All of it. You can't really get around physics.

Iridium has historically targeted low-power, omnidirectional terminals (antennas can be larger at lower frequencies without requiring steering than at higher frequencies).

They recently had some forays into steered, high-bandwidth antennas with their Certus line and their second-generation satellites that now allow native packet switching (the first gen was circuit-switched at 2.4 kbps only), but that brings you into the bandwidth-limited regime, and is honestly just a waste of scarce L-band spectrum and much better served by all the Ku- and Ka-band LEO competitors.

It's going to be interesting to see if Rocketlab start also serving that market, like some of their main competitors already are.