Do ERCOT's primary reports confirm the ~226 GW (Nov 2025) large-load queue, the ~63 GW end-2024 baseline, and the ~77%-data-center share?
claim-ercot-large-load-queue-quadrupled-2025-phantom-load rests on Latitude Media (Tier 2, Catherine Boudreau, Dec 2025) reporting on ERCOT's own large-load interconnection queue data. The numbers are quantitative and load-bearing, so the sourcing floor requires verification against the ERCOT primary, not the trade-press reproduction.
Specific verifications needed:
- Pull the ERCOT large-load figures directly. ERCOT publishes large-load interconnection tracking (e.g. its Large Load Interconnection Study / large-load reports and monthly updates to the board and the Public Utility Commission of Texas). Confirm the "about 226 gigawatts... as of November [2025]" total and the "end of 2024, at 63 GW" baseline — i.e. the ~4× year-over-year growth.
- Confirm the data-center share. Verify "about 77% of that load" is attributable to data centers in ERCOT's own categorization, and how ERCOT defines/attributes the category.
- Confirm the "phantom load" framing quantitatively, if possible. ERCOT and PUCT have discussed speculative/duplicative interconnection requests and reforms (e.g. the "Batch Zero" batching process) to filter them. Is there an ERCOT estimate of how much of the queue is deliverable vs. speculative, or is the gap only inferable from the ~23 GW of generation actually added in 2024–25?
- Confirm units and vintage. 226 GW is several times ERCOT peak demand; verify this is cumulative requested large load in the queue (not net, not co-optimized), and pin the exact snapshot date.
Why it matters: this is the present-day mechanism that anchors the 1935→2026 Texas-grid thread — the claim that the isolated, distinctively-regulated Texas market is where AI data-center demand is stress-testing the grid fastest. If the figures hold against ERCOT's primary, the note can move off seedling. Distinct from question-verify-lbnl-queued-up-interconnection-figures, which concerns the national generation queue (LBNL "Queued Up"), not ERCOT's large-load (demand-side) queue.
Progress log
- Answered by claim-ercot-board-presentation-confirms-226gw-63gw-large-load-queue, claim-ercot-primary-source-73-percent-data-centers-not-77-percent, claim-ercot-queue-only-2-percent-observed-energized-nov-2025, and claim-ercot-large-load-queue-grew-to-239gw-by-december-2025. What settled it: three ERCOT primary PDFs (a Dec 2025 board deck, a Q&A document, and a system-constraints report) were read directly. 226 GW and 63 GW confirmed verbatim. The ~77% data-center-share figure did NOT hold — ERCOT's own deck says ~73%, a genuine correction to the vault's existing note, most likely explained by conflation with an unrelated 77%-solar-plus-storage statistic in the same deck. The requested-vs-built gap is now quantified directly (~2% 'Observed Energized'), though ERCOT's own documents never use the word 'phantom' — the qualitative 'speculative filing' framing remains Tier-2 sourced and is the one sub-thread this question did not fully close.