Gas Printed Its Tightest Build of the Month and Still Could Not Hold $2.80

September NYMEX natural gas opened at $2.778/MMBtu and traded lower after the storage print, threatening to snap a two-day advance that had carried the prompt month to its highest level in nearly four weeks.

The Energy Information Administration reported a 16 Bcf injection into storage for the week ending August 14, nearly matching a consensus that had clustered between 14 and 15 Bcf. The five-year average build for that calendar week is 29 Bcf. That makes the print 13 Bcf tighter than seasonal norm, and it moved nothing.

Working gas now stands near 3,169 Bcf. The prior week's report showed 3,153 Bcf as of August 7 — 25 Bcf below the same period last year and 198 Bcf, or 6.7%, above the five-year average of 2,955 Bcf. A build 13 Bcf under the seasonal average narrows that surplus to roughly 185 Bcf.

Narrowing a 198 Bcf surplus by 13 Bcf per week takes fourteen weeks to erase. The injection season ends in ten.

The composition of the build was the interesting part. A hefty South Central withdrawal limited the overall figure, which means the headline number understates how loose the balance was elsewhere. Salt-dome storage drawing during a Gulf Coast heat event is a regional demand signal, not a national tightening.

Price action tells the story better than the data. September closed at $2.690 on Monday after falling 2% to around $2.68 — the lowest since August 7. Tuesday it broke near-term technical resistance. Wednesday it rallied above $2.80 to its highest since July 24 on hotter forecasts, with Houston temperatures projected near 100°F from August 20 through 23. Thursday it opened at $2.778 and immediately gave ground, trading back below Wednesday's low.

The rally ran into a wall at exactly the level where every summer bounce has stalled.

The reason is supply. Lower-48 dry gas production hit 112.0 Bcf/d on Wednesday, up 2.7% year over year, with the August average running 111.6 Bcf/d against July's record monthly pace of 110.7 Bcf/d.

Production at 112 Bcf/d Is Absorbing Everything

The supply side has not flinched once during this entire summer and that is the single most important fact in the market.

Lower-48 output averaged 111.6 Bcf/d through August, exceeding July's monthly record of 110.7 Bcf/d, with Wednesday's print at 112.0 Bcf/d representing a 2.7% year-over-year increase. Producers have not adjusted to lower prices. Shale wells keep producing, and associated gas from Permian oil drilling keeps arriving regardless of what Henry Hub does.

That last point is the structural problem and it is worsening. With WTI at $86.40 and Brent above $93 on the Iran escalation, Permian oil economics are outstanding. Every incremental oil well drilled for a $86 barrel produces associated gas that reaches the market whether the gas price is $2.68 or $4.68. Oil-directed drilling is a gas supply engine that operates entirely independent of gas prices.

The takeaway constraint that used to bottle that gas in the basin is being removed. The Hugh Brinson pipeline reaches full capacity at 1.5 Bcf/d by September 1 — Permian gas heading straight to Henry Hub precisely as summer cooling demand fades into the shoulder season.

Permian takeaway is reaching the point where the market can clear, which unlocks more oil production and the associated gas that comes with it. That is a supply flywheel, and it turns faster when crude is at $86.

Run the arithmetic against demand. Gas flows to the nine major LNG export facilities averaged 17.3 Bcf/d in August against 17.2 Bcf/d in July — a 0.1 Bcf/d increase. Production rose roughly 0.9 Bcf/d over the same stretch.

Supply added nine times what export demand added.

Power burn should be the offset and it is underperforming. Total Lower 48 power generation came within 0.1 GW of its all-time weekly high during the latest storage week, yet gas-fired generation sat about 2% below its own record because stronger wind and solar output displaced it. Solar generation rose 21% and wind 6% in the first half of 2026.

Record power demand producing below-record gas burn is the clearest statement available that renewables are eating the marginal load.

The EIA Cut Its Third-Quarter Forecast by Fifty Cents

The agency's own revision is the most damning input for the bull case and it arrived nine days ago.

The August Short-Term Energy Outlook forecasts the Henry Hub spot price to average around $2.87/MMBtu in the third quarter of 2026 — down 50 cents compared with the prior month's forecast. The stated reasons: reduced LNG feedgas demand and record natural gas production, which the agency expects will leave inventories at their highest level heading into winter since 2016.

The revision path is brutal when laid out in sequence. The July STEO had projected 3Q26 at $3.37/MMBtu with 4Q26 at $3.57. The August edition cut the third quarter to $2.87 and now expects the spot price to remain below $3.00/MMBtu until November, averaging $3.03/MMBtu over the remaining five months of the year — nearly 50 cents lower than the prior forecast.

