Grid Interconnect Queue

The electricity grid interconnection queue is the process by which new generators (solar, wind, nuclear, gas) apply to connect to the transmission grid and receive permission to inject power. In the US, this process is managed by regional grid operators (MISO, PJM, CAISO, ERCOT, SPP, etc.) and is fundamentally broken at scale.

As of 2026, the US interconnection queue has swelled to a 2,600 GW backlog — a figure that exceeds total installed US generating capacity. Nearly 80% of queued projects never reach commercial operation (they withdraw due to cost and timeline uncertainty). The median wait time to reach commercial operation is approaching five years; some segments face 12-year timelines. Interconnection costs can consume 30-37% of a project's total budget.

This is not a niche regulatory issue. It is the physical bottleneck that prevents new power supply from reaching AI datacenters and other load growth, even when capital is available and generation is contracted. A hyperscaler that wants 500 MW of new nuclear power delivered by 2028 cannot have it — the grid simply cannot absorb it that fast.

FERC Order 2023 (2023) required reforms to interconnection processes — replacing the serial "first-come, first-served" queue with cluster-based studies. Implementation is ongoing but slow. PJM reopened its long-closed interconnection queue in April 2026 with a promised 1-2 year study timeline, down from the prior 5+ year backlog.

Key Claims

• US interconnection queue: 2,600 GW backlog as of 2026. Evidence: strong (Berkeley Lab LBNL Queued Up, Enkiai)
• Over 2,060 GW of capacity actively seeking interconnection as of end of 2025. Evidence: strong (Zero Emission Grid)
• Median wait time approaching 5 years; some datacenter projects face 12-year timelines. Evidence: strong (Enkiai)
• ~80% of queued projects withdraw, primarily due to unpredictable delays and high grid upgrade costs. Evidence: strong (LBNL)
• Interconnection costs = 30-37% of total project budget for withdrawn projects. Evidence: moderate (Novogradac)
• PJM reopened its queue in April 2026 with reformed cluster-based study processes. Evidence: strong (Advanced Energy United)
• FERC interconnection reform (Order 2023) mandates cluster studies — policy fix, but grid physics and labor constraints slow implementation. Evidence: strong (FERC)
• The queue is precisely why hyperscalers buy nuclear directly: nuclear PPAs (restarts, co-located SMRs, advanced reactors) are how they secure firm, carbon-free baseload outside congested interconnection queues. Evidence: moderate (multi-source synthesis, hyperscaler-nuclear-procurement)

Why This Matters for AI Power

The interconnection queue problem is structural and not solvable by capital alone. Even a perfectly-funded nuclear restart or new SMR faces the queue. Three workarounds are being tested:

1. Co-location — Site the datacenter directly adjacent to the generator (Amazon/Talen Susquehanna; Talen/Amazon model), bypassing the transmission grid entirely and using a direct connection.
2. Behind-the-meter — Generator and load share a common point of connection, sidestepping the queue but creating regulatory complexity about whether this constitutes "retail" power sales.
3. Existing interconnection rights — Buy or transfer existing Capacity Interconnection Rights (CIRs) from retiring plants to new projects (the Constellation/FERC waiver model for Crane, transferring 760 MW of CIRs from Eddystone to TMI).

Open Questions

• Can FERC's Order 2023 reforms actually reduce queue timelines to 1-2 years at scale?
• What is FERC's position on co-location / behind-the-meter datacenter arrangements? (Active rulemaking as of 2026)
• How does transmission capacity expansion (HVDC, grid modernization) change the calculus over 10 years?
• Does load growth itself break the queue math? (If 50 GW of new datacenter load needs interconnection concurrently, the queue explodes)

Related Concepts

• Datacenter Power Demand — the load that is colliding with the queue
• IPPs & Merchant Power — companies navigating the queue
• Baseload vs Intermittent — queue applies to all generators

Changelog

• 2026-06-15 — Initial compilation from web research (LBNL, FERC, Enkiai, Novogradac, Advanced Energy United)
• 2026-06-24 — Compiled new sources (hyperscaler-nuclear-procurement)