Natural Gas Storage Builds Just 15 Bcf vs 20 Est — Salt Caverns Drew 20 Bcf in a Week

Fundamentals · 2026-08-27

Working gas in storage 3,184 Bcf as of Friday August 21, a net build of just 15 Bcf against a 20 Bcf consensus and the third straight week of shrinking injections after 36 and 16; stocks now 30 Bcf below last year, -0.9 percent, but 167 Bcf above the five-year average of 3,017, +5.5 percent; East +19 and Midwest +18 building while South Central drew 19 Bcf and salt caverns alone drew 20; salt stocks -9.2 percent year over year and only 2.1 percent above their five-year average; Mountain and Pacific sit 11.2 and 11.4 percent above theirs; standard error on the total net change is 0.9 Bcf.

What Is This?

Want the full explainer? Economic News Events, Explained breaks down this release and every other one we cover, in plain English.

Summary

Working gas in underground storage was 3,184 Bcf as of Friday, August 21, a net injection of just 15 Bcf against a consensus of 20. That is the third consecutive week of shrinking builds — 36 Bcf, then 16, now 15 — in the middle of injection season, when the industry is supposed to be filling for winter. Prices rallied on the print.

The two headline comparisons point in opposite directions, and both are true. Stocks are 167 Bcf above the five-year average of 3,017, a 5.5 percent surplus, which reads comfortable. Stocks are also 30 Bcf below where they were a year ago, down 0.9 percent, which reads tight. The five-year average is dragged down by genuinely difficult years, so the honest summary is that inventories are normal by historical standards and slightly worse than the recent past — and the direction of travel over the last three weeks has been one way.

The national number is where the report stops being useful, because +15 Bcf is a net of two regions filling and two emptying. East added 19 Bcf and Midwest added 18 — the consumption regions are building normally. Against that, South Central drew 19 Bcf and Pacific drew 3. Mountain added a single Bcf. Sum the components and you get +16; independent rounding brings the published national total to 15.

Inside South Central is the number this report is actually about. Salt-cavern storage drew 20 Bcf in a single week. Salt caverns are the fast-cycle, high-deliverability half of Gulf Coast storage — they can be filled or emptied within days rather than over a season, which makes them the shock absorber for the region that holds every American LNG export terminal and a large share of gas-fired power generation. A 20 Bcf salt draw in the third week of August is not a seasonal pattern. It says Gulf Coast demand outran Gulf Coast supply that week, and it is the reason the national build was so thin. The non-salt half of South Central was essentially flat at +1 Bcf, so the entire regional draw is the fast-cycle storage.

This is not a noise story. EIA publishes standard errors, and the one attached to the salt net change is 0.2 Bcf. On a 20 Bcf move that is about as definitive as weekly energy data gets. The total net change carries a standard error of 0.9 Bcf, so the 5 Bcf shortfall against consensus also sits well outside sampling variability. Pacific carries a standard error of exactly zero because EIA surveys every operator in that region rather than sampling.

The regional distribution of the surplus is the structural point. Nationally stocks are 5.5 percent above the five-year average, but that surplus is not sitting anywhere useful. Mountain is 11.2 percent above its five-year average and Pacific is 11.4 percent above — the West is full. Salt is only 2.1 percent above its and South Central as a whole only 3.0 percent above, while both are down 9.2 and 5.9 percent respectively against last year. Storage geography matters because pipeline capacity between these regions is finite: gas sitting in Wyoming and California cannot serve a Louisiana export terminal on any useful timescale. The country has a 167 Bcf cushion and it is in the wrong place.

Year over year, every region tells the same story about where the tightness is. East is +3.1 percent and Midwest +4.5 percent against last year — comfortable. Mountain is -6.3 percent, South Central -5.9 percent, salt -9.2 percent and non-salt -4.8 percent. The regions that export and generate are running below last year; the regions that simply consume in winter are running above it.

Set against the week's other energy data, the picture is consistent. The crude report showed commercial inventories flat while the Strategic Petroleum Reserve fell 3.7 million barrels and products ran 6 to 14 percent below their five-year averages. Now gas storage builds are decelerating three weeks running with the fast-cycle buffer drawing hard. Neither series is in crisis, and neither has much slack left in the part of the system that responds quickly. That matters for the August inflation print on September 11, with energy already the gap between headline PCE at 3.7 percent and core at 3.3 percent.

