Absolutely, designing a mix for today's demand is guaranteed failure.

Designing a mix for a target date with predicted demand characteristics actually has a shot at working. We can make quantified statements about distribution estimates for demand that includes quantified variability and create a mix target that meets our tolerance for risk, cost, and sustainability.

I frankly think that armchair debate over solar vs nuclear - which is 99% of these discussions - has exactly zero probability at arriving at an optimal mix. Pigeon, chessboard, &c.