> ## Documentation Index
> Fetch the complete documentation index at: https://fruitstand.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# EIA Energy Data in Snowflake

> 3.8M+ EIA energy series in Snowflake: natural gas, petroleum, electricity, the hourly grid and EIA forecasts, with revision history. Free 30-day trial.

> [!TIP]
>
> **Try it free for 30 days.** Get full use of every table, series and revision during the trial.

## Overview

The EIA Energy Data listing delivers every public-domain series published by the [U.S. Energy Information Administration](https://www.eia.gov) (EIA): more than 3.8 million series and 250 million values across 28 EIA datasets, ready to query in Snowflake with no API keys or rate limits to manage. Start with a free 30-day trial with full use of the data.

It's built for energy and commodity analysts, quant researchers and data teams who need EIA data next to their own data in Snowflake: gas storage and production, grid demand, refinery runs, prices and EIA's forecasts, without writing and maintaining an EIA pipeline.

Data is standardized into analytic‑ready tables (series metadata, current values, revision history) with every value's period parsed into a real date, the raw value kept next to the numeric one, and automated pipeline quality checks (row parity with EIA's published data, format validation, version ordering) so you can query, join, model and build dashboards right away, without building or maintaining your own EIA ingestion pipeline.

> [card] Coming soon to the Snowflake Marketplace
>
> **Free for 30 days** with full use, then **\$150/month** flat rate — no pipelines to build.

**Pricing**: one flat subscription of \$150/month, billed monthly in advance through Snowflake. It includes every view, all 3.8M+ series with full history, the revision history and every refresh. There are no per-query, per-row or per-user fees; your own Snowflake warehouse pays for the queries you run.

* **Free 30-day trial** with full use of every view, series and revision
* **3.8M+ EIA series** from 28 datasets: natural gas, petroleum, electricity, coal, international, state energy, emissions and more
* **Hourly electric grid data (EIA-930)**: demand, generation by fuel and interchange for every balancing authority, from 2019
* **Forecasts and projections**: the Short-Term Energy Outlook, 12 editions of the Annual Energy Outlook and 4 of the International Energy Outlook
* **Full available history**, back to 1859 for some petroleum series
* **Revision history**: every value each data point has had since we started loading it, with `valid_from` and `valid_to`
* **Checked hourly**: new EIA data is in the listing within about an hour of EIA publishing it
* **Automated pipeline quality checks** (row parity with EIA's published data, period and value validation, version ordering)

Fruit Stand also publishes [FRED economic data](https://fruitstand.dev/datasets/fred-series.md), [historical stock prices](https://fruitstand.dev/datasets/historical-stock-prices.md), [fund and ETF returns](https://fruitstand.dev/datasets/fund-returns.md) and [earnings call transcripts](https://fruitstand.dev/datasets/earnings-call-transcripts.md) on the Snowflake Marketplace.

## What is EIA data?

The [U.S. Energy Information Administration](https://www.eia.gov) (EIA) is the independent statistics agency of the U.S. Department of Energy. It collects data from energy companies, utilities, grid operators and other agencies, and publishes the official U.S. numbers on how energy is produced, stored, moved, priced and used. Its data is in the public domain, and it's the standard reference for anyone who analyzes energy markets.

What's available, at a glance:

* **Natural gas**: production, weekly storage, imports and exports, consumption and prices by state
* **Petroleum**: crude oil and product supply, weekly inventories (including Cushing, Oklahoma), refinery runs, imports by country, and retail gasoline and diesel prices
* **Electricity**: generation by fuel, state and power plant, retail sales and prices, and the hourly U.S. grid (demand, generation and interchange for every balancing authority)
* **Coal and nuclear**: coal production, consumption and prices, and the daily status of every U.S. nuclear reactor
* **State and total energy**: the State Energy Data System and the Monthly Energy Review, covering all fuels together
* **International**: production, consumption and reserves for countries around the world
* **Emissions**: energy-related CO2 emissions by state and sector
* **Forecasts**: the monthly Short-Term Energy Outlook and the long-range Annual and International Energy Outlooks

