Henry Hub Defends $2.84 as the Storage Cushion Drains 31 Bcf in 2 Weeks — Freeport's Return Is the Untraded Catalyst

Natural gas trades at $2.91/MMBtu into the close of August, holding the highest level in five weeks after a run that started with the smallest storage build of the summer. The market held around $2.90 through Thursday and Friday, having climbed from $2.81 during the prior report week, with the September NYMEX contract settling at $2.842/MMBtu before rolling.

The move is small in absolute terms and significant in composition. Gas rallied roughly 10 cents on a genuine supply-demand signal — a 15 Bcf injection against a 33 Bcf five-year average — rather than on a weather headline that reverses in 48 hours.

The 2026 arc puts $2.91 in context. Henry Hub averaged $7.72/MMBtu in January on extreme cold and a record 2,020 Bcf withdrawal season punctuated by Winter Storm Fern. By spring it was back below $3. The second quarter averaged $2.83. That is a 62% collapse in four months, and the market has spent the summer chopping in a $2.70 to $3.10 band while production set records.

The longer history frames the volatility. Henry Hub bottomed at $1.63/MMBtu in June 2020, spiked to a 14-year high of $9.85 in August 2022 on Russia-Ukraine supply fears, crashed below $2 in early 2023, sat below $2 again in early 2024, and ran to $7.72 in January 2026 before this round trip. This is the most violent major commodity in the complex and nothing about that has changed.

The current setup is a classic shoulder-season standoff. Domestic supply is at an all-time high with Lower 48 output averaging a record 111.4 Bcf/d in August. Storage sits 6% above the five-year average. And the EIA cut its 2026 Henry Hub forecast by more than 6% in the August Short-Term Energy Outlook.

Against that, injections have collapsed to less than half the seasonal norm for three consecutive weeks, Freeport LNG is returning from maintenance, and forecasters see above-average temperatures across nearly the entire United States through September 10.

The thesis: gas at $2.91 is priced for a comfortable refill that the last three storage reports no longer support. The bearish inputs are all known and published. The bullish ones are arriving.

The 15 Bcf Build Against a 33 Bcf Five-Year Average

The week ending August 21 produced the tightest storage print of the injection season, and it did not get the attention it deserved.

Energy firms added 15 Bcf to storage. The five-year average injection for that week is 33 Bcf. Consensus expected 20 Bcf. Last year's build for the same week was 17 Bcf.

That is a 55% miss against the seasonal norm and a 25% miss against the forecast. All regions posted increases except Pacific and South Central Salt, both of which withdrew.

The three-week sequence is what matters more than the single print. Week ending August 7: +36 Bcf. Week ending August 14: +16 Bcf. Week ending August 21: +15 Bcf. Against five-year averages that run in the low-to-mid 30s, the market has injected 31 Bcf across two weeks where it should have injected roughly 66.

Compare that to earlier in the season. Week ending July 3 delivered +61 Bcf. July 10: +43 Bcf. July 17: +32 Bcf. July 24: +28 Bcf. July 31: +33 Bcf. The pace has decelerated every step since early July and has now fallen off a cliff.

The cause is straightforward and it is demand-side. A persistent heat dome across the South supported strong regional cooling demand, record-high temperatures hit the Southwest, and operators in the South Central region drew heavily on supply in storage during the week as sweltering heat and recovering LNG activity competed for molecules.

The reason it has not moved price further is equally straightforward. Production is at a record, and the market has been treating the surplus cushion as sufficient to absorb any injection shortfall.

That calculus changes if the pace persists into September. Two more sub-20 Bcf weeks against 40-plus Bcf averages would erase roughly 50 Bcf of surplus in a fortnight, and the surplus is the only thing keeping this market at $2.91.

The next report lands Thursday, September 3, covering the week ending August 28.

Storage at 3,184 Bcf Is Still 167 Bcf Above Normal

The cushion is real, it is shrinking, and both facts matter.

Working gas in storage totaled 3,184 Bcf as of Friday, August 21. That is 167 Bcf — 6% — above the five-year average of 3,017 Bcf, and 30 Bcf or 1% below last year's level at the same point. Total working gas remains within the five-year historical range.

