You need to consider short term intermittency (the wind not always blowing and the sun being down 50% of the time).  That needs to be covered by some form of short term storage, most of which are expensive at scale.  Or long distance grid interconnects.

But worse than that, you have, at least in some areas, massive seasonal variations.  I've looked at, but can't find it right now, a document by California's Energy Commission where they claim a factor of 5 difference between peak solar and wind production at high vs low times of the year.  That seemed high, for California's weather, but higher latitude zones have massive summer/winter solar variations (German winter sun?  hah!).

That's something that no storage is going to help with, you'll need to either source elsewhere or get around by overcapacity.

These factors don't show up that much when renewables are a small proportion, but they become more important as fossils are retired entirely (which they should).  The bottom line is that, right now, seasonal variations and day-to-day intermittency is going to make a full-renewable system difficult to pull off.

Let's not forget that Germany's energy wende swallowed 130B$ and then still increased emissions, because their baseline backup went to coal, some of it lignite.

I am all for taking global warming seriously, but let's not repeat the silliness of the ethanol subsidies for dubious gains once the full cycle of production is accounted for.  Every dollar badly spent is a dollar not available for better solutions.  We can't afford to do that very often.  Even replacing older coal plants with natural gas is risky if it locks us into natural gas for the next 50 years - natural gas is not carbon neutral by any means.

In a perfect world, we'd gradually increase carbon taxes and fund better systems from revenue.  In a less perfect world, we'd do revenue-neutral carbon pricing + dividends.  In our current world, we don't price carbon, but subsidize various technologies, not all of which are grid scale ready and not all of which *actually reduce emissions much*.

We really need to get it right, because the CO2 numbers of the chosen solution mix make sense, not just because it *feels* right.  And even the $ numbers need watching, because the wrong tech will not scale outside rich countries.