## Coverage

As of November 2026:

| Dataset | Series | Values | Frequencies | History |
| - | - | - | - | - |
| `EBA`: hourly electric grid (EIA-930) | \~1,500 | \~79.2M | Hourly (UTC) | 2019 to present |
| `ELEC`: electricity generation, fuel, sales, plants | \~785,000 | \~61.5M | Monthly, quarterly, annual | 2001 to present |
| `PET`: petroleum supply, stocks, refining, prices | \~181,000 | \~17.2M | Daily, weekly, monthly, annual | From 1859 |
| `INTL`: international energy | \~104,000 | \~5.2M | Monthly, quarterly, annual | From 1949 |
| `NUC_STATUS`: nuclear reactor status | \~500 | \~3.5M | Daily | 2007 to present |
| `SEDS`: State Energy Data System | \~48,000 | \~2.6M | Annual | 1960 to 2024 |
| `COAL`: production, consumption, prices | \~189,000 | \~1.6M | Quarterly, annual | 2000 to present |
| `NG`: natural gas, incl. weekly storage | \~16,000 | \~1.1M | Daily, weekly, monthly, annual | From 1900 |
| `PET_IMPORTS`: crude oil imports by origin and destination | \~46,000 | \~1.0M | Monthly, annual | 2009 to present |
| `STEO`: Short-Term Energy Outlook | \~4,400 | \~0.75M | Monthly, quarterly, annual | 1960, with forecasts to the end of next year |
| `TOTAL`: Monthly Energy Review | \~2,000 | \~0.63M | Monthly, annual | From 1949 |
| `EMISS`: CO2 emissions | \~4,900 | \~0.21M | Annual | 1970 to 2021 |
| `AEO.2014` to `AEO.2026`: Annual Energy Outlook (12 editions) | \~2.45M | \~75.4M | Annual | Projections to 2040 or 2050 |
| `IEO.2017` to `IEO.2023`: International Energy Outlook (4 editions) | \~41,000 | \~1.4M | Annual | Projections to 2050 |

Series EIA licenses from third parties (for example the Thomson Reuters spot prices in `NG` and `PET`) aren't public domain and are left out. Every other EIA series in these datasets is included.

## Schema

The listing has three views. Every series ID starts with its dataset, for example `NG.NW2_EPG0_SWO_R48_BCF.W`. Join the observation views to `eia_series_vw` on `series_id` to get each value's name, units and geography.

### `eia_series_vw`

One row per series, including series EIA has stopped publishing.

| Column | Type | Description |
| - | - | - |
| `series_id` | `varchar` | EIA series ID, prefixed with the dataset (e.g. `NG.NW2_EPG0_SWO_R48_BCF.W`) |
| `dataset` | `varchar` | EIA dataset (e.g. `NG`, `PET`, `ELEC`, `EBA`, `AEO.2026`) |
| `name` | `varchar` | Series name as published by EIA |
| `description` | `varchar` | Longer description, where EIA provides one |
| `units` | `varchar` | Units of the values |
| `units_short` | `varchar` | Abbreviated units |
| `frequency` | `varchar` | `A` annual, `Q` quarterly, `M` monthly, `W` weekly, `4` four-week average, `D` daily, `H` hourly |
| `source` | `varchar` | Source agency as published by EIA |
| `geography` | `varchar` | Geography code(s), e.g. `USA` or a state |
| `geography2` | `varchar` | Second geography, for flows between two places |
| `geoset_id` | `varchar` | EIA ID of the set of series that differ only by geography |
| `iso3166` | `varchar` | ISO 3166 country or subdivision code |
| `lat` | `varchar` | Latitude, for plant- and site-level series |
| `lon` | `varchar` | Longitude, for plant- and site-level series |
| `latlon` | `varchar` | Latitude and longitude together |
| `start_period` | `varchar` | First period of the series |
| `end_period` | `varchar` | Last period of the series |
| `last_historical_period` | `varchar` | STEO only: the last period with actual data. Later periods are forecasts. |
| `series_last_updated` | `timestamp_tz` | When EIA last updated this series |
| `is_active` | `boolean` | False once EIA has stopped publishing the series. Its last values stay in the listing. |
| `inactive_at` | `timestamp_ltz` | When EIA was first seen to have stopped publishing the series; null while active |
| `src_created_at` | `timestamp_ltz` | When this series first arrived in the listing |
| `src_updated_at` | `timestamp_ltz` | When this row last changed. Filter on it to load only changed series. |

### `eia_observations_vw`

The current value of every data point: one row per `(dataset, series_id, period)`.