The surplus has been eroding steadily. Week ending July 31 showed a surplus of 195 Bcf. August 7 widened it to 198 Bcf. Then August 14 cut it to 185 Bcf and August 21 to 167 Bcf. That is 31 Bcf of cushion lost in two weeks — the direct arithmetic consequence of injecting 31 Bcf when the norm is 66.

The year-over-year position tells the same story from a different angle. Stocks ran 12 Bcf below last year at the end of July, 25 Bcf below on August 7, 28 Bcf below on August 14, and 30 Bcf below on August 21. The deficit is widening while the five-year surplus narrows — both moving the same direction.

Season context explains why the market has been relaxed. The 2026 injection season began around 1,829 Bcf in late March, near the five-year average, with the first build of 36 Bcf arriving two weeks earlier than normal for the week ending March 27. The market needed to inject roughly 2,000 Bcf over 30 weeks — about 67 Bcf per week — to reach a comfortable pre-winter target, and it has run comfortably ahead of that pace for most of the summer.

The EIA still expects storage to end the injection season on October 31 at 7% above the prior five-year average, which the agency describes as a comfortable cushion that reduces near-term price risk.

That forecast was built on injection assumptions the last three weeks have not met. Nine weeks remain in the season.

Production at a Record 111.4 Bcf/d Is the Ceiling

Every bullish argument in this market runs into the same wall.

Lower 48 output averaged a record 111.4 Bcf/d so far in August, up from 110.7 Bcf/d in July. The EIA's August Short-Term Energy Outlook has U.S. dry gas production setting a new annual record at 111.2 Bcf/d for 2026, a 3.3% gain over the 2025 record. Marketed natural gas production is on track for its own annual record, expected to average nearly 123 Bcf/d for the year.

The sources of that growth are specific. Increased associated gas production in the Permian basin — gas that comes up with oil and responds to crude economics rather than gas prices — and stronger Haynesville output responding directly to Gulf Coast feedgas demand.

That composition is what makes the supply response so hard to break. Permian associated gas keeps flowing at any Henry Hub price because the operator is drilling for oil. With Brent at $90.69 after the Larak Island strike, Permian activity has every incentive to keep running, which means the gas keeps arriving regardless of what happens at Henry Hub.

The forward numbers extend the pressure. Lower 48 gross production averaged 117.2 Bcf/d in the first quarter of 2026, up 4% year over year, with the EIA forecasting 118.9 Bcf/d for full-year 2026 and 124.0 Bcf/d in 2027.

The balance math is the summary. Forecast supply growth outpaces demand growth by 0.5 Bcf/d in 2026, with demand including exports rising less than 1% at +0.6 Bcf/d while supply including imports rises nearly 1% at +1.1 Bcf/d. That reverses in 2027, when demand growth of +2.5 Bcf/d exceeds supply growth of +0.9 Bcf/d, falling behind by 1.6 Bcf/d and putting upward pressure on prices.

So the structural tightening arrives next year. This year, strong domestic production continues to cap every rally, and it capped this one at $2.91.

Freeport Took 2 Bcf/d Off the Board and Is Bringing It Back

The single most tradeable near-term variable is a maintenance outage that is ending.

Freeport LNG maintenance removed roughly 2 Bcf/d of nominal capacity from the market through August. The EIA expected that work to conclude in late August, allowing more U.S. supply to reconnect with international markets where European and Asian prices remain substantially above Henry Hub.

The return has started. Freeport feedgas nominations neared 2 Bcf/d on Thursday, returning roughly 0.8 Bcf/d of demand to the Gulf Coast just as summer heat lingers into September.

That 0.8 Bcf/d is not a rounding error against a market injecting 15 Bcf per week. Fifteen Bcf across seven days is 2.14 Bcf/d of net surplus. Adding 0.8 Bcf/d of feedgas demand removes more than a third of that, and full Freeport restoration at 2 Bcf/d would eliminate it entirely.

The EIA's forecast already partially accounts for it. The August STEO projects U.S. LNG exports averaging 16.5 Bcf/d in the third quarter of 2026, slightly below the previous forecast specifically because of Freeport maintenance, with exports continuing to increase through 2027.

