I think at scale, the method used in Chile with the Cerro Dominador[0] using CPVCSP could be adopted. Their system was about $1B for 210MW and has ~17.5 hours of storage.

So ~$4.76B for 1GW or ~$4.76T for 1000GW.

This also ignores future upgrades with would use sCO2 turbines vs typical steam and higher delta_C molten salts (typical delta_C now is only 300 for KNO3 + NaNO2 + NaNO3, but systems using CaCl2 + MgCl2 would have a delta_C of ~1500, and meta-material salts could be engineered with even higher delta_C) so the same solar fields can have even higher outputs.

[0] https://en.wikipedia.org/wiki/Cerro_Dominador_Solar_Thermal_...