By "cheaper power", you mean: building efficient solar panels that can withstand launches, building satellites and rockets, filling the rockets with rocket fuel, burning the rocket fuel to put the solar panels into orbit, recovering the rockets and refurbishing them after launch, then constantly managing the satellite orbit to optimize sun exposure while also cooling the electronics, and eventually watching it all burn up in reentry? Are you sure that power is cheaper than $0.20/kWh?
As for the NIMBY factor... How many satellites are we talking? Just to make it fun, let's say there are 10 million of them we want to launch. To place them on earth, we could find any number of 5x5 km plots of land in the middle of several deserts around the globe, where they could be installed hundreds of km from anyone, where no one would be impacted.
Placing datacenters in any of those remote locations would still be easier and cheaper than this crazy idea of launching them into space. Yes, there are challenges with cooling in those places, and there are challenges with access and maintenance, but any of those challenges are so much more solvable than doing the same thing in space.
Anyone who thinks space is easier in any way just doesn't understand the magnitude of the project.
But I think that's the point: people can understand what a hot day is or what it means to have a 500 mile drive for a technician to repair something-- they have a gut feeling that putting a datacenter with 10 million nodes into the middle of the Sahara desert would be so difficult it would be infeasible. By abstracting it away into a realm that people don't really understand, it seems to be easier to bamboozle potential investors who can be dazzled into thinking it can happen.