If you've been paying attention to the energy sector lately, you've probably noticed something odd. We're building data centres faster than ever, but the electricity to power them? That's stuck in traffic.
That's how we wrap up PJM interconnection queue crisis of 2024-2025.
The Numbers Are Staggering
PJM, the grid operator serving 65 million people across 13 states, now has roughly 195 GW of projects stuck in its interconnection queue. That's not a typo. We're talking about enough generation capacity to power most of the Eastern Seaboard... just sitting there, waiting for approval.
The timelines are brutal. Projects are waiting 4 to 5 years just to get through the queue. And according to GridLab's analysis, this backlog is costing ratepayers approximately $3.5 billion in unnecessary capacity costs.
Data centers are responsible for about 94% of PJM's projected load growth through 2030.
Let that sink in.
Interconnection Queue Capacity by ISO/RTO
Active queue capacity in GW, 2024-2025 | Data centers driving unprecedented backlogs
Why Data Centers Are Breaking the Grid
Northern Virginia's "Data Center Alley" alone hosts over 586 operating data centers. The AI boom isn't helping either. Every ChatGPT query, every cloud backup, every Netflix stream... it all needs power. A lot of it.
PJM's 2025 Long-Term Load Forecast projects summer peak demand climbing by 70 GW over the next 15 years. For context, their all-time record was 165 GW back in 2006. Now they're forecasting 210 GW by 2035.
The market monitor put it bluntly... data center load accounted for $6.5 billion (40%) of the $16.4 billion in costs from the December 2025 capacity auction.
How Does PJM Compare to ERCOT and NYISO?
It's not just PJM struggling here.
ERCOT in Texas saw its large load queue grow by 300% in 2024 alone. They're now assessing over 233 GW of large load interconnection requests, with more than 70% coming from data centers. The difference? ERCOT isn't subject to FERC rules, so they operate under their own playbook.
Data Center Share of Load Growth by Region
PJM's data centers account for 94% of load growth through 2030, adding ~30 GW of demand. This concentration creates unprecedented grid strain.
NYISO in New York has a different challenge. As of mid-2025, they have 29 large load projects in queue totaling nearly 6,055 MW. The average wait time for active projects is 6.53 years... with some stretching to 7.4 years. Not exactly startup-friendly timelines.
The good news? In 2024, interconnection queues across the U.S. actually shrank for the first time in years.
Why? Not because demand disappeared, but because stricter rules cleared out speculative projects. Higher deposits, milestone requirements, and FERC Order 2023 reforms are forcing developers to get serious.
China's Approach Looks Different
Meanwhile, China is taking a different path entirely.
They launched the "Eastern Data and Western Computing" initiative, relocating data centers to western regions with abundant renewable energy, natural cooling, and cheaper land. By 2025, total data center power consumption in China is expected to reach 100 to 200 terawatt-hours.
Queue Processing Timeline Comparison
NYISO projects wait 6.5+ years on average. China's centralized "Eastern Data, Western Computing" approach enables parallel buildout, avoiding queue bottlenecks entirely.
The critical difference? China's central planning allows grid infrastructure to move in parallel with data center buildout. No queue backlogs. No 5-year permitting delays.
Is it replicable in the U.S.? Probably not. But it's worth noting that while we're stuck arguing about transmission rights, they're building.
What Needs to Change
PJM isn't sitting idle. ISOs should be partnering with companies like Espresso for processing interconnection applications. But the real solutions require structural reform. Faster software-driven study timelines. Parallel study paths for large loads. Integrated planning models like SPP's Consolidated Planning Process.
The bottom line is simple. Data centres can be built in 18 months. New power plants take 5 to 7 years. That math doesn't work... and until we fix the queue, we're going to keep seeing capacity crunches and rising electricity costs.
The AI revolution needs electrons. And right now, those electrons are stuck in the queue.
About the Author
Dr. Sayonsom Chanda
Dr. Sayonsom Chanda is an electrical engineer and senior scientist with more than a decade of experience in developing AI, ML, and other advanced computing solutions for the electric utility industry in US and India. He is also an energy policy thinker and a published author with more than 20 papers and 1 book.
