Interconnection Queues Explained
Interconnection Queues Explained
If you want to understand where the AI data center buildout will actually slow down, stop watching announcements and start watching interconnection queues. This is the paperwork bottleneck standing between a planned data center and a live one — and it is the single best predictor of which projects are real.
What Is an Interconnection Queue?
An interconnection queue is the formal process a grid operator or utility uses to study, approve, and schedule any new large electricity load — including AI data centers — before it can connect to the transmission or distribution system. Every request has to go through engineering studies to confirm the grid can safely handle the added demand, and often what upgrades (new substations, transmission lines, transformers) are required first. In most of the US, this is a first-come, first-served process organized in "batches" or study cycles, though several grid operators are moving to readiness-based models that require proof of site control and financing before a project holds a queue slot.
Typical Timelines and Attrition Rates
The headline numbers in interconnection queues are almost always requests, not reality — and the gap between the two is enormous.
- Nationally, the median time from queue entry to commercial operation reached 55 months (roughly 4.6 years) in 2024, up from 36 months in 2015 and 22 months in 2008, according to Lawrence Berkeley National Laboratory's annual "Queued Up" report.
- Of projects that entered US interconnection queues between 2000 and 2019, only about 19% reached commercial operation by the end of 2024; the rest were withdrawn or are still waiting.
- Attrition is even starker in the two markets most relevant to AI data centers: Carbon Direct's 2026 analysis found only about 40% of ERCOT projects and 24% of PJM projects that began the queue process by 2020 had reached an interconnection agreement or gone operational — meaning 60% and 76%, respectively, had not (and most never will).
- The Federal Energy Regulatory Commission targets 8–11 months to reach an interconnection agreement. Actual average wait times run to roughly 20 months in ERCOT and 40 months in PJM, with active large-load projects in data-center growth zones waiting 36–48 months.
The takeaway: a headline queue number like "400+ gigawatts of requests" tells you almost nothing about how much power will actually be delivered, or when.
Why Queues Matter More Than Power Prices for AI Infrastructure
Site selection teams and investors often focus on electricity rates, but rates are irrelevant if a facility cannot get connected to the grid at all. Queue position, transmission headroom, and utility capital plans determine whether a project can energize in two years or ten. A data center with cheap power quoted on paper but no near-term interconnection slot is a stranded asset until the grid catches up — and that risk is exactly what shows up later as delayed go-live dates, cost overruns, and, for public companies, guidance misses.
How Gridlas Uses Queue Data in the Powering AI Report
The Powering AI report tracks interconnection queue depth, batch-study timing, and attrition by region so readers can separate speculative site announcements from projects with a realistic path to power. Rather than repeating a utility's press release, we go to the underlying queue filings, rate-case dockets, and grid-operator reports to show what's actually been approved to energize versus what's simply been requested.