Power availability, not chip supply, is now the binding constraint on AI buildout

Grid interconnection queues are becoming the real bottleneck for new AI data center capacity.

WI
Wolfsberg Index Staff
September 2, 2026 · 6 min read
AI & Tech

For the first time in the current AI infrastructure buildout cycle, developers say grid interconnection timelines — not chip availability — are the binding constraint on how quickly new data center capacity can come online.

Interconnection queues in several major data center markets now stretch multiple years, a timeline that increasingly shapes where new capacity gets sited rather than proximity to fiber or existing data center clusters.

Some operators are responding by co-locating with dedicated generation, including gas turbines and, in a smaller number of cases, behind-the-meter arrangements that bypass the public interconnection queue entirely.

Utilities, for their part, are recalculating long-term demand forecasts upward after years of flat load growth, a shift that is itself reshaping utility capital spending plans well beyond the data centers driving it.