| Column | Type | Description |
| - | - | - |
| `dataset` | `varchar` | EIA dataset |
| `series_id` | `varchar` | EIA series ID |
| `period` | `varchar` | Period as EIA publishes it: `2026` (A), `2026Q3` (Q), `202608` (M), `20261030` (W/D), `20261111T23` (H) |
| `period_start` | `date` | First day of the period |
| `period_ts` | `timestamp_tz` | Hourly series only: the hour, in UTC |
| `frequency` | `varchar` | Frequency of the series |
| `value` | `float` | Numeric value; null when EIA published a marker or no value |
| `value_raw` | `varchar` | Value exactly as EIA published it, including markers such as `NA`, `--` and `W` (withheld) |
| `version_at` | `timestamp_tz` | When EIA published this value, or last revised it |

### `eia_observations_history_vw`

Every value each data point has had since we started loading it: one row per `(dataset, series_id, period, valid_from)`. The row with `is_current = true` matches `eia_observations_vw`.

| Column | Type | Description |
| - | - | - |
| `dataset` | `varchar` | EIA dataset |
| `series_id` | `varchar` | EIA series ID |
| `period` | `varchar` | Period as EIA publishes it: `2026` (A), `2026Q3` (Q), `202608` (M), `20261030` (W/D), `20261111T23` (H) |
| `period_start` | `date` | First day of the period |
| `period_ts` | `timestamp_tz` | Hourly series only: the hour, in UTC |
| `frequency` | `varchar` | Frequency of the series |
| `value` | `float` | Value during this version; null when EIA published a marker or no value |
| `value_raw` | `varchar` | Value exactly as EIA published it during this version, including markers |
| `valid_from` | `timestamp_tz` | When EIA published this value |
| `valid_to` | `timestamp_tz` | When EIA revised or removed it; null while it's still current |
| `is_current` | `boolean` | True for the value that is current today |

## Example queries

All example output below is real, pulled from the listing on **November 20, 2026 at 01:50 UTC**. Timestamps are shown in UTC. Each query finished in under 1.5 seconds on an X-Small warehouse. Run the same queries yourself during the free 30-day trial once the listing launches on the Snowflake Marketplace (coming soon).

### Natural gas in storage

Weekly Lower 48 natural gas working storage, in billion cubic feet:

```sql theme={null}
select period_start, value
from eia_observations_vw
where series_id = 'NG.NW2_EPG0_SWO_R48_BCF.W'
order by period_start desc
limit 5;
```

| period\_start | value |
| - | - |
| 2026-11-13 | 3,882 |
| 2026-11-06 | 3,862 |
| 2026-10-30 | 3,801 |
| 2026-10-23 | 3,724 |
| 2026-10-16 | 3,644 |

The last three years of the same series:

**Lower 48 working gas in storage, weekly**

### Find series by name

Crude oil stocks at Cushing, Oklahoma:

```sql theme={null}
select series_id, name, units, frequency
from eia_series_vw
where dataset = 'PET'
  and name ilike '%cushing%stocks%'
  and is_active;
```

| series\_id | name | units | frequency |
| - | - | - | - |
| PET.W\_EPC0\_SAX\_YCUOK\_MBBL.W | Cushing, OK Ending Stocks excluding SPR of Crude Oil, Weekly | Thousand Barrels | W |
| PET.MCRST\_YCUOK\_1.M | Cushing, OK Ending Stocks of Crude Oil, Monthly | Thousand Barrels | M |
| PET.MCRST\_YCUOK\_1.A | Cushing, OK Ending Stocks of Crude Oil, Annual | Thousand Barrels | A |

### Hourly electricity demand

Hourly Lower 48 electricity demand, in megawatthours, from EIA-930:

```sql theme={null}
select period_ts, value
from eia_observations_vw
where series_id = 'EBA.US48-ALL.D.H'
order by period_ts desc
limit 24;
```

First 5 of 24 rows:

| period\_ts | value |
| - | - |
| 2026-11-19 19:00 UTC | 460,080 |
| 2026-11-19 18:00 UTC | 467,824 |
| 2026-11-19 17:00 UTC | 469,789 |
| 2026-11-19 16:00 UTC | 473,333 |
| 2026-11-19 15:00 UTC | 478,352 |

The last seven days, with the daily demand cycle:

**Lower 48 electricity demand, hourly**

### STEO forecasts vs. actuals

STEO world petroleum production, in million barrels per day, with EIA's forecasts marked apart from actuals:

```sql theme={null}
select o.period_start,
       o.value,
       o.period_start > to_date(s.last_historical_period, 'YYYYMM') as is_forecast
from eia_observations_vw o
join eia_series_vw s on s.series_id = o.series_id
where o.series_id = 'STEO.PAPR_WORLD.M'
  and o.period_start >= '2024-01-01'
order by o.period_start;
```

