Wheredatacenterscanactuallygetpower.
GridSolve re-scores every substation in the U.S. interconnection queue by withdrawal-adjusted capacity — the megawatts likely to remain once the dead projects clogging the queue drop out.
Utility hosting-capacity maps count dead queue projects as if they were real.
A developer looking at published capacity sees megawatts already spoken for by speculative projects — most of which will never be built. Promising sites look full; full sites look promising.
Lawrence Berkeley National Laboratory finds roughly 75% of proposed capacity in U.S. interconnection queues ultimately withdraws before reaching commercial operation.
In this data, GridSolve's model estimates a base withdrawal rate of 61%. Adjust for that, and the map of where power is actually deliverable looks very different.
Likely to drop from the queue
Realistically available power
Only the deliverable share is power you can actually count on.
From a clogged queue to deliverable megawatts, in three passes.
- 01
Model the queue
A survival model is fit on historical interconnection projects — who withdrew, when, and why.
- 02
Predict withdrawals
Each active project gets a withdrawal probability from its cluster, service type, and technology.
- 03
Re-score capacity
Published capacity is discounted by expected withdrawals to yield deliverable megawatts per substation.
A few features separate the projects that get built from the ones that vanish.
Hazard ratios above 1 raise the chance a project withdraws. Projects tangled in a cluster study are the clearest red flag — the queue's equivalent of a crowd that never moves.
What predicts withdrawal
hazard ratioSee which substations still have room.
Explore substations nationwide ranked by capacity that's actually deliverable — not just what the queue claims.