LTE supports a very wide range of bandwidths (0.2 MHz for NB-IoT up to multiples of 20 MHz for carrier aggregation) and modulation rates.
Given the transmit power constraints on both sides of satellite-to-mobile, Starlink/T-Mobile will most likely be targeting the extreme low end on both bandwidth and modulation rates.
The website says text messages in 2024, voice and data starting in 2025, and IoT in 2026.
So it seems they have had the same thought on the usefulness for remote telemetry data.
I don't know a lot about LTE signals but given the distances to the satellite and wide coverage area each satellite is responsible for it makes sense that performance is significantly diminished compared to your local cell tower.
There's only a few thousand satellites up in orbit right now. Maybe 4000-5000? Half of them are on the other side of the planet at any given time... The service is already strained in a lot of areas. It's not gonna be easy to give bandwidth to another few million phones.
Not just half, but most of them will be below the horizon at any point in time – their orbits are pretty low.
But since mobile phones don't really have directional antennas anyway (unlike Starlink terminals), having more than one satellite in sight above the horizon doesn't really help with capacity anyway. (You need steerable beams on both sides to make use of spatial multiplexing.)
I would've thought that an LTE tower would have more throughput, but I suppose they might need far more error correction than usual?