The rows around the switch from actuals to forecasts:

| period\_start | value | is\_forecast |
| - | - | - |
| 2026-05-01 | 94.33 | false |
| 2026-06-01 | 99.28 | false |
| 2026-07-01 | 101.66 | false |
| 2026-08-01 | 100.77 | true |
| 2026-09-01 | 103.08 | true |
| 2026-10-01 | 104.47 | true |

**World petroleum production (STEO), monthly**

### Revisions to one data point

Every value one hour of grid demand has had, as EIA revised it:

```sql theme={null}
select period, value, valid_from, valid_to, is_current
from eia_observations_history_vw
where series_id = 'EBA.US48-ALL.D.H'
  and period = '20261111T23'
order by valid_from;
```

| period | value | valid\_from | valid\_to | is\_current |
| - | - | - | - | - |
| 20261111T23 | 484,286.77 | 2026-11-18 20:04 UTC | 2026-11-19 10:03 UTC | false |
| 20261111T23 | 484,287.77 | 2026-11-19 10:03 UTC | 2026-11-19 20:03 UTC | false |
| 20261111T23 | 485,038.77 | 2026-11-19 20:03 UTC | | true |

## Revision history

EIA revises values after it first publishes them, and only ever publishes the latest number. `eia_observations_vw` gives you that latest number. `eia_observations_history_vw` keeps every earlier one: each time EIA changes or removes a value, the old row is closed with a `valid_to` and the new one starts. That lets you see how a number was reported before it was revised, or rebuild a series as it was known at a past moment:

```sql theme={null}
select series_id, period, value
from eia_observations_history_vw
where series_id = 'EBA.US48-ALL.D.H'
  and valid_from <= '2026-11-19 00:00 +00:00'::timestamp_tz
  and (valid_to > '2026-11-19 00:00 +00:00'::timestamp_tz or valid_to is null)
order by period desc
limit 3;
```

| series\_id | period | value |
| - | - | - |
| EBA.US48-ALL.D.H | 20261118T19 | 462,583 |
| EBA.US48-ALL.D.H | 20261118T18 | 469,049 |
| EBA.US48-ALL.D.H | 20261118T17 | 473,076 |

History starts when we began loading each dataset in November 2026. Revisions EIA made before then aren't known.

## Data Ingestion

Data is ingested into Snowflake with best practices in mind.

**1. Change detection**

EIA's bulk manifest is checked every hour. A dataset is downloaded only when EIA has published a new version of its file, so an unchanged dataset is never reloaded.

**2. Raw snapshot**

Each new file is saved to a dated S3 folder exactly as EIA published it, then copied into Snowflake.

**3. Standardize**

Periods are parsed into dates (and UTC timestamps for hourly data), numeric values are parsed while the raw value is kept, and series metadata is flattened into columns.

**4. Insert-only versioning**

New and revised values are appended next to the old ones, never written over them. That is what builds the revision history, and it means a refresh only adds the values that actually changed.

**Refresh cadence**: each dataset updates on EIA's own schedule. The hourly grid data (`EBA`) republishes about twice a day, natural gas storage weekly, the Short-Term Energy Outlook monthly, and the Annual Energy Outlook once a year. New data is loaded within about an hour of EIA publishing it.

## Data Quality

* **Row parity**: every load checks that each dataset's series and value counts match what EIA published, and that the published counts for all 28 datasets match the listing's views.
* **Format validation**: every staged row is checked for a known frequency, a period that matches its frequency and parses to a date, a numeric value that parsed, and no hourly or daily period far in the future.
* **Unique grain**: every `(dataset, series_id, period)` has exactly one current value.
* **No hollow series**: every series in `eia_series_vw` must have values in `eia_observations_vw`.
* **Truncated-load guard**: a dataset update with far fewer series or values than its last load, or missing key series like Lower 48 gas storage, is stopped before it reaches the listing.
* **Versions only move forward**: a dataset can't be replaced by an older EIA release than the one already published.
* **Public domain only**: licensed series are checked to never appear in the listing's views.

The row-parity and truncated-load checks run before any new values are added, so a bad update never reaches the listing's tables.

## Disclosures

Fruit Stand Data LLC is not endorsed by or affiliated with the U.S. Energy Information Administration. EIA data is in the public domain. Use [contact@fruitstand.dev](mailto:contact@fruitstand.dev) for questions.

