As for cost of delivered energy from solar, there is a final reduction that can still be obtained even if the cost/watt were to plateau immediately: extension of the lifespan of the system. This could be obtained by, over time, identifying and addressing all the degradation modes of the components of the system.
But I also think you are mis-applying a mental model about the cost of manufactured goods to the cost of solar plant development. In a competitive free market, the cost of manufactured goods should eventually be reduced to a sum of their inputs plus any profit.
But solar development is inherently an exercise in value capture. The project developers are trying to capture as much value as possible for their investors. The only thing that matters is that solar stays competitive in the energy market. As long as that happens, someone somewhere in the value chain can try and capture more value as other costs decrease. For example, what is stopping land owners from raising rents for solar projects? Right now, land is relatively cheap because there is not a lot of competition for it. But there are only so many parcels that are near transmission lines with good geographical features, and I fully expect land prices to increase as solar penetration increases. But this is really just a guess.
If you are interested in current project cost breakdowns, figure 30 in this report does a pretty good job[1]. Module costs were the primary driver of cost reductions in the last 20 years. But they now account for only 40% of total project costs. So even if module prices fell by 50%, that would only result in a 20% decline in total project cost. While a 20% drop in prices would be nice, I don't think it really changes the math much with respect to nuclear (the math is already overwhelmingly in favor of solar).
Also, I do agree that the average lifespan of these systems are going to be much longer than 20 years and that has not been priced in. But I don't think that is going to really change things anytime soon. Financial models and risk models are incredibly crude instruments. Using them to model even 5 to 10 years in the future is already extending them well beyond what they were designed for. But we have to put an end date on these models because at some point every project is going to cease to exist. And 20 years is what the banks like, so that is what we use.
I will note that there have been past attempts to call a floor in the cost of PV, for example the period from 2002-2007. That was not the floor.
Current cost breakdowns are nice, but don't show that costs can't be further reduced.
> For example, what is stopping land owners from raising rents for solar projects?
What stops them is the enormous quantity of land that's available. In the US, for example, land can be < 1% of the cost of a utility-scale PV installation.
