Six consecutive ERCOT demand records point to a structural gas call that front-month prices underreflect
Six straight weeks above the 2023 record in Texas reveal a structural load story that gas markets have not finished pricing.
Hourly load across ERCOT averaged 74.5 GW for the week ending August 22 (2026), a record for any week in the grid's history, the U.S. Energy Information Administration reported on September 3 (2026). That followed ERCOT's new all-time hourly peak of 91.089 GW, set on July 22 (2026), which surpassed the previous record of 85.508 GW, set on August 10, 2023, by more than 6%.4,1
Six consecutive weeks, from the week ending July 25 (2026) through the week ending August 29 (2026), came in above that 2023 record, EIA said. The week ending August 29 (2026) averaged 73.7 GW, just off the weekly peak but still far above anything ERCOT had sustained before this summer. This is not one anomalous day. Six weeks of demand has rewritten the baseline.3,4
Gas traders have responded to the heat. On August 19 (2026), NYMEX Henry Hub front-month futures surged to $2.837, up 2.20% on that session, after updated forecasts extended heat across the South and West through late August and into early September. Buyers broke through resistance at $2.790. On Tuesday (2026-09-08), NYMEX Henry Hub front-month was trading at $2.88/MMBtu, up 0.35% on the day. That is not a distressed level given what the physical load data shows.2
But the gas-demand story embedded in this summer's ERCOT data is more layered than a straight heat trade. Natural gas supplied 48% of ERCOT's power at the July 22 (2026) peak. Solar supplied 32%. Doug Lewin, Google's Texas lead for energy market development, made the comparison explicit: when PJM neared its own peak demand record it declared an energy emergency, while ERCOT did not. His explanation was solar. Texas solar resources had reached a deployment record of 35 GW on that day, Lewin noted.4
That is a meaningful suppressor of gas burn. Texas has around 53.5 GW of solar capacity installed, according to the Solar Energy Industries Association. At the July 22 (2026) peak of 91 GW, natural gas generation was running near 43.7 GW in absolute terms — substantial, but not at the system's limit, partly because solar was simultaneously carrying close to 29 GW. Strip out solar at that scale and ERCOT would have needed more gas, or would have been closer to its operational edge.4
ERCOT projected peaks above 90 GW every day during the week of September 7 (2026), with temperatures expected to remain consistently above 100 degrees Fahrenheit, and anticipated no shortages. The confidence reflects installed capacity, not comfort. At that level of sustained peak demand, any shortfall in solar output, whether from cloud cover or a shift in load timing, translates directly into incremental gas dispatch.4
The longer view makes the gas question harder to dismiss. ERCOT's own long-term load forecast projects peak loads reaching 154 GW by 2035, roughly 70% above this summer's record. That volume of residual load, the demand solar and wind cannot serve, falls on gas unless battery storage, demand response, or new generation scales in parallel. This summer's record is not a ceiling.4
NYMEX Henry Hub front-month has recovered from its August lows but remains at levels that price Texas power demand as a seasonal event. ERCOT weekly load data for the week of September 7 (2026) will show whether averages hold above 70 GW as temperatures stay elevated. If they do, and if solar output softens through the autumn shoulder season, the case for structurally higher gas dispatch from Texas becomes harder to set aside.3,42