The bigger the project, the likelier it dies.
The bigger the project, the likelier it dies.
Kenneth · 2026-08-12
Source: LBNL Queued Up 2026 Edition (CC BY 4.0), through end-2025. Analysis my own.
A question nobody seems to publish: does project size predict failure?
It does, and the relationship is clean.
| Nameplate | Withdrawn | Built | Failure rate | n |
|---|---|---|---|---|
| Under 20 MW | 1,809 | 574 | 76% | 2,383 |
| 20–99 MW | 1,822 | 407 | 82% | 2,229 |
| 100–299 MW | 993 | 122 | 89% | 1,115 |
| 300 MW and above | 602 | 89 | 87% | 691 |
**A sub-20 MW project is more than twice as likely to reach service as one over
100 MW** — 24% against 11%.
The intuition runs the other way. Large projects have serious developers, real
capital and professional interconnection counsel behind them. Small ones are
often speculative.
The data says the advantage sits with small anyway, and the plausible reason is
network upgrade cost: **a large interconnection request is likelier to trigger
system upgrades whose allocated cost kills the economics.** That is a hypothesis.
This dataset does not contain upgrade costs and I have not tested it.
If you are siting, the queue rewards being small more than it rewards being
good. Four 40 MW projects have materially better odds than one 160 MW project,
for the same capacity.
If you are underwriting, size belongs in the model. A 300 MW request priced
at the composite 82% is being priced at the wrong end of a seven-point spread.
If you hold a large position that survived, you are rarer than you think.
*Method: terminal outcomes only, bucketed on mw_1. Requests without a capacity
figure excluded. Free at emp.lbl.gov/queues.*
*This is published research. Every figure is a count from a free dataset at emp.lbl.gov/queues — check it yourself rather than taking my word for it.