The American power grid has a paradox: unprecedented investment appetite, record project pipelines — and a connection process measured in half-decades. Understanding why requires holding two institutional readings of the same data at once.
More generation and storage seeks connection than the entire installed U.S. fleet — while the median request-to-operation path has doubled past four years, and historically only about 13% of requested capacity gets built.
Where the institutions split
"Risks are intensifying."
The 2025 Long-Term Reliability Assessment puts 13 of 23 assessment areas at elevated or high risk within five years, projects 224 GW of summer peak growth by 2035 driven by data centers, and counts 105+ GW of confirmed retirements concentrated in already-stressed regions.
"Too pessimistic."
A March 2026 review argues NERC undercounts resources in advanced development and likely double-counts data-center loads that bid into multiple regions simultaneously — concluding most regions can meet 2030 needs.
The megawatts get covered. The study engineers don't.
Every project in that 2,000+ GW queue requires load-flow, short-circuit, and stability studies run by transmission-planning and power-systems engineers at ISOs, utilities, and consultancies — a small, aging talent pool whose skills are learned over years on the job. The queue backlog is substantially a specialized-labor bottleneck wearing an administrative costume.