Taiwan

The Expected Buildout line answers one question: of the pipeline announced today, how much actually gets built, and when? It is not a scenario and not a target tracker — it is the database's live pipeline, passed through two filters measured from the industry's own delivery record.

Whether a project completes. Every project in development carries a completion weight, banded Delivering / Expected / Prospective / Fragile. A proposal — a site where no authority has yet granted rights — carries none and sits outside this line entirely; it is listed, not counted. The starting point is the historical completion rate of its lifecycle stage — of projects that reached construction, 97% delivered; financed but not yet building, 92%; consented without financing, 75%; early development, 61%; on hold, 15%. That base rate is then adjusted by what the project has actually secured (offtake, grid, consents, investment decision) and decays if a project sits quiet past its stage's normal dwell time.

When it delivers. Each project also carries a completion-date distribution — a shape, not a single year, right-skewed because offshore wind slips late far more often than it runs early. The peak sits on a rule-based expected date chained from the project's latest achieved milestone; the width of the late tail is fitted to how far past their announced dates real projects actually landed: builders slip tight (68% on time, n=158), financed projects tighter still (72%, n=101), consented-only projects carry the fat tail — 25% land three or more years late (n=133). A stated date faster than any comparable project has ever achieved is floored to the achievable window and flagged, never taken at face value.

The line is the sum, across every live project, of capacity × completion weight × the probability it delivers in each year. For projects already under construction, a share of expected capacity is counted at first power — turbines start generating before a wind farm is fully commissioned — so a market's chart can show capacity arriving a year before a project's stated completion date. Delivered history is metered fact; the line inherits every assumption above and nothing else.

Honest limits. The weights are calibrated on a fleet whose failures have so far been overwhelmingly pre-financing — completion rates after FID rest on a record with few failures, so they are ceilings, not guarantees. The slip fitted on consented projects is the loosest (the "three or more years late" bucket spans many years). Policy shocks arrive faster than base rates move — the weights see a stop-work order only as its lifecycle consequences accrue. And the line redistributes as the database learns: it is re-derived on every data refresh, not frozen.

Full per-project workings — each project's weight, band, expected date and distribution — are on its project page. Calibration v2.0-phase0 · generated 2026-09-09; every figure above is read from the generated calibration file, never typed.