Run the decline backwards. A pool 10 percent larger in 2024 stood near 2,290 GW, so the net reduction was about 229 GW [1] against the 750-plus GW that walked away [5], which leaves roughly 520 GW of fresh requests arriving behind the departures [2]; that residual also absorbs whatever actually energised, and by Lawrence Berkeley National Laboratory's own history that is the smallest term in the sum [10]. The pool now sits about 21 percent below the end-2023 peak, a drop of some 539 GW [3]. The line got shorter because developers left it.
Gas did most of the arriving. An 86 percent increase implies a base near 136 GW, so roughly 117 GW of gas requests entered in a single year [4], while solar shed about 181 GW and storage about 143 GW [5]. Newness has a price in a queue: 18 percent of gas capacity holds a draft or executed interconnection agreement, some 46 GW [6], against 32 percent of solar and 33 percent of wind [8]. The capacity added fastest is the capacity furthest from a signed contract.
Anything underwritable sits inside the 536 GW that already has a draft or executed agreement [8], and applying the lab's post-agreement attrition takes that to roughly 322 GW [7]. Push the 2000-2020 conversion rate across the whole active pool and the implied eventual build is about 268 GW [8]. LBNL's data covers about 98 percent of installed US generation, compiled with Interconnection.fyi across seven RTOs and 50 further balancing areas, so this is the national picture rather than a regional slice [13]. A queue position, the lab is careful to say, still needs landowner agreements, permits, a power purchaser, equipment suppliers and financiers [14].
The reform reading has evidence behind it: record agreement volumes in four markets [9], and FERC's 2023 order in force across most regions since the following year [4]. LBNL itself goes no further than saying the reforms and market conditions "may have played a role" in the exits [5], which is the defensible position, because in queue data a developer who quit over process cost and a developer who quit because the offtake never appeared produce the same minus sign. What the numbers support is a less speculative queue rather than capital moving from waitlist into construction, and the lab's caution that most requests are never built [12] applies to the smaller pool as much as it did to the peak [3].
Two lines in next year's edition settle it. Post-agreement withdrawal below 40 percent, with gas's agreement share climbing off 18 percent, would mean throughput genuinely improved; another year of heavy exits against a flat agreement-backed pool would mean the queue is simply clearing out.