The longer arc is larger. LNG exports grow a forecast 9% — 1.3 Bcf/d — in 2026 and 11% — 1.7 Bcf/d — in 2027, driven by the ramp-up of three new facilities: Plaquemines LNG, Corpus Christi Stage 3, and Golden Pass LNG. Plaquemines and Corpus Christi Stage 3 continue ramping toward full operations, and Golden Pass is expected to begin operations in 2026.

That is the demand side that turns the 2027 balance negative and drives the EIA's forecast of a 33% price increase next year.

For September specifically, the trade is simple. Freeport coming back is a known, dated, quantifiable demand increase landing into a market whose injection pace has already collapsed. It is the cleanest bullish input available and it is not in the price.

LNG Feedgas at 17.1 Bcf/d Is 12.8% Below the Record

The gap between current and peak feedgas is the size of the coiled spring.

Average gas flows to the nine major LNG export facilities ran 17.1 Bcf/d in August, down slightly from 17.2 Bcf/d in July. For the week ending August 19, feedgas averaged 17.1 Bcf/d — 12.8% below the early-year daily record of 19.6 Bcf/d, according to Rystad Energy preliminary data.

That 2.5 Bcf/d gap between current flows and the record is almost entirely maintenance-driven, with Freeport accounting for the bulk of it. Recovering to the record adds 2.5 Bcf/d of demand to a market currently netting roughly 2.14 Bcf/d of surplus.

The arbitrage economics support full utilisation. European and Asian prices remain substantially above Henry Hub, which means every available cargo has a profitable destination. There is no commercial reason for U.S. feedgas to run below capacity once maintenance completes.

The AGA framed the market's central tension precisely: it is shifting from summer weather toward whether record production can keep pace with recovering LNG exports and winter demand.

That framing is the correct one for anyone positioning into September. The weather trade is a two-week story that ends with the shoulder season. The LNG trade is a structural demand increase that compounds through 2027.

The counterweight sits in the same data. LNG demand is showing signs of recovery but flows have not yet moved higher — 17.1 Bcf/d in August against 17.2 in July is a marginal decline, not a rebound. Until the weekly feedgas number prints above 18 Bcf/d, the recovery is a forecast rather than a fact.

Watch the daily nomination data through the first week of September. Freeport back at full rate with the other eight terminals running normally puts feedgas near 19 Bcf/d, and that number arriving alongside sub-20 Bcf storage builds is what breaks $3.00.

Europe at 60.8% Full Is Below the Five-Year Minimum

The international picture stands in stark contrast to the bearish domestic one, and it is the reason the export pull will not weaken.

European Union storage inventories were 60.8% full as of August 15, according to S&P Global Energy — below the five-year minimum for that point in the season. That is a genuinely tight position entering the final weeks of the European injection season, and it means EU buyers must compete aggressively for cargoes through autumn.

Tightening global supply-demand balances against a well-supplied U.S. market is the definition of an arbitrage that pulls molecules out of Henry Hub.

The historical precedent for what tight EU storage does to prices is recent. Reduced EU storage running at 48% against a five-year average of 63% in January drove TTF futures to a weekly average of $12.40/MMBtu, with East Asian LNG cargoes at $10.73/MMBtu — both multiples of Henry Hub.

The current spread works the same way. As long as European and Asian benchmarks sit substantially above U.S. prices, every incremental Bcf of American liquefaction capacity is contracted, loaded and shipped, and the domestic surplus drains into international storage rather than into U.S. inventories.

That is the mechanism converting a 2026 domestic glut into a 2027 domestic tightness. The EIA has demand growth exceeding supply growth by 1.6 Bcf/d next year driven mainly by feed gas demand from U.S. LNG export facilities, reducing gas in storage.

The risk to that framing is a warm European winter. EU storage at 60.8% is only a problem if December is cold. A mild heating season in Europe collapses TTF, closes the arbitrage, and leaves U.S. cargoes competing for a shrinking pool of buyers.

For now, the pull is real and it is one-directional. American gas is the marginal supply to a market that does not have enough.

