Definition

The administrative and engineering process by which proposed power generation, storage, or large-load projects request connection to the US transmission grid. Each Regional Transmission Organization (RTO) — PJM, ERCOT, MISO, SPP, ISO-NE, NYISO — maintains its own queue. Projects are studied for grid impact, assigned interconnection costs, and scheduled for energization. The queue is public, updated quarterly or annually, and dispositive: a project with PPAs but no queue position is a press release; a project with queue position is real.

Why It Matters for the Newsletter

The interconnection queue is the document class that decides which AI data center projects actually deliver in 2026-2030. Financial press AI capex coverage rarely cites it; the queue is the first place the binding constraint shows up and the last place chip-layer analysts look. Vocabulary central to the AI Buildout Grid Constraint thesis. Provides the analytical edge for the operator-class reader: read the PUC and FERC filings, track the queue depth, ignore the press release.

Evidence & Examples

Tensions & Counterarguments

  • Vertical integration as queue bypass. Google’s acquisition of Intersect Power and Microsoft’s Brookfield deal both sidestep traditional interconnection queues by co-locating generation with load (behind-the-meter). Whether this becomes the dominant pattern or remains supplemental to grid-scale connection is open.
  • Withdrawal-rate framing. The 70% withdrawal rate is sometimes cited as evidence the queue is “self-cleansing.” LBNL’s analysis suggests instead that withdrawal indicates structural failure: projects reach the front of the queue at price points or timelines that no longer support the economics that justified them.
  • The queue isn’t a single entity. “The queue” is shorthand for seven different RTO queues running on different rules; reform of one (FERC’s December 2025 PJM order) doesn’t reform the others. Cross-RTO comparison requires reading each separately.

Key Sources