Less fossil.
Or just a bigger grid?
A falling fossil share can hide rising fossil output. Follow the electricity through 58 grids to see who changed fastest—and what the numbers can actually explain.
Six findings. Every calculation traceable.THE BIG PICTURE
Two measures. Different stories.
01 / FIELD NOTES
Six ways to read the transition.
Fixed 2010–2024 findings.
Choose a note to load its comparison.
02 / THE COMPARATIVE ATLAS
A mix is a moving picture.
Who moved fastest?
Largest declines first. Click a row to focus that grid.
Ten leaders + your selections. Fixed cohort of 58. Negative = a decline.
Full change ledger ↗What generated the electricity?
Domestic generation shares. All bars run from 0 to 100%.
“Other clean” is a calculated remainder including bioenergy. Hatching marks an unresolved breakdown when detailed data are missing. “Clean” includes nuclear and renewables; it does not mean zero impact.
03 / FOLLOW THE ELECTRICITY
Every change has to balance.
An accounting explanation.
It does not identify policy causation.
Waterfalls use independent TWh scales to keep both grids legible; compare the printed values. Positive contributions raise fossil output. Negative contributions reduce it. Rounding residuals are shown when material.
Open the clean-generation ledger
04 / A THOUGHT EXPERIMENT
Can clean additions outrun demand?
Start from the selected observation year. Move two assumptions and watch the accounting unfold over ten years. Both grids receive the same rates, scaled to their own size.
Percent of baseline generation added as TWh each year. This is generated electricity, not installed capacity.
The model is intentionally small. See its assumptions.
Demand compounds at the selected annual rate. Clean output rises by a constant TWh increment each year; existing clean output never retires. Net imports stay at the selected year's observed value. Fossil generation fills the remaining balance, floored at zero. Any excess appears as surplus requiring curtailment or additional exports, not negative fossil generation. No hourly reliability, storage, prices, investment, fuel switching, weather, grid constraints or emissions are modeled. Future points are illustrative calculations, never appended to the observed-data table.
05 / SHOW YOUR WORK
The ledger is part of the story.
— means not reported or not calculable; it is never treated as zero. Reported zero remains 0.000. All generation and demand values are TWh; shares are percentages; changes in share are percentage points. The detailed columns can be missing even when total clean output is known.
A reproducible definition of “fastest”
58 country and territory grids with at least 20 TWh of generation in 2010 and complete total, fossil and low-carbon data in all 15 years. Speed = share change ÷ elapsed years. No regional aggregates or population weights. Alternate baselines keep this same cohort.
Baseline sensitivity table ↗What the data cannot tell us
Annual source data may include estimates and revisions. The export has no row-level uncertainty intervals or estimate flags. Geography, starting mix, trade, weather and demand differ. A ranking is descriptive, not a policy scorecard or a causal estimate.
Independent raw-row verification ↗A pinned, attributable snapshot
Our World in Data, combining Ember (2025) and Energy Institute (2024). Archived at commit f419e8f on 30 May 2025. OWID and Ember use CC BY; original third-party terms remain applicable. This edition deliberately ends in 2024.