The Internals

Working gas in underground storage, Lower 48 states, billion cubic feet:

Region · Aug 21, 2026 · Aug 14, 2026 · Net change · Year ago · Percent vs year ago · Five-year average · Percent vs five-year average

East · 727 · 708 · +19 · 705 · +3.1 · 692 · +5.1

Midwest · 866 · 848 · +18 · 829 · +4.5 · 818 · +5.9

Mountain · 238 · 237 · +1 · 254 · -6.3 · 214 · +11.2

Pacific · 293 · 296 · -3 · 298 · -1.7 · 263 · +11.4

South Central · 1,061 · 1,080 · -19 · 1,127 · -5.9 · 1,030 · +3.0

South Central, salt · 248 · 268 · -20 · 273 · -9.2 · 243 · +2.1

South Central, non-salt · 813 · 812 · +1 · 854 · -4.8 · 788 · +3.2

Total, Lower 48 · 3,184 · 3,169 · +15 · 3,214 · -0.9 · 3,017 · +5.5

Implied flow equals net change in every region this week, meaning there were no reclassifications between categories. Totals do not equal the sum of components because of independent rounding — the regional net changes sum to +16 against a published national total of +15.

The three-week injection path, which is the trend the headline hides:

Week ending · Net change · Total working gas

August 7 · +36 Bcf · 3,153 Bcf

August 14 · +16 Bcf · 3,169 Bcf

August 21 · +15 Bcf · 3,184 Bcf

Where the build came from and where it went:

Direction · Regions · Combined change

Building · East +19, Midwest +18, Mountain +1 · +38 Bcf

Drawing · South Central -19, Pacific -3 · -22 Bcf

Net, before rounding · All five regions · +16 Bcf

Region By Region

The surplus and the deficit sit in different places, which is the whole point of this release:

Region · Percent vs five-year average · Percent vs year ago · Read

Pacific · +11.4 · -1.7 · Full against history, stranded from demand centers

Mountain · +11.2 · -6.3 · Full against history, falling against last year

Midwest · +5.9 · +4.5 · Comfortable on both measures

East · +5.1 · +3.1 · Comfortable on both measures

South Central, non-salt · +3.2 · -4.8 · Thin, and below last year

South Central, total · +3.0 · -5.9 · Thin, and below last year

South Central, salt · +2.1 · -9.2 · The tightest storage in the country

Total, Lower 48 · +5.5 · -0.9 · Normal versus history, slightly worse than last year

Read that table top to bottom and the structure is unmistakable. The two regions carrying double-digit surpluses against their five-year averages are the two furthest from the Gulf Coast export and generation complex. The tightest storage in the country, on both measures, is the salt caverns that serve it. The national 5.5 percent cushion is real but it is not fungible.

EIA's published sampling variability, which is why the salt figure can be taken at face value:

Region · Coefficient of variation for stocks, percent of working gas · Standard error for net change, Bcf

East · 0.6 · 0.4

Midwest · 0.8 · 0.3

Mountain · 2.4 · 0.4

Pacific · 0.0 · 0.0

South Central · 0.8 · 0.6

South Central, salt · 1.2 · 0.2

South Central, non-salt · 1.0 · 0.6

Total, Lower 48 · 0.4 · 0.9

Three things follow. The 20 Bcf salt draw carries a standard error of 0.2 Bcf, so it is measured about as precisely as anything in weekly energy data. The 5 Bcf shortfall against the 20 Bcf consensus is roughly five and a half standard errors on the national net change, so the miss is real rather than survey noise. And Mountain carries the loosest stock estimate at a 2.4 percent coefficient of variation, so its 11.2 percent surplus should be held a little more loosely than the others.

Impact on Natural Gas (NG)

Impact on USD

Impact on US Indices (ES / NQ / YM)

Impact on Gold

What To Watch

TLDR

EIA Weekly Natural Gas Storage Report (week ending August 21, released August 27):

The +15 Bcf headline undersells this report twice over. It is the third consecutive shrinking injection, and underneath it the Gulf Coast salt caverns that feed every American LNG export terminal drew 20 Bcf in a single week — measured to within 0.2 Bcf, so there is no arguing it away. The country holds a 167 Bcf cushion against the five-year average, but 11 percent of it sits in Mountain and Pacific storage that cannot reach the Gulf, while salt sits just 2.1 percent above normal and 9.2 percent below last year. Watch the September 3 report for a fourth thin build, Warsh at Jackson Hole tomorrow, and August inflation on September 11.

_For informational purposes only. Not investment advice._


Read this on ptmtrading.io — Phantom Trading, a trading mentorship community for futures and CFDs.