I've read about a couple of solar projects that projected x/kWh and turned out to be much worse in production.
Pakistan's big deployment is one example, it was planned to be 100MW but only produced 18MW. Mostly because they didn't factor in the dust that would cover the panels, keeping them clean was very expensive (I'm curious how this applies to other proposed desert deployments). Nor factoring in the typical malaise of public projects adding significant delays/costs, which is hardly unique to Pakistan when it comes to major infrastructure in 2020.
https://en.wikipedia.org/wiki/Quaid-e-Azam_Solar_Park
Then there's the famous Solyandra which got half a billion in US tax payer money and flopped without ever being competitive price-wise: https://en.wikipedia.org/wiki/Solyndra
Tesla's famous Australian battery factory only has enough to power 30k homes for about 1-2hrs each day. It was on-schedule, which is rare, and also makes money but scaling it up to millions of homes to make a real dent in coal would be much more challenging, especially sourcing enough lithium.
I think it's safe to take any cheery prediction from green projects and add 1.5x time/production costs and/or minus 30% of the expected output.
Note: I'm not trying to rain on the parade, I'm otherwise all for this stuff as long as it's realistic on a large scale.