Maybe they found a frequency that is not absorbed by the athmosphere that much. Seems far-fetched with the gas mix that the athmosphere is, but I guess that such frequencies would be company IP.

Instruments that sweep all frequencies have been available since shortly after the dawn of radio and public atmospheric absorption tables have been available for nearly as long. Also, absorption bands tend to be very broad in this region, so after 10 or 20 data points there's nowhere for a magic frequency to hide.

You are correct. There's no magic at work here. We don't break the laws of physics, we just flex them with clever engineering... like most innovators that came before us.

Path loss through free space at these relatively low frequencies (<6 GHz) isn't from air/water vapor losses. It's from diffraction spread.

This company can say all they want about near field tech, but the beam waist diameter relative to wavelength determines the diffraction spread. And that aspect of path loss is proportional to the distance in wavelengths even if there's no "absorption" by the air components. For any reasonable link at 2.4-5.8 GHz the length in wavelengths will be tens of thousands.

The equations governing diffraction are relatively straight forward. We are operating within the near-field (or more accurately in the Frensel range). I'm sure you can do the math and see how focusing a phased array can reduce diffraction at this range :)

Just make sure to do your liability math because in a about a year you're going to have a lot of investors very angry at you.