EIA-923 generation data lag, explained

If you've ever tried to true up a generation forecast against EIA-923 and found the most recent month missing, you already know the drill: the data exists, it's just not out yet.

EIA-923, the Power Plant Operations Report, collects monthly fuel receipts, fuel consumption, and net generation from utility-scale plants. EIA publishes it roughly 60 to 65 days after the end of the reporting month. January numbers show up in late March. A plant trip, a derate, a fuel-switch decision made on March 3 doesn't show up in federal data until the back half of May, and even then it can get revised again when the next release rolls out.

Why the lag exists in the first place

The delay comes down to the mechanics of survey collection. Plant operators file Form EIA-923 monthly, but EIA has to receive, validate, and reconcile thousands of individual submissions before anything goes public. Some respondents file late. Some get flagged for data quality checks and corrected in a later release. The number you see in month one is often not the number that survives to the annual final file.

Add in the fact that EIA-923 runs on a different clock than EIA-860, the annual survey that reports generator-level capacity, fuel type, retirements, and planned additions, and you get two datasets that never quite line up. EIA-860 preliminary data for a given year typically comes out the following spring, with the final version landing months after that. So if a plant retired a unit or brought a new one online mid-year, you might not get a clean confirmation in the federal record until well into the next year.

For an analyst building a supply stack or checking a coal-burn assumption against actuals, that means working with two lagging series on two different schedules, neither of which tells you what a plant is doing this week.

What desks actually do while they wait

Most shops don't sit idle for 60 days. They triangulate: ISO/RTO dispatch data where it exists, EPA CEMS hourly emissions data for a rough generation proxy, rail and barge coal shipment reports, and whatever operator guidance leaks out in earnings calls or trade press. CEMS helps for emitting units, but it's an emissions proxy, not a direct generation number, and it says nothing about fuel inventory on site or whether a plant is running hot, running light, or down for maintenance.

None of these fill the real gap, which is plant-level confirmation of current status. Is the unit actually running today, not last reported quarter. Is the coal pile drawing down the way a dispatch model assumes, or sitting flat because the plant is in an extended outage nobody's disclosed yet.

That's the gap imagery-based monitoring is built to sit in. Instead of waiting on a monthly filing, you can track stack and cooling-water signatures and coal-pile volume at a named plant on a weekly cadence, and chart that against your generation and commodity assumptions in near real time. Power Plant Activity turns satellite and aerial imagery into exactly that kind of weekly read, built to run alongside EIA-923 and EIA-860, not replace the audit trail they eventually provide.

The federal filings still matter. They're the record of truth, the number that reconciles everything else once it's final. But if you're pricing a fuel-switch trade or checking whether a derate story is real before the market catches up, a 60-day-old number doesn't help you this week. Worth taking a look at how a weekly plant-level signal fits into that gap.