The EIA Cut 2026 to $3.44 and 2027 to $3.18

The official forecast has been revised down all year, and the market currently trades below even the reduced number.

In the August 2026 Short-Term Energy Outlook, the EIA lowered its 2026 Henry Hub spot price forecast by more than 6% from the July STEO, moving from $3.67 to $3.44 per MMBtu. That annual forecast has now fallen more than 20% since the February 2026 STEO estimate of $4.31, which had been influenced by sustained heating demand, record storage withdrawals and the price spikes during Winter Storm Fern.

The 2027 revision is larger. The agency cut its forecast 11.5% to $3.18/MMBtu from $4.60, citing stronger-than-expected storage builds and production growth.

That second cut is analytically important. The January STEO had projected Henry Hub decreasing about 2% to just under $3.50 in 2026 before rising sharply to just under $4.60 in 2027 — a 33% annual increase driven by LNG feedgas demand outrunning supply. The agency has since kept the 2026 number roughly intact while gutting the 2027 recovery, which means it now expects production growth to absorb the LNG ramp rather than be overwhelmed by it.

Front-month gas at $2.91 sits 15% below the $3.44 annual forecast. Either the market is pricing a weaker fourth quarter than the agency expects, or the agency has not yet cut enough.

Bank forecasts sit higher and were made earlier. Goldman Sachs raised its 2026 Henry Hub forecast to $4.15/MMBtu in January, citing a colder-than-expected winter tightening storage. Morgan Stanley has carried a structural target near $5/MMBtu. Channel-based technical work puts the mid-range near $4/MMBtu as the most likely destination heading into winter 2026-27, with the upper boundary near $5.

Every one of those numbers is above spot by 37% or more.

The next STEO publishes September 9. Whether the agency revises the fourth-quarter assumption after three consecutive sub-20 Bcf injections is the datapoint to watch.

January's $7.72 Print Is What the Tail Looks Like

Anyone modelling this market on a $2.91 base needs to hold January in front of them.

Henry Hub averaged $7.72/MMBtu for the month of January 2026. That was driven by extreme cold and a record 2,020 Bcf withdrawal season, with temporary spikes during Winter Storm Fern. The Henry Hub spot price rose $1.86/MMBtu in a single report week during that stretch, from $3.12 to $4.98, while the February NYMEX contract climbed $1.76 from $3.120 to $4.875 and the 12-month strip jumped 65 cents to $3.970.

That is a 165% monthly average move above current spot, produced entirely by weather.

The relevant lesson is not that it will repeat. It is that this market has no upper bound when storage draws exceed capacity to deliver, because there is no strategic reserve and no spare production that can be dispatched inside a week. The Permian and Haynesville produce what they produce.

Storage at 3,184 Bcf heading toward a 7%-above-average season-end is the buffer against that outcome, and it is why the market is comfortable at $2.91. A cushion of 167 Bcf covers roughly ten days of peak winter withdrawal at the record rate seen last season.

Ten days is not a lot of margin.

The asymmetry is what should drive positioning. Downside from $2.91 is bounded by production economics — sustained prices below $2.50 shut in dry-gas drilling in the Haynesville within a quarter, though Permian associated gas keeps flowing regardless. Upside is bounded by nothing except weather.

That is why option markets in this commodity price such enormous skew, and why the four-month distance between now and January makes September the cheapest month of the year to own optionality.

The 2026 experience — $7.72 in January, below $3 by spring, $2.83 in the second quarter — is a complete illustration of both tails inside eight months.

Weather: Above-Normal Across Nearly the Entire U.S. Through September 10

The near-term demand signal is unusually strong for the calendar date, and it is the reason gas holds $2.90.

The Commodity Weather Group flagged well-above-average temperatures across the eastern two-thirds of the United States from August 31 through September 4, which is likely to sustain gas demand from power generators as air-conditioning use remains elevated. It followed that on Thursday with a broader call: above-average temperatures across nearly the entire United States from September 1 through 10.

Record-high temperatures were forecast across the Southwest through the weekend. A persistent heat dome across the South has been supporting strong regional cooling demand and slowing the pace of storage injections all month.