**Source gaps**: some EIA series have periods with no value, or a marker such as `NA` or `W` (withheld to avoid disclosing individual company data) in place of a number. These are passed through as EIA published them: `value` is null and `value_raw` shows the marker.

## F.A.Q.

Frequently asked questions from our clients.

> [accordion] Is there a free trial?
>
> Yes. The listing comes with a free 30-day trial with full use: every view, every series and the full revision history, the same data paying subscribers get. After 30 days it's \$150/month flat. Nothing is charged when the trial ends unless you choose to subscribe. The Snowflake Marketplace listing is coming soon.

> [accordion] What is the EIA?
>
> The U.S. Energy Information Administration is the statistical agency of the U.S. Department of Energy. It collects and publishes data on energy production, consumption, prices, stocks, imports and exports, and electricity generation, plus forecasts like the Short-Term Energy Outlook and long-range projections like the Annual Energy Outlook.

> [accordion] How often is the data refreshed?
>
> EIA's releases are checked every hour, and new data is loaded within about an hour of EIA publishing it. How often that happens depends on the dataset: the hourly grid data about twice a day, petroleum and natural gas weekly, most other datasets monthly, and the outlooks when EIA releases a new edition.

> [accordion] When does replication across cloud regions happen?
>
> Replication of refreshes across cloud regions begins at 5:00 AM Pacific each day. If you consume this listing from a Snowflake region other than its primary region, expect each day's updated data to arrive after that replication run, so the hourly grid data there is current as of that morning.

> [accordion] How far back does the historical data go?
>
> Each series carries its full history as EIA publishes it. That's back to 1859 for some petroleum series, 1900 for some natural gas series, 1949 for the Monthly Energy Review and international data, and 2001 for electricity. The hourly grid data starts on January 1, 2019.

> [accordion] How is the data obtained?
>
> Directly from EIA's open data, the same data the EIA website and API serve.

> [accordion] Why use this listing instead of the EIA API?
>
> You certainly can use the EIA API to load data into Snowflake yourself. The API needs a key, is rate-limited, and serves data route by route, so loading millions of series means thousands of calls and a pipeline to keep in sync. You'd also have to parse each dataset's own period and value formats, handle markers like `NA` and `W`, detect changes and keep revisions. This listing does all of that, so you can start querying right away, and the free 30-day trial lets you check it against your own use case first.

> [accordion] What does it cost to build this ourselves vs. buy the listing?
>
> Building it yourself means a pipeline that tracks 28 EIA datasets, parses each one's period and value formats, detects new releases every hour, keeps every revision and checks each load, then someone to keep it running as EIA changes things. That's weeks of engineering up front and ongoing maintenance after. The listing is \$150/month with all of it done, and the free 30-day trial lets you check it first.

> [accordion] Does this include revision history?
>
> Yes. `eia_observations_history_vw` keeps every value each data point has had since we started loading it in November 2026, with `valid_from` and `valid_to`. See [Revision history](#revision-history) above.

> [accordion] Are EIA's forecasts included?
>
> Yes. The Short-Term Energy Outlook (`STEO`) has monthly forecasts through the end of next year, and `last_historical_period` on each STEO series marks where actuals end. The Annual Energy Outlook editions from 2014 to 2026 (`AEO.2014` to `AEO.2026`) and the International Energy Outlook editions (`IEO.2017` to `IEO.2023`) carry long-range projections to 2040 or 2050, each edition as its own dataset so you can compare them.

> [accordion] What isn't included?
>
> Two things. First, the 36 series EIA licenses from Thomson Reuters, including the WTI and Henry Hub daily spot prices: they aren't public domain, so they're left out. EIA's own price series are included, such as weekly retail gasoline and diesel prices, natural gas prices by state, and the monthly Henry Hub and WTI prices in the Short-Term Energy Outlook. Second, revisions EIA made before we started loading each dataset in November 2026 aren't in the revision history.

> [accordion] What happens when EIA stops publishing a series?
>
> It stays in the listing with its last values, and `is_active` turns false in `eia_series_vw`. Filter on `is_active` to see only series EIA still updates.

> [accordion] Can I request additional EIA data to be made available in Snowflake?
>
> Yes! Please reach out to [contact@fruitstand.dev](mailto:contact@fruitstand.dev) with the requirements. We are more than happy to make it available.

> [card] Coming soon: try EIA Energy Data free for 30 days
>
> Full use of all 3.8M+ series and their revision history, then **\$150/month** flat rate.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.