Can't give real answers without data on Starlink reliability (or non-consumer features), but personally as someone who depended on ATC both as small time pilot and more regularly as passenger, I'd be wary of sat radio as main backhaul.
Of course Verizon can fsck it up as well, but multiple leased fiber lines, possibly direct MPLS with guarantees on latency and bandwidth, or maybe even DWDM between major locations sounds more reliable to my instincts.
Most importantly I don't trust the hatchet job way this is done, not just on Starlink use, but the whole hatchet job with FAA.
With density of flights in USA, intra-ATC Comms are non trivial part of safety - passing flight information on structured messages rather than by hand between areas is... More than important.
I haven't before, but due to certain private interests I can explain what they mean by "legacy copper TDM infrastructure".
A lot of interconnections between US ATC services were done back with the good old Bell System over then new T1/T3 digital carriers, plus analog leased lines in some cases, that provided high reliability and high quality of service links for both data and voice (which could include voice connection to remote radio transceivers).
This is, of course, being phased out even from "compatibility mode" setups where you get old style trunks that are backed by newer infrastructure. So the network infrastructure is literally dying out.
A comparable setup with modern tech would be MPLS fiber backbone with capacity reservation (thus actually getting stable latency and bandwidth guarantees), which would allow creation of virtual meshes with different QoS requirements that would provide both stable voice service (including from centralized ATC to remote radar transceivers), guaranteed performance data links (for example for radars etc), slower but still guaranteed service bulk data transfers, etc.
I might have also mentioned DWDM, which essentially lets you create a physical fiber optic path where switching is done by optical elements based on frequency of the signal (so you can pack multiple signals into one fiber then route them to different ports etc. etc.) - very useful if you want no variability in connection while still be able to reroute them fast.
That exactly is the point of the program. A shift to an IP network.
Where Starlink comes in is in the intermediate step. Before Verizon finishes running their MPLS network, there's a program called RTRI that provides an off ramp for TDM, so when Lumen stops serving, they have something. It's contracted to L3Harris and runs over SDWAN while using any network as an overlay. Satellite (including GEO), Cable and 5G (anything available really)
Of course Verizon can fsck it up as well, but multiple leased fiber lines, possibly direct MPLS with guarantees on latency and bandwidth, or maybe even DWDM between major locations sounds more reliable to my instincts.
Most importantly I don't trust the hatchet job way this is done, not just on Starlink use, but the whole hatchet job with FAA.