Resistance losses aren’t a huge problem (standard currently available HVDC can do 3.5% loss per 1000km, or 1-(0.965^20) = 51% loss between opposite points of the planet, which is fine given how cheap PV is).
The cost for both is the big thing: even just assuming overnight electricity replacement, 12 hours * 2 TW * $100/kWh = $2.4 trillion; the 8 GW English Channel link cost $1.6/km[0][1], which a naive scaling up to 20 Mm and 2 TW would cost $8 trillion. Then the question is “which lasts longer, and which is harder to refurbish at end of life?”. (Numbers are just a guideline. On the one hand there is less demand at night and there will probably be wind and geothermal in the mix, on the other hand we want something that allows everyone to have an energy use currently only available to rich people and this doesn’t cover anything except mere electricity).
I have not been able to find the size of the global energy industry in dollars per year to compare this to, only an un-cited claim of “$5 trillion per year … subsidizing the fossil fuel industries”.
That said, at this scale I have to ask about the economics of making an orbital ring, transferring power kinetically in what is essentially a ballistic superconductor[2], and now you can put the PV in space on the ring without having to solve the political aspects of “can I convince everyone else I’ve not sneakily designed my power satellite to be used as a death ray, and can I be sure everyone else’s can’t all be told to collectively focus on one place even if I trust the owners’ intentions?” that usually follows any discussion of space-based solar.
[0] https://en.wikipedia.org/wiki/High-voltage_direct_current#Co...
[1] I’m ignoring converter stations because over 1 Mm they amortise to negligible
[2] i.e. purely classical and without the fun quantum magnetic effects: https://en.wikipedia.org/wiki/Ballistic_conduction