Hedge Funds' Record Short in US Gas Is Betting on Weather — AI Turbines and Qatar Now Set the Price

Two different records got blurred into one headline. Money managers' net short in US natural gas futures reached 223,591 contracts in the CFTC report covering the week ended Oct. 27 — the most bearish positioning since February 2020, when the Covid-19 outbreak crushed demand. Separately, short-only positions across seven Henry Hub contracts hit 599,731 contracts on Oct. 20, the highest since CFTC records began in May 2013, while bullish wagers fell to their lowest in almost two years. The net short grew 16% in a single week, from 192,728 contracts.

The reasoning is not hard to find, and it is not contrarian. NOAA's winter outlook for December 2026 through February 2027 favours warmer-than-average temperatures across the northern United States, and funds are leaning on the same El Niño logic that pushed October and November toward the third-warmest comparable period since 1950. The short exists because a warm winter means less heating gas.

The problem is that the warm-winter case is already the base case. The EIA's October outlook has US working gas at 3,850 Bcf at the end of October, 2% above the five-year average, and Henry Hub averaging $3.48 in 2026 and $3.16 in 2027, a 9% decline. The record short is not an outlier view; it is the consensus view expressed at maximum size.

QatarEnergy declared force majeure on LNG shipments in March 2026 after Iranian drone strikes on the liquefaction hubs at Mesaieed and Ras Laffan, halting Qatari liquefaction and intensifying competition for cargoes between the Atlantic and Pacific basins. Europe absorbed the loss into storage: EU gas was 72.4% full on Oct. 3, the lowest for the date since records began in 2011, and IEEFA calculates the bloc may have to cut winter gas demand by 7% year on year. The causal link between the Qatari outage and the storage deficit is an inference, but it is the one the market is pricing.

Brussels says the system can cope. After the Gas Coordination Group reviewed ENTSOG's 2026-27 winter supply outlook on Oct. 8, officials said storage is below historical levels but the EU gas system is flexible enough to meet demand, with infrastructure able to absorb higher LNG imports. The market has already voted differently: TTF front-month gas stood at €103.58/MWh on Oct. 30, up 5.78% in a session and 40.1% over the month, while US natural gas futures fell more than 15% in the week to Oct. 23.

For US gas equities, the European leg is mostly a volume story rather than a price story. Cheniere Energy earned $1.90 billion of adjusted EBITDA in the third quarter on 685 TBtu loaded, 17% more than a year earlier, and raised full-year 2026 guidance again to $8.75-9.25 billion while lifting the quarterly dividend by more than 12%. LNG that is already liquefied gets paid whether the winter in Ohio is mild or not.

Global Energy Monitor's August tracker puts US gas-fired power capacity in development at 378 GW, up 50% from 252 GW in six months, with Texas alone accounting for 122 GW and 77 GW of that tied to data centres. Capacity actually under construction rose 76% in the first half of 2026 to 52 GW, of which 16.9 GW is intended to directly power data centres. Under construction alone now stands at 52 GW, and unlike heating load it never switches off.

The distance between proposals and steel is what matters for the next twelve months. The EIA's February tally put planned US gas-fired capacity additions for 2026 at 6.3 GW, of which 3.3 GW is combined cycle — a fraction of the 378 GW pipeline. Almost none of the AI gas bid has reached the point of ordering turbines.

That ordering lag is the reason the record short has survived this long, and the reason it is dangerous. AI-driven gas burn is weather-insensitive but not time-insensitive: it arrives when turbines are delivered and plants are commissioned, which is 2027 and 2028, not this winter. The short has to be right about the weather before it can be wrong about everything else.

The October outlook has US dry gas production rising from 112.2 Bcf/d in 2026 to 116.1 Bcf/d in 2027, while consumption rises from 92.4 to 93.8 Bcf/d and LNG exports from 17.6 to 18.6 Bcf/d. Production grows 3.9 Bcf/d; demand grows 1.5 Bcf/d.

The surplus has to land somewhere, and it lands in storage. Inventories end October at 3,850 Bcf, 2% above the five-year average — a cushion of roughly 75 Bcf against a winter forecast by a climate model running a strong El Niño. By Oct. 23 the surplus had already narrowed to 38 Bcf, about 1% of the five-year average. There is essentially no surplus left to absorb a cold snap.

The sharpest single finding sits in EQT's hedge book. The third-quarter release shows hedged volume of 1.2 Bcf/d for Q4 2026, falling to 0.7 Bcf/d for Q1 2027 before rising to 1.7 Bcf/d in Q2 and 1.9 Bcf/d in Q3 — roughly four-fifths of 2027 covered on a gas-equivalent basis, with Q1 2027 barely a tenth of the 6.3 Bcfe a day EQT sold in the third quarter alone. The quarter most exposed to a cold snap is the quarter EQT has left open.

Which produces an uncomfortable conclusion: in a squeeze, the producer is the wrong equity. Cheniere Energy earned $1.90 billion of adjusted EBITDA in the third quarter and lifted 2026 guidance to $8.75-9.25 billion, while EQT exited the quarter with $5.6 billion of total debt, a realized price of $2.72 per Mcfe against a $2.96 NYMEX strip, and 21 Bcfe of strategic curtailments taken to defend price. Toll roads get paid for the squeeze; producers get squeezed inside their own hedges.

Two things are not disclosed and matter. Neither the CFTC report nor the Bloomberg write-up names the funds or shows whether their short is paired against long TTF or JKM exposure, and EQT does not publish the strike prices on its hedge book, so the mark-to-market effect of a rally on reported revenue cannot be sized from the release. Anyone claiming to know who is short, or how much it costs them, is guessing.

This is a timing trade dressed as a structural one, and it sits on a curve that has already moved: TTF is up 40.1% in a month, US futures have dropped more than 15% in a week, and the short is at its biggest since Covid. When the consensus is right about the weather and wrong about everything else, the equity sits in the toll roads and the turbines, not the producers.