Full-year expectations moved with it. July's STEO had 2026 averaging $3.67/MMBtu and 2027 at $3.49. Both of those numbers now look stale against a third quarter running under $2.90.

For context, Henry Hub averaged $3.53/MMBtu in 2025 and $7.72/MMBtu in January 2026 alone, when Winter Storm Fern drove the largest weekly net withdrawal in the history of the weekly storage report.

The inventory projection is the anchor. The agency expects natural gas inventories to reach a record 3,985 Bcf at the end of October 2026 — an increase of 19 Bcf versus the July STEO and 5% above the five-year average, per the EIA's August Short-Term Energy Outlook.

A record end-of-October inventory strips away any winter-scarcity premium bulls might otherwise lean on. The futures curve agrees: contracts through September 2026 remain below $3.00/MMBtu.

The next STEO lands September 9. On current trajectory, the revision risk is to the downside again.

LNG Is the Demand Story That Keeps Underdelivering

Export demand is supposed to be the structural floor under this market and it has not shown up at the scale required.

The EIA cut its third-quarter LNG export forecast to 16.5 Bcf/d, down 0.2 Bcf/d from the prior month. Actual feedgas flows to the nine major facilities have run 17.3 Bcf/d in August against 17.2 Bcf/d in July, with some measures near 18 Bcf/d.

Maintenance at Freeport LNG began July 10 and is expected to complete in late August, affecting 2.0 Bcf/d of nominal export capacity. That is the single largest identifiable demand outage in the market and its return is the clearest near-term bullish catalyst on the calendar.

The problem is what happens after. Even with Freeport fully operational, exports would remain limited due to slow growth in additional export capacity despite US price spreads to Europe and Asia remaining elevated. The bottleneck is liquefaction trains, not demand.

International pricing confirms the arbitrage is wide open. LNG vessel traffic through the Strait of Hormuz slowed considerably after strikes on vessels resumed on July 7, pushing international prices to levels last reached in early April. European gas prices have soared on Middle East supply shortages, keeping upside inflation risks live across the eurozone.

A world short of gas and an American market at $2.78 is a transportation problem, not a pricing signal.

Struggling feedgas demand on the US Gulf Coast has also been mirrored in Canada and Mexico simultaneously, which points to a systemic capacity issue rather than a US-specific one.

The forward pipeline is substantial. Venture Global's proposed CP2 expansion has moved into federal environmental review, potentially adding 11.7 million tons per year of peak liquefaction and nearly 1.9 Bcf/d of transportation capacity. Cheniere raised its full-year outlook on strong export demand.

None of that reaches the physical market before 2027. The domestic market does not care yet because the gas is not leaving fast enough, and until it does, 112 Bcf/d of production has nowhere to go but into storage.

Four Bullish Prints and the Contract Still Cannot Hold $2.80

The storage series has been running tighter than seasonal norms for weeks and price has not responded. That divergence is the tell.

The July 24 week delivered a 28 Bcf injection against a 33 to 38 Bcf consensus range — a bullish surprise that extended a trend of prints coming in below expectations. Working gas stood at 3,084 Bcf, 32 Bcf below year-ago levels and 185 Bcf above the five-year average of 2,899 Bcf. The September contract sat at $2.77 that morning.

The August 7 week produced 35 to 36 Bcf against a 33 Bcf estimate, taking inventories to 3,153 Bcf and widening the surplus to 198 Bcf. The August 14 week just delivered 16 Bcf against a 14 to 15 Bcf consensus and a 29 Bcf five-year average.

Three of the last four prints have come in at or below the five-year seasonal build. Across the same stretch the September contract has gone from $2.77 to $2.690 to $2.778.

Net movement in a month: roughly one cent.

That is the definition of a market where the supply overhang overwhelms the flow data. One moderate storage number can force shorts to cover for a session. It does not fix a surplus built over months. Buyers need repeated tight builds paired with hotter weather and rising feedgas, and they have received the first without enough of the other two.

The July 30 pattern is instructive. Futures settled higher after a smaller-than-expected injection, then edged lower the following session as traders looked past it and refocused on elevated production, softer LNG export demand and a mixed weather outlook.

Same script, three weeks later.

Wednesday's rally to the highest level in nearly four weeks stalled short of the key moving average overhead, buyers took the trend change but could not hold it through the overnight session, and the market opened Thursday back below Wednesday's low on the weak side of the resistance zone that was supposed to become support.

