Mapping Solar Power
The grid and the market cannot see rooftop solar. They only see the demand that never arrives.
Rooftop solar is now a large share of installed capacity, and almost none of it is metered in real time.
It never appears in a generation feed. It shows up as demand that did not arrive, in exactly the sunny hours when solar already sets the price.
Residual load forecasts — demand minus wind and solar — start from official capacity figures. Those figures lag the real installed base, so the error is built in before the first cloud.
Distribution network operators face the same blind spot at street level. Without knowing where the panels are, they cannot tell how much power flows back through each feeder on a sunny afternoon.
We apply AI to satellite and aerial imagery, and to the gap solar leaves in measured load, to estimate how much capacity is really installed, where it sits and what it produces.
That capacity base feeds our forecasts, so traders and utilities work from residual load built on the real fleet, not last year's register. Grid operators get a map of where solar sits, so they can plan the network and spend on hardware only where it is needed.