I assume your path loss figure is assuming dB per km at sea level due to absorption by the atmosphere.
2.4 GHz is attenuated by water vapour, which there is less of at higher altitude and none in space. So I'd think the loss to atmospheric absorption looking up to a satellite would be much, much less than it is at sea level.
On the contrary, I've used the standard "free space path loss" formula, aka path loss in vacuum. You'd have to add additional losses for water vapor and atmosphere, I have no idea how much additional loss they would incurr. Maybe 5dB?
Doesn't the basic path loss formula assume non-directional antennas?
Using a calculator [1] that takes gain into account, looks like a narrow beam antenna at 35 dBi could bring it down to a comfortable -75dBm - and I imagine you could have even higher gain with phased array antennas. The real problem is aiming the beams, and SpaceX has already proven this is possible.
figures 2-2-8 and 2-2-9 graph attenuation due to oxygen and water vapor expected for frequencies 1-30GHz, on a vertical total atmospheric path (sea to vacuum)
page down for water droplet and rain losses per kilometer at various droplet sizes
I guess they get 21dB from directional antenna, another 21dB from message coding (essentially repeating each bit 128 times to improve SNR)?