Power-sector demand is where this transmits. Rystad data had natural gas demand for electric power generation averaging 45.6 Bcf/d for a peak summer week, more than 15% above the comparable period — and every incremental degree-day in September adds to that at a time when the market should be injecting aggressively.

The bearish counterweight is the calendar. Traders are weighing the approaching shoulder-season decline in demand against strong LNG feedgas and hotter regional weather revisions. September heat delays the shoulder season; it does not cancel it. By the second week of October, cooling demand collapses and heating demand has not begun, which is structurally the weakest demand window of the year.

That transition is the risk to any long position taken on the current weather forecast. Ten days of above-normal temperatures buys ten days of injection suppression, and then the market has to find a new reason.

The AGA's read is that the heat dome's effect on total demand and Henry Hub prices has been muted, which is a fair characterisation of a market that has gained 10 cents on three weeks of tight builds.

Watch the six-to-ten-day forecasts through September 5. A shift to normal temperatures with Freeport still ramping is a $2.75 tape.

Regional Basis: SoCal +$1.19 While Texas Eastern M-3 Fell $0.31

The regional dispersion during the last report week was extreme, and it identifies where the physical stress actually sits.

The largest increase across U.S. hubs was $1.19/MMBtu at SoCal Border Average. The largest decline was $0.31/MMBtu at Texas Eastern M-3.

The Southern California move reflected a combination of extreme heat, stronger gas-fired power generation, and pipeline capacity constraints affecting westbound flows. That is a delivery problem, not a supply problem — the gas exists, it cannot get there fast enough.

The storage data confirms the regional pattern. In the week ending August 21, all regions posted injections except Pacific and South Central Salt, both of which withdrew. Earlier in the summer the same split appeared: for the week ending July 24, Mountain withdrew 2 Bcf, Pacific withdrew 7 Bcf, and South Central withdrew 9 Bcf, with South Central salt alone withdrawing 14 Bcf while nonsalt stocks sat 6.0% below last year.

South Central salt facilities are the fastest-cycling storage in the system and they are the buffer for Gulf Coast LNG feedgas swings. Persistent salt withdrawals during peak injection season signal that the export terminals are pulling harder than the pipeline system can deliver.

The practical implication for the Henry Hub forecast: national inventory surplus does not remove local basis risk, pipeline-constraint risk, LNG-feedgas risk or high-demand risk. A comfortable 167 Bcf national cushion coexists with a Pacific region drawing down and Gulf Coast salt caverns being emptied.

Those regional stresses are where a winter price event originates. The national number stays comfortable right up until a specific basin cannot deliver, and then the front month repriced 165% in a month, as it did in January.

For September, the basis blowouts are a warning rather than a trade. They tell you the system is running tighter than 6%-above-average implies.

The Level Map: $2.81, $2.90, $3.00, $3.44

The technical structure is a tight range with well-defined boundaries and a clear catalyst hierarchy.

Immediate support is $2.842, the September contract's final settlement. Below that, $2.81 was the Henry Hub spot low during the last report week and marks the bottom of the recent range. Beneath $2.81, the second-quarter average of $2.83 has acted as a magnet all summer, and $2.70 is the floor the market has defended since spring.

Immediate resistance is $2.91 to $2.90 — the five-week high the market is currently testing. Clearing it decisively opens $3.00, which is psychological and has capped every rally since spring. Above $3.00, the next reference is $3.15, then the EIA's $3.44 annual forecast for 2026.

Beyond $3.44, the structure is thin. Channel work puts the mid-range near $4.00/MMBtu as the most likely destination heading into winter 2026-27, with the upper boundary near $5.00 aligning with Morgan Stanley's structural target.

The trading framework is straightforward. Gas has spent the summer between roughly $2.70 and $3.10, and $2.91 sits in the upper third of that band. A breakout requires either a sub-10 Bcf injection or Freeport returning to full rate — both plausible within two weeks. A breakdown requires a shift to normal temperatures plus a return to 40-plus Bcf builds.

Seasonality argues for caution on longs. NYMEX Henry Hub futures expire three business days before the first day of the delivery month, closer to delivery than WTI crude, which stops trading around ten days out. That compressed expiry makes front-month gas unusually sensitive to prompt physical conditions in the final week of each contract.

