This is arguing that all of Earth's energy needs could be supplied by orbital solar-power, as early as 2050. Is that realistic in any way?

It seems to me that this is a bit like the iPhone, before it existed. Almost all of the technology exists, but nobody has put it together in the right way yet. Is that true? I wish I knew.

Oh, the technology absolutely exists and it wouldn't be terribly hard to build one, but whether it's worthwhile to do so with all the headaches caused by microwaving birds, satellites, and airplanes is a different question.

Can we get microwaves (or whatever atmosphere blindspot frequency) through dozens of kilometers of atmosphere without it diffusing/scattering or losing a bunch to absorption?

Would need very large ground target to absorb and convert to electricity and hope someone or some tech error doesn't change the coordinates. The standard cosmic ray/solar flare bitflipping or supervillain hacker.

Lose energy 2 steps more than a cell on the ground

  Sunlight -> Battery (solar cell)
  Battery -> Transmitter (magnetron?)
  Transmitter -> Receiver (through atmosphere)
  Receiver -> battery/grid (Receiver a PV cell or funky molten salt/water steam turbine?)

How many football fields of solar cell in space do we need to match a rooftop on the ground?

Just get started on that Dyson sphere and cut out the middle man...

I haven't done the math myself to double check, but the ground station receivers would have to be multiple kilometers across, and atmospheric energy losses would be significant. Honestly, after a certain level of complexity you really do have to admit Nuclear is safer, cheaper, and simpler.

> Is that realistic in any way?

the Sun provides much more power than any satellite could. Solar panels + batteries seem dramatically more realistic to me