> It sounds like they had to reverse-engineer the MIPI protocol used on the Pi 5 to do this (since it goes through the RP1 chip), and the way it's architected, you can daisy-chain multiple QuadRF modules together, letting each module calculate it's own phase shift.

How are they planning on distributing a shared, highly precise clock for that purpose? That's already a PITA if you do QO-100 modes that need high precision, but usually there it's enough to have one good clock that you feed to the LNA... but here? Every single one of these modules needs a very precisely identical timing signal and the kind of chips you can use to multiplex a reference clock signal are pretty expensive.