Weekly US Steel Production Data: Where It Comes From and How to Read It

The estimate matters because of how it is built: AISI compiles it from weekly tonnage reported by companies representing roughly 50 percent of domestic capacity, combined with recent monthly data for the remainder. The institute says outright that the figure is intended for trend assessment rather than precise accounting - which is exactly the right frame for anyone wiring it into a model.

Datadory catalogs AISI Industry Data as the weekly pulse of the vertical: one estimate per week, five districts deep, with the monthly shipment series and SIMA import-permit summaries landing beside it. Every figure arrives as a typed row with a field dictionary and a coverage statement attached.

A recent week shows the shape of the data. For the week ending August 15, 2026, AISI estimated 1,833,000 net tons at 79.4 percent capability utilization, against 1,764,000 net tons in the prior-year week, with year-to-date production at 59,163,000 net tons. The district split for that week put Southern at 830 thousand net tons, Great Lakes at 479, Midwest at 318, North East at 140 and Western at 66 - Southern and Great Lakes together carry roughly two-thirds of the reported tonnage.

The remaining fields are monthly, not weekly: steel mill shipments released with month-over-month change, SIMA import permit tonnage split total and finished, and finished import market share as a share of apparent consumption.

Capability utilization is the number that turns a tonnage print into a cycle signal. The weekly release states it as the percentage of installed capability utilized during the week - 79.4 percent in the August 15, 2026 observation - and because the denominator (capability) moves slowly while the numerator (production) moves weekly, the series behaves like a high-frequency capacity gauge.

Two cautions come straight from the source documentation. First, since roughly half of domestic capacity reports actual weekly tonnage and the rest is bridged with recent monthly data, week-over-week changes mix measurement with real movement. Second, the year-ago and prior-week comparisons AISI prints alongside each estimate are the cleanest way to use it: they hold the estimation method constant within any single comparison.

For context outside the weekly print, two federal ledgers anchor the levels: the USGS Iron and Steel Statistics and Information record carries continuously cast share and raw steel production in its annual Mineral Commodity Summaries chapter, and Data Series 140 runs annual US steel series back to 1900. Weekly gives you direction; annual tells you where the level sits.

Datadory catalogs Steel Import Monitor (SIMA) as the Wednesday-side complement to the Friday estimate; Eurostat's dissemination series plays the matching role for European production indices.

No weekly global crude steel series exists in the cataloged record. The closest global analogue to the AISI weekly is worldsteel's Monthly crude steel production (71 countries) release, which covers about 70 countries representing roughly 98 percent of world output and lands three to four weeks after month end. Each release carries two tables: regional production in million tonnes with year-over-year and year-to-date changes, and a top-10 country ranking led by China, India, the United States, Japan and Russia. The June 2026 release covered 70 countries producing 155.7 Mt.

That lag is structural, not sloppiness - national statistics offices report monthly at best. So a practical monitoring stack pairs the weekly AISI estimate for US direction with worldsteel's monthly release for global placement, then uses annual series to anchor levels: UNSD Industrial Commodity Statistics (steel) provides annual crude steel series for about 103 reporting countries, and the USGS world production table in each Mineral Commodity Summaries chapter covers pig iron and raw steel for roughly eleven named countries plus an others line every January.

The stack resolves by grain. Weekly prints give direction, monthly releases give global placement, annual series hold the level - and none substitutes for another, because each is measuring at a different rhythm by design.

The weekly series rewards assembly: six figures a week, five districts, one utilization rate, compounding into a panel that answers cycle questions no monthly release can. A workable build:

1. Treat each weekly observation as a typed row - Saturday-ending week date, estimated net tons, capability utilization percent, adjusted year-to-date tonnage, and the five district values in thousands of net tons.
2. Normalize districts to their canonical names (North East, Great Lakes, Midwest, Southern, Western) so a multi-year panel joins cleanly.
3. Reconcile weekly sums against the monthly anchors - AISI mill shipments and the Annual Statistical Report totals of 91 million net tons shipped and 90 million net tons produced for 2025 - flagging weeks whose revision drift exceeds your tolerance.
4. Append SIMA Wednesday volumes keyed by country and HTS product group so trade flow sits beside production in the same table.
5. Store year-ago and prior-week comparators alongside every row; they are the cleanest cycle signals because they hold the estimation method constant within each comparison.

Steps 2 through 4 are exactly the assembly work Datadory ships as a product: the weekly estimate, its district split and the SIMA trade frame arrive as one reconciled table, delivered daily, weekly, or hourly - your call.

Four personas pull on this feed hardest.

Commodity and procurement analysts read the utilization rate as a tightness gauge: when it runs hot, mill lead times stretch and spot premiums widen, so buying calendars shift forward. The year-over-year and week-over-week comparators matter more than any single print, for the estimation reasons above.

Macro and fixed-income modelers fold weekly tonnage into industrial-production nowcasts, where a fresh weekly estimate beats waiting on monthly factory data, and capability utilization plugs straight into capacity-constraint terms.

Trade analysts pair Wednesday import volumes with Friday production to frame apparent consumption: domestic output plus arrivals, minus exports, is the demand number sitting beneath both series.

Start with the steel data hub - every one of the 25 pooled records with field dictionaries, sample rows and coverage statements - and the ranked shortlist of the best steel datasets. For the Wednesday side of the market, the EU steel & basic-metals production statistics (dissemination API) comparison sets the European production-index frame beside the SIMA trade monitor, and the profile behind the global monthly release is at the World Steel Association source page.

Then go deeper on the two layers this page touches: AISI Industry Data documents every weekly field down to the district split, and Steel Import Monitor (SIMA) carries the Wednesday-side volumes by country and HTS group.