Grid access · resource ledger
Interconnection Queue
AI did not create the five-year generation queue, but it is forcing a separate large-load reckoning.
Scapegoated5 yearsmedian request-to-operation time for projects built in 2023
What the evidence supports
LBNL found a five-year median for generation projects completed in 2023, with nearly 2,600 GW waiting—over 95% zero-carbon resources. That structural backlog predates today’s AI wave. FERC is now separately reforming large-load connection rules.
- Mechanism
- Generation queues study projects supplying the grid; large-load processes study customers drawing from it. They interact, but collapsing them into one ‘AI queue’ is a category error.
- Who pays—or gains
- Delayed generators, large loads and ultimately customers pay for slow studies and network upgrades. Better queue management is a supply-side reform with broad benefits.
- Binding constraint
- Study throughput, speculative projects, transmission planning, cost allocation and incomplete tariff rules for flexible or co-located loads.
- Strongest caveat
- The scapegoated grade applies to the claim that AI caused the existing five-year generation queue—not to the claim that new data centers add grid work.
- What would change the grade
- Regrade if comparable queue data show AI-driven large-load studies displacing generation projects or materially lengthening completion times.
Evidence file