 MW at HE10 to 43,172 MW at the HE18 demand peak and a day maximum of 47,031 MW at HE21, while wind rose from 1,635 MW at HE14 to 12,899 MW by HE24 (Noreva hourly grid data, July 2026). Batteries were one contributor among several, but the one built specifically to bridge the hours where solar disappears fastest.

ERCOT’s Evening Ramp on July 22, 2026: Solar Decline, Battery Response, Price (Noreva hourly grid data, July 2026)

What would the evening have looked like without that battery ramp? The dataset does not support a specific counterfactual price, and none is offered here. Directionally: 11,316 MW of discharge at HE21 landed in the exact hour solar bottomed out, so that block would otherwise have leaned on gas already climbing toward its day maximum, on a thinner reserve margin, or on a higher ORDC adder – all pointing toward a steeper price curve than the one that actually cleared. Noreva’s ERCOT coverage frames this directly: the battery fleet creates wide charge/discharge spreads because it is built to buy cheap midday solar and sell into the evening scarcity window (Noreva, ERCOT LMP coverage).

PJM’s dataset carries no battery column, and this piece makes no PJM battery claim it cannot show: PJM’s storage fleet is small relative to ERCOT’s, and the dataset does not break it out (Noreva hourly grid data, July 2026). On behind-the-meter generation, PJM’s own account credits demand response with roughly 6,113 MW of the July 2 peak (PJM Inside Lines, July 10, 2026). Behind-the-meter gas generation tied to new data-center campuses (see Noreva’s coverage of how data-center builders are solving the power problem) is a documented structural trend, but neither ISO reports it hour by hour, so its contribution to these two particular hours sits outside what any published dataset can show.

What Does This Mean for Capacity Accreditation and the Next Auctions?

PJM’s scarcity event happened inside a capacity market already pricing this tightness: three consecutive Base Residual Auctions (2026/2027, 2027/2028 and 2028/2029) cleared at their FERC-approved caps (PJM results releases of July 22, 2025, December 17, 2025 and July 14, 2026), a run covered in Noreva’s PJM capacity coverage. A nine-hour, $1,600-plus event reinforces why the market has paid up for firm capacity; it does not suggest the auctions mispriced the risk. The question this event raises is accreditation: how much of the capacity that cleared actually performs the way it was credited to perform on a day like July 2 is the argument in Noreva’s ELCC Crunch piece, where solar’s accredited value has compressed to 7 to 8 percent of nameplate under PJM’s 2027/2028 ELCC class ratings (fixed-tilt 7 percent, tracking 8 percent).

ERCOT has no capacity market to clear, which makes July 22 a different kind of evidence: a live demonstration of what the battery fleet is worth in the hours that matter, just as battery accreditation is under review. American Clean Power Association guidance pushing operators toward more conservative state-of-charge and discharge-rate limits can reduce a project’s accredited capacity even as its real-world performance, as seen July 22, looks strong (Noreva, battery standards coverage). That is the gap to underwrite: what a battery discharged on the day it mattered versus what it gets credited for. Noreva’s broader treatment of firm-capacity accounting across PJM, MISO, NYISO and ISO-NE is in UCAP, ELCC, Firm Capacity: What Each ISO Pays For in 2026, with auction prep in How to Prepare for a Capacity Auction.

How Does Noreva Turn the Evening Ramp Into a Forecast?

Both July 2026 records make the same point Noreva’s methodology is built around: the hour that sets the demand record and the hour that sets the price are often two different hours. Noreva’s PJM LMP curves are built on dispatch modeling because gas sets the marginal cost in more than 60 percent of PJM’s peak hours (Noreva, PJM LMP Merchant Curve coverage) – exactly the mechanism behind the $1,868.67/MWh hour on July 2. Noreva’s ERCOT curves separate structural trends from event-driven pricing using a scarcity framework calibrated to historical event frequency, including battery charge-and-discharge economics. Both sit inside Noreva’s Power hub across all seven US ISOs, alongside the Capacity hub. Noreva’s merchant curves refresh monthly as load, generation mix and accreditation assumptions move.

Frequently Asked Questions

What drove the record demand in PJM and ERCOT in July 2026, and how did it affect prices?

Heat drove both records. PJM’s July 2 peak of 168,158 MW came during its hottest-ever average RTO temperature, 97 degrees (PJM Inside Lines, July 10, 2026); ERCOT’s July 22 peak of roughly 91 GW came during a regional heat wave that also produced a Southwest Power Pool record five days later (EIA, Today in Energy, August 3, 2026). The price impact differed sharply: PJM cleared 9 hours above $500/MWh at a $1,868.67/MWh maximum; ERCOT stayed under $209/MWh throughout (Noreva hourly grid data, July 2026).

Was the price impact of the July 2026 demand records as expected, worse than expected, or better than expected?

In PJM, the outcome matched what a thermal-heavy, near-ceiling system predicts: price severity tracked heat persistence and forced-outage levels more than the peak megawatt figure, and a near-zero congestion component confirmed the event as system-wide scarcity rather than a local one. In ERCOT, the outcome was milder than the “record demand” headline suggests: the record hour cleared near $48/MWh, and the price peak arrived four hours later, after solar faded – ERCOT’s scarcity pricing is timed to reserve compression, not the demand peak.

How much did batteries and behind-the-meter generation limit the price impact of record demand?

In ERCOT, batteries discharged a peak 11,316 MW at HE21, the same hour solar collapsed to 1,001 MW from a 33,484 MW midday peak – directly offsetting a meaningful share of the solar cliff (Noreva hourly grid data, July 2026). PJM’s dataset does not break out battery output, and its storage fleet is small relative to ERCOT’s; PJM’s own account credits roughly 6,113 MW of demand response with helping serve the July 2 peak (PJM Inside Lines, July 10, 2026). Behind-the-meter output is not reported hourly by either ISO, so its effect on either event cannot be read from published data.

Did PJM or ERCOT set an all-time demand record in July 2026?

Both did. PJM served an estimated 168,158 MW on July 2, 2026, above its 2006 record of 165,563 MW (PJM Inside Lines, July 10, 2026). ERCOT reached roughly 91 GW on July 22, 2026, 6 percent above its prior record of 85.5 GW (EIA, Today in Energy, August 3, 2026).

How does Noreva forecast PJM and ERCOT power prices around scarcity events?

Noreva’s PJM LMP curves model dispatch economics and nodal basis because gas sets the marginal price in most PJM peak hours. Noreva’s ERCOT curves layer a fundamental dispatch model with a scarcity-event framework, including battery economics. Both refresh monthly as part of Noreva’s merchant curve coverage. Book a demo to see the current PJM and ERCOT curves.

Methodology note: figures cited as “Noreva hourly grid data, July 2026” come from an hourly-resolution dataset covering PJM RTO (July 1-3, 2026) and ERCOT (July 21-23, 2026): average hourly demand, generation by fuel type, battery position (ERCOT only), net interchange, real-time price, and the dataset’s day-ahead demand forecast. Because the series reports hourly averages, its demand sits at or below the instantaneous peaks PJM and EIA report separately.

Price the Next Record Day With Noreva

July 2026 showed two grids answering the same kind of record with two different price outcomes – the difference came down to hourly resource mix more than to the peak megawatt figure itself. Noreva’s PJM and ERCOT curves are built on that hourly reality: dispatch, scarcity pricing, and the transaction data behind them.

Book a demo to see Noreva’s current PJM and ERCOT price forecasts.