I suspect the 'multi-component' learning curves is the main driver of the slowing rates. Most of the media follows the mass manufactured components learning rate, rather than total installed cost as well, which makes the learning look faster than the realized cost improvements actually are...
The next biggest is outright physical limits. We can do no better than Carnot efficiency, which puts a hard cap on one improvement factor for example.
I suspect the 'multi-component' learning curves is the main driver of the slowing rates. Most of the media follows the mass manufactured components learning rate, rather than total installed cost as well, which makes the learning look faster than the realized cost improvements actually are...
The next biggest is outright physical limits. We can do no better than Carnot efficiency, which puts a hard cap on one improvement factor for example.
That is very likely the case. I think about how, for residential solar PV, soft costs rather than the cost of the panels themselves are now dominant.