October is now the front month, and October is the weakest demand month of the year.

Range call into the September 9 STEO: $2.75 to $3.10, with a break above $3.00 requiring the storage pace to stay under 20 Bcf.

The Week That Prices September: Storage, STEO, and the Shoulder

Three scheduled events determine whether gas holds $2.90 or gives it back.

The EIA Weekly Natural Gas Storage Report lands Thursday, September 3, covering the week ending August 28. Consensus will build around the recent pattern, and the five-year average for that week runs in the mid-40s Bcf as injection season peaks. A third consecutive sub-20 Bcf print against a 40-plus average would cut the surplus below 140 Bcf and force a repricing.

The EIA Short-Term Energy Outlook publishes Wednesday, September 9. The August edition cut the 2026 Henry Hub forecast to $3.44 from $3.67 and the 2027 forecast to $3.18 from $4.60. Whether the agency revises the fourth-quarter path after three tight builds and Freeport's return tells you how the official balance is shifting.

Then the shoulder season itself. Above-average temperatures across nearly the entire United States from September 1 through 10 delay the demand collapse. What follows is the structurally weakest window of the year, when cooling demand ends and heating has not started.

The macro backdrop is a secondary but real input. Gas is a physical commodity with a domestic buyer base, so it trades the rate path less directly than gold or crypto — but industrial demand and power-sector consumption both respond to growth. The Chicago Business Barometer collapsed 10.5 points to 47.1 in August, the first contraction reading in four months. ISM Manufacturing prints Tuesday, September 1, and the August employment report lands Friday, September 4.

Crude matters too, and it moved hard. Brent traded $90.69 on Monday, up 2.93% after U.S. forces struck Iranian rocket launchers on Larak Island. Sustained high oil prices keep Permian drilling active, which keeps associated gas flowing regardless of Henry Hub — the single most important bearish transmission channel from the oil market into gas.

Baker Hughes rig counts publish Friday.

Natural Gas Price Forecast: $2.75 Downside, $3.15 Upside, Constructive Above $2.84

Gas at $2.91 is a market with a shrinking cushion, a record supply base and a demand increase arriving on a known schedule.

The bull case has five legs. The August 21 injection came in at 15 Bcf against a 33 Bcf five-year average and a 20 Bcf consensus — the third consecutive week running at less than half the seasonal norm. The five-year surplus has narrowed from 198 Bcf on August 7 to 167 Bcf on August 21, a 31 Bcf erosion in two weeks, while the year-over-year deficit widened to 30 Bcf. Freeport LNG is returning from maintenance that removed roughly 2 Bcf/d, with nominations already back near 2 Bcf/d and 0.8 Bcf/d of demand restored. Feedgas at 17.1 Bcf/d sits 12.8% below the 19.6 Bcf/d record, and EU storage at 60.8% full is below the five-year minimum, guaranteeing the export pull persists. And the Commodity Weather Group sees above-average temperatures across nearly the entire United States through September 10.

The bear case has four. Lower 48 production hit a record 111.4 Bcf/d in August, up from 110.7 in July, with dry gas output on track for a record 111.2 Bcf/d annual average and marketed production near 123 Bcf/d. The EIA cut its 2026 Henry Hub forecast more than 6% to $3.44 and its 2027 forecast 11.5% to $3.18, citing stronger storage builds and production growth. Storage remains 6% above the five-year average with the agency still projecting a 7%-above-average season end on October 31. And October is the front month, which is the weakest demand window of the calendar year.

The verdict is constructive above $2.84 and neutral below it. Base case through the September 9 STEO: $2.75 to $3.15, midpoint near $2.93. Upside target on a daily close above $2.91 is $3.00, then $3.15; clearing $3.15 with feedgas above 18.5 Bcf/d and a sub-15 Bcf storage print opens the EIA's $3.44 annual forecast. Downside target on a close below $2.84 is $2.81, then $2.70; a break of $2.70 signals the market has decided the shoulder season arrives early.

Trade the Thursday storage print. Everything bearish here is already published.

That's TradingNEWS