Failed breakouts at the same level repeatedly are distribution, not accumulation.

The Weather Trade Is Real and It Is Running Out of Calendar

Heat is the one genuinely supportive input right now and its shelf life is measured in weeks.

Short-term temperature outlooks turned more bullish, with Houston temperatures expected to average around 100°F from August 20 through 23 and warmer-than-normal conditions forecast through early September. That drove Wednesday's move above $2.80 to the highest print since July 24.

The problem is that the calendar is working against it. Cooling degree days peak in mid-August and decline sharply through September. Even sustained above-normal temperatures produce less gas burn each week as the sun angle drops.

Traders are already turning attention toward a shoulder season that could leave balances even looser once cooling demand fades, with production hovering near record levels and LNG demand still below what may be needed to absorb growing supply.

That is the setup. Peak demand is now, supply is at a record, and the surplus is still 185 Bcf.

Weather models have moderated repeatedly this month. On August 17 milder forecasts and expectations for larger storage injections drove a 2% decline to around $2.68, the lowest since August 7, as models pointed to less intense heat across much of the country.

The West and South being hot is not the trade. The market needs sustained national heat, and it needs it paired with something structural.

The hydro offset deserves attention as a genuine bullish input. Hydropower generation rose 9% in the first half of 2026 but is forecast to fall 3% in the second half because of intensifying drought conditions in the western United States. Less hydro means more thermal generation at the margin.

Against that, new capacity keeps arriving on the renewable side, including the 3.7 gigawatt SunZia wind farm, with renewable generation expected to keep growing at similar rates through 2027.

The generation stack is structurally displacing gas faster than drought is restoring it.

What the Winter Strip Is Actually Pricing

The front month is the wrong place to express a view in this market. The curve is where the disagreement lives.

The EIA expects the Henry Hub spot price to remain below $3.00/MMBtu until November and average $3.03/MMBtu across the final five months of the year. That is a forecast of essentially no seasonal ramp until the calendar forces one.

Against that, the base-case scenario framework most widely used assigns roughly 55% probability to a $2.50 to $4.50/MMBtu range, with storage ending the injection season 7% above average, third-quarter prices hovering near $2.80 to $3.00, and a fourth-quarter ramp toward $4.00 to $4.50 as LNG feedgas demand peaks and heating demand returns.

The gap between $2.78 spot and a $4.00 to $4.50 fourth-quarter expectation is the entire trade.

The cold-winter case carries roughly 25% probability at $5.00 to $8.00/MMBtu. That scenario requires an early or severe winter drawing storage below the five-year average by November, which triggers the same dynamic that pushed January 2026 to $7.72/MMBtu — with LNG export demand above 17 Bcf/d leaving little margin for error. Morgan Stanley's $5/MMBtu structural target sits in that band.

The warm-winter case carries roughly 20% probability at $2.00 to $2.80/MMBtu, where a mild winter leaves storage at or above the five-year average into spring 2027 and suppresses any fourth-quarter recovery.

At $2.778, September is trading inside the bear case.

The asymmetry is what makes the winter strip interesting. Downside from here is bounded by production economics — a break and sustained close below $3 opens the path toward $2/MMBtu, a level most consider unsustainable given LNG export dynamics but possible in a severe warm-winter scenario. Upside is unbounded because January 2026 already proved $7.72 is reachable when storage cracks.

Buying the front month at $2.78 is a bet on weather in the next three weeks. Buying the winter is a bet on a distribution where the tail is fat in one direction only.

The Structural Floor Has Moved and That Matters

The long-term picture is genuinely different from the summer picture and it deserves separating.

US LNG exports at 17 Bcf/d and rising mean sub-$2/MMBtu prices are an increasingly historical anomaly. The structural floor under Henry Hub has risen permanently because a meaningful share of American production now competes against European and Asian pricing rather than against domestic alternatives alone.

The long-term forecasts reflect it. Henry Hub is projected at $3.80/MMBtu by 2030, $4.20/MMBtu by 2040 and $4.95/MMBtu by 2050, with LNG export growth past 20 Bcf/d offsetting pressure from renewables while production rises modestly. Independent modeling runs higher, at $5.40/MMBtu in 2030 and $6.35/MMBtu by 2040, driven by sustained Asian and European LNG demand plus US data centre power needs outpacing supply growth.

The data centre load is the newest variable and the least well modeled. Electricity generation growth is being driven primarily by increasing demand from large customers including data centres, and that load is baseload rather than peaking — it runs at 90%-plus utilization regardless of season or weather.

