https://www.nrdc.org/experts/kevin-steinberger/debunking-thr...
For all that article's efforts to quibble with definitions, the fact is that a grid that is all or mostly renewables (except for hydro and geothermal, which are great for the places that are suitable for them, but not available everywhere) will have blackouts. Lots of things can be demand-managed but lots of things can't. If you want electricity to be available 24/7 then the grid absolutely does need to "overpay" for generators that are available 24/7 (and sure, nuclear plants have unexpected shutdowns like anything else, but those outages are going to be uncorrelated with each other), and whether you call that "baseload" or some other term is neither here nor there.
I’m even hopeful of a future where these natural gas plants will have carbon capture employed. But you are right, as it stands natural gas power plant is not a solution to the climate crisis.
Firstly, you talk about $ per capacity in MW, however solar capacity factor is 20%. Only use a levelized cost per MWh.
Secondly, solar needs storage which you think is trivial, but which turns out to be wayyyy more expensive than the solar panels. Some reports[1] only compare short-term (peaker generation) for storage. Nuclear is expensive, but battery storage is far more expensive to cover daily or longer load variation (non-peaker). That report mentions in a footnote a levelized cost of storage of crazy high $1613/MWh to $3034/MWh.
[1] https://www.lazard.com/perspective/levelized-cost-of-energy-...
You used MW in that comment - your new comment is not clarifying whether you made a mistake or what your actual point is.
LCOE is an averaged cost/MWh, which is absolutely inappropriate to use when discussing whether nuclear is expensive or not, because LCOE ignores usage patterns (which you obviously know, but are hand-waving away as “money left over”). From your Wikipedia link: “One of the most important potential limitations of LCOE is that it may not control for time effects associated with matching electricity production to demand” and “To ensure enough electricity is always available to meet demand, storage or backup generation may be required, which adds costs that are not included in the LCOE”.
Finally, almost any plot that uses “installed capacity” is deceptive by design: because capacity factors make solar/wind appear 5x better, which is not a trivial difference on a graph. Instantaneous capacity is usually not relevant (except during “peaker” loads).
So no, we are not going to be needing an absurd amount of any one mineral (except maybe silicon) as they all have alternatives which quite often are even better then lithium and cobalt.
If storage is cheap enough then solar will be an important part of the mix. But unless you have the right geography for geothermal or hydro (which probably means flooding a bigger area than the chernobyl exclusion zone before you've even started, but leaving that aside), nuclear is still the only viable option for clean, reliable baseline power. It's not an either/or, we should be building both.