This is a very important point. With load-following natural gas (and some coal) plants available, the returns on renewable energy are excellent, and the cost per kW "average" output is quite low. These values cannot be linearly extrapolated out to 100% capacity without energy storage which, even if a viable solution is invented, will not be free.
Bear in mind also that current wind and solar installs are being built in the most economical locations to build wind and solar: places that get good amounts of average wind and average light, and which are close enough to populated areas to avoid lots of transmission loss.
Future wind and solar installations (albeit with some benefit from hypothetical improvements in PV efficiency) will not be as efficient on a per-unit basis, and utilities will increasingly have to build not to the average power output (made possible only because of load-following natural gas, as well as coal) but rather to some low percentile.
An example: if a 100% renewable grid must service an area with a flat average demand of 500 GW, then if renewable output is greater than 500 GW 95% of the time, the other 5% of the time will require firing up a fossil source or turning off customers' power. And this assumes demand is constant, which it isn't-- demand peaking is just as much of a problem as supply dipping.
Bear in mind also that current wind and solar installs are being built in the most economical locations to build wind and solar: places that get good amounts of average wind and average light, and which are close enough to populated areas to avoid lots of transmission loss.
Future wind and solar installations (albeit with some benefit from hypothetical improvements in PV efficiency) will not be as efficient on a per-unit basis, and utilities will increasingly have to build not to the average power output (made possible only because of load-following natural gas, as well as coal) but rather to some low percentile.
An example: if a 100% renewable grid must service an area with a flat average demand of 500 GW, then if renewable output is greater than 500 GW 95% of the time, the other 5% of the time will require firing up a fossil source or turning off customers' power. And this assumes demand is constant, which it isn't-- demand peaking is just as much of a problem as supply dipping.