A persistent new demand floor is exactly what a market carrying a 185 Bcf surplus needs, and it arrives incrementally rather than in a step.

The EIA's own longer view supports the direction. Natural gas-fired electricity generation increased 2% in the first half of 2026 and is forecast to increase further in 2027 as prices remain relatively low, while coal generation continues declining on the shift toward lower-cost gas.

Cheap gas destroys coal. Coal destruction creates permanent gas demand. That is a self-reinforcing loop that operates at $2.78 and stops operating at $5.00.

The key long-term risk is the pace at which renewables displace gas in the US power mix, which most mainstream scenarios project as gradual rather than abrupt through 2035.

Nothing in that framework helps a September contract that cannot hold $2.80.

Levels, Targets and What Kills the Setup

Three scenarios with defined triggers.

The bull path requires a daily close above $2.80 followed by a clean break of the moving average that capped Wednesday's rally. That level has rejected every summer bounce and converting it opens $2.90, then $3.00 — the psychological line and the level the EIA does not expect to be reclaimed until November. Above $3.00 the next reference is the $3.176 area where the August contract traded in early July. The catalyst that delivers it: Freeport's 2.0 Bcf/d returning in late August, a genuinely hot September, and two more sub-20 Bcf builds.

The base case is range work between $2.68 and $2.85. The market chops through the shoulder season with production at 112 Bcf/d absorbing every heat event, the surplus narrowing slowly from 185 Bcf without disappearing, and the prompt month trading around the EIA's $2.87 third-quarter forecast. Base-case band into the September contract expiry: $2.65 to $2.90.

The bear case triggers on a daily close below $2.68, Monday's low and the lowest print since August 7. Losing it opens $2.60, then $2.50 as the base-case scenario floor. A sustained close below the psychological levels beneath that opens the path toward $2.00, which would require a materially warm end to the injection season plus the Hugh Brinson pipeline delivering its full 1.5 Bcf/d into a market that cannot absorb it.

The dates that matter: September 1 for Hugh Brinson reaching full capacity, late August for Freeport maintenance completion, September 9 for the next STEO, and the weekly storage print every Thursday at 14:30 GMT.

Watch the surplus rather than the price. At 185 Bcf above the five-year average with ten weeks of injection season remaining, builds need to run roughly 19 Bcf per week below seasonal norms just to reach October at the five-year average. The last four weeks have averaged closer to 8 Bcf below.

The math does not close in time.

The Verdict: Sell the Rallies, Own the Winter

September natural gas at $2.778 is priced correctly for a market where every bullish input is being absorbed by supply.

The bullish inputs are real. The 16 Bcf build for the week ending August 14 came in 13 Bcf under the five-year average of 29 Bcf, and three of the last four prints have been at or below seasonal norms. A hefty South Central withdrawal limited the overall build. Houston temperatures are running near 100°F through August 23 with warmer-than-normal conditions forecast into early September. Hydro generation is set to fall 3% in the second half on western drought. Freeport's 2.0 Bcf/d returns in late August. International prices are elevated with the Hormuz disruption slowing LNG vessel traffic. And Cheniere raised its full-year outlook on strong export demand.

None of it is working. The contract closed at $2.690 Monday, broke resistance Tuesday, rallied above $2.80 Wednesday to a near four-week high, and opened Thursday at $2.778 before trading back below Wednesday's low.

The reason is 112.0 Bcf/d of production against 110.7 Bcf/d in July, itself a record. LNG feedgas added 0.1 Bcf/d month over month while supply added roughly 0.9 Bcf/d. Record power generation produced gas burn 2% below its own record because solar rose 21% and wind 6%. Hugh Brinson delivers another 1.5 Bcf/d of Permian gas to Henry Hub on September 1. And the EIA cut its third-quarter forecast 50 cents to $2.87 while projecting a record 3,985 Bcf at end-October — the highest heading into winter since 2016.

The trade is defined by $2.80 above and $2.68 below. Clearing $2.80 on a close opens $2.90 and then $3.00. Losing $2.68 opens $2.60 and then $2.50.

Fade rallies into $2.85 in the front month. Every summer bounce has stalled in the same place and nothing in the supply data has changed to make this one different.

Own the winter instead. The strip prices a fourth-quarter ramp toward $4.00 to $4.50 and a cold-winter tail at $5.00 to $8.00 with roughly 25% probability, and January 2026 already proved $7.72 is reachable when storage cracks. A 185 Bcf surplus is a cushion, not a guarantee — Winter Storm Fern erased more than that in a single week.

The summer is lost. The asymmetry is in November.

That's TradingNEWS