AI Infrastructure Energisation Gap
Share of publicly announced AI/datacentre MW with explicit evidence of energisation in the open records of each covered market, reported separately per market
US · observed cohort
70%
Texas observed cohort — 2 of 8 projects, single issuer (IREN)
Milestone: substation energisation — evidence found, not capacity operating
- With energisation evidence
- 1,400 of 2,000 MW
- Stage unknown
- 30% of the cohort — no disclosure either way
- As at
- 2026-09
Not published for these capacity bases, because every project in them has an unknown stage: IT capacity (50 MW); Gross connection capacity (4,230 MW)
Eight projects from two issuers, all in Texas. This is an observed cohort and is not a United States denominator. No national ratio may be computed from it.
Definition
- What it measures
- Share of publicly announced AI/datacentre MW with explicit evidence of energisation in the open records of each covered market, reported separately per market
- Headline metric
- Energisation Evidence Ratio, % of tracked MW, stated per market (US panel leads; GB panel second)
- Supporting metrics
- per market: MW with energisation evidence; MW approved but not energised; MW in study or review; MW with no studies submitted; unknown-stage MW; record coverage. US stage ladder as published by ERCOT: Observed Energized / Approved to Energize but Not Operational / Planning Studies Approved / Under ERCOT Review / No Studies Submitted
- Cadence
- Monthly
- Panel
- Deliverability
- Primary sources
- ercot Large Load Interconnection Status Update (US); neso exact Open connection datasets and ofgem OGL records (GB); cited public announcements for the announced-MW denominator
- Licence position
- US: clear – ERCOT Terms of Use section 5 reuse grant, retrieved 2026-09-10, re-verified twice yearly. GB: clear for exact NESO Open/OGL records. Announcement facts: citation-level only.
- Release gate
- every row has a cited record, stage evidence and observation date; unknown stage reported separately; each market panel carries its own market label, source and stage-definition note; no cross-market blending; the page states which markets are covered and does not describe the figure as global
What this does not cover
PROVISIONAL EVIDENCE PROXY. A missing public energisation record is unknown, not proof of non-energisation; queue position is not energisation. Two markets are not the world: US and GB stage definitions are not identical and the panels are not comparable like-for-like. Covered markets are named on the page and the figure is never labelled global.