In his article "2 remarkable facts that illustrate solar power’s declining cost" Roberts starts off well:
First, there’s no such thing as an electricity source that is cheapest in all circumstances, nor is there likely to be such a source any time soon. All sources have advantages and disadvantages; they all have circumstances in which they excel.
But goes downhill from there, concluding with the following nonsensical remark:
“Solar is winning.”
What does that even mean?
Industrial solar is going to play an important role in future grids but unless it’s located in one of the sunniest of places, it won’t be the least expensive source. And even when it is the least expensive source, it will never be the only component of a grid. More expensive components, be they nuclear, wind, gas, and possibly some amount of much more expensive storage, will have to be part of that grid ...especially at night. So, what matters is the total cost of operating all of the components of a grid. The fact that you may have lowered the cost in very sunny places of just one component (solar) will not necessarily lower your electric bill. Evidence to date strongly suggests the opposite.See Figure 1.
Why? One reason would be integration costs, which are not included in LCOE calculaltions. The price charged by the solar farm owner does not reflect all costs to the grid operator (which all get passed on to the consumer) of the additional transmission needs of industrial intermittent sources. The integration costs for wind can double the final price charged to customers. If the wind farm is charging $22/mWh and the cost of installing new transmission lines to export excess power produced when it's not needed locally is $22/mWh (see Figure 2), the real cost is twice that being charged by the wind farm. Integration costs of intermittent sources tend to be much higher than for dispatchable sources because of the need for more transmission infrastructure to export excess power produced at times when it is not needed in the local grid.