Power and grid buildout

POWER opens the power and grid buildout explorer. Pro unlocks the full dataset and captured history; free access previews three rows per section. The Queue, History, Outcomes, Loads, Utilities, Capacity and Coverage views share filters for geography, technology and status. A selected project opens its original source fields, source link and preserved revisions. Exact company matches link to DES, FA, G, SPLC and TBO; GPU opens compute rental prices. No data is imported from The Buildout.

What the numbers mean

Board results may reuse a stored-data read for up to 30 seconds. Source dates and the board's generation timestamp remain attached to that result; refreshing a pane does not trigger collection. The first request after expiry can take longer than a repeated read.

Queue MW are requested or approved interconnection capacity. They are not a construction forecast or spare grid capacity. Approval and a signed interconnection agreement do not establish commercial operation. Proposed service dates retain the source's precision, and missing values remain blank. Storage power in MW differs from storage energy in MWh. Hybrid projects retain the source's capacity basis; a sum of generation and storage nameplate differs from export rights.

Loads separates requested, approved and operating MW where the publisher discloses them. Capacity segments can be separate parts of a single request, and regional summaries are identified in their evidence. Record counts therefore need not equal unique projects. Utility exposure uses the utility named in the source. It does not imply that every request is a datacentre, that requested capacity will be consumed, or that a project company belongs to an assumed listed parent.

Outcomes groups project records by request year. The completion and withdrawal fractions use known active, approved, completed and withdrawn statuses, with unknown states counted separately. The status filter does not change the rate denominator. Regional summaries and partial capacity segments are excluded. These are observed cohort dispositions, subject to source retention and unresolved recent projects, rather than estimated ultimate success probabilities.

History and evidence

Every successful source capture is retained. Corrections append revisions, and original values remain reviewable. Source publication dates and collection timestamps are separate. A file fetched today does not establish what its queue looked like yesterday; an absent project is not automatically marked withdrawn. Observed charts start only when enough actual captures exist. The annual benchmark charts published regional queued capacity from 2014 through 2025 by reporting year. It converts GW to MW, excludes duplicate national totals, and gaps incomplete series rather than treating missing components as zero. Some hybrid-storage quantities are estimated by the publisher; their evidence retains that note.

Historical benchmarks are separate from current direct queues. LBNL overlaps ISO records, so its historical records must not be added to the current ISO totals. Annual capacity, generation and utility data retain their native reporting period. A generator's capacity, electricity production in MWh, utility customers and sales in MWh are different measures. Utility revenue is reported retail-sales revenue in USD, converted from the source's thousands of dollars; it is not total company revenue.

Evidence carries the primary-source URL, native record or literal extracted line, source locator, source period, observation time and confidence. Confidence describes deterministic source normalization, not the probability that the project will be built. Ticker links require an unambiguous exact entity alias. Unmapped developers and special-purpose project companies remain unmapped.

Coverage

The Coverage view reports each register's availability and most recent successful collection. Current direct sources, historical benchmarks and operating/utility context remain distinguishable. Access restrictions, unavailable endpoints and confidential project lists are shown explicitly. They are not empty queues.

US adapters target the public registers of PJM, ERCOT, MISO, CAISO, NYISO, ISO-NE and SPP where access is permitted, with LBNL historical queues and EIA-860/861/923 context. Verification collected PJM and CAISO directly. ERCOT, MISO, NYISO, ISO-NE and SPP were inaccessible from the collection environment; their failures remain explicit in Coverage, while their published LBNL benchmark records remain available separately. No public project-level ERCOT large-load register was verified. Global adapters cover public UK connections, Irish contracted generation, Norwegian capacity reservations, Swedish regional requests and French renewable development summaries, Ontario connection assessments and New Zealand generation/storage and load pipelines. Ontario assessment approval is separate from commissioning; New Zealand stages, incremental capacity and retirement events retain their own classifications. Source-level availability and the reporting period are authoritative; no global completeness is implied. ENTSO-E requires a token and remains disabled. Australia's generation information page was inaccessible during verification and is also disabled. Confidential or account-only sources are not collected.

Collection is sequential with a declared User-Agent, request bounds, independent source backoff and operator kill switches. A source failure leaves the last successful snapshot available and marked stale. Pane refresh reads stored data and never triggers a source scrape.

Commands

Annual board totals use only the latest published year. The offshore-wind subtotal is retained as evidence but excluded from totals and history aggregates because the published wind total already includes it. Annual history retains each reporting year.