8. The grid gap
The average grid got 18% cleaner since 2000. The gap between the cleanest and dirtiest grids closed by only 8%, so the penalty for being at the wrong end is not going away.
Every other finding is about direction: who fell, who rose, who decoupled. This one is about spread, and it points the other way.
The panel is 80 countries, each with a grid above 10 TWh in every year since 2000.
The three lines fall roughly in parallel, so the distance between them barely changes. The dirtiest grids have come down by about a tenth, and the cleanest were already near the floor.
Because the mean fell faster than the spread, the spread relative to the mean rose by 23%: in proportional terms the world's grids are further apart than in 2000. Neither reading is an artefact of the start and end points: fitted across all twenty-five years, the fall in the mean and the rise in relative spread both carry p-values below 0.001.
So what. A transition that improves every grid at roughly the same proportional rate leaves the ranking intact, and with it the penalty for being at the wrong end. Anything priced off where electricity is consumed (the Scope 2 line of a disclosure, the embedded carbon in an imported tonne, the siting of a plant) does not get cheaper to get wrong as the world decarbonises. On this trajectory a location premium is a permanent feature of the next two decades.
Who acts: whoever signs off site selection, long-term supply agreements or a decarbonisation roadmap that assumes convergence. Cost of getting it wrong: building a twenty-year plan on the expectation that the gap closes on its own.