Henry Hub Horizons
Storage enters summer 144 Bcf above the five-year average. Whether it stays there is a demand-stack question.
Storage enters summer 144 Bcf above the five-year average. EIA says it ends the season above; NGSA says below. The injection season is the referee.
Working gas in storage reached 2,483 Bcf on May 22, 144 Bcf above the five-year average and 21 Bcf above a year ago. A surplus that size often reads as the cap on a summer rally. This year it reads as a starting line, because the two most-watched summer outlooks disagree on where the injection season leaves it.
EIA's May Short-Term Energy Outlook has inventories ending the injection season about 7% above the five-year average. NGSA's summer outlook, issued in mid-May, has them finishing near 3.7 Tcf, slightly below the historical band. From the same May starting point, one outlook ends the season above the benchmark and the other below it. The split is less about supply than about whether record supply can outrun record demand.
Supply is the less-contested half. EIA and NGSA both see record production across Appalachia, Haynesville, and Permian associated gas, with U.S. dry output near a record 111.7 Bcf/d. March already printed 110.9 Bcf/d, the highest for the month on record. Taken alone, that is a surplus-building number.
The demand side is growing faster. NGSA has total summer demand up 6.4 Bcf/d, against supply growth of 4.4. LNG exports lead, rising 4.3 Bcf/d to about 20 Bcf/d as Plaquemines keeps ramping and Corpus Christi Stage 3 and Golden Pass add feedgas. Power burn adds 2.0 Bcf/d to a record 40.3 Bcf/d, the largest domestic demand sector, lifted by coal retirements and data-center load. When demand outpaces supply, storage absorbs the gap. On NGSA's balances, summer injections run 1,772 Bcf against 2,210 Bcf last summer, roughly 20% lighter.
The early prints lean tighter. The May 22 injection landed at 92 Bcf, below the mid-90s consensus, and the July contract climbed to a multi-month high near $3.37 in early June, on forecast heat and that smaller build. A prior Horizon argued the five-year average has stopped describing the system it measures. The 2025 season is the recent precedent: production ran near record highs, yet storage ended October only 4% above the five-year average. The same baseline now sits under an even heavier demand stack.
Storage above the five-year average is a fact about May. Whether it is still a fact about October depends on a demand stack that does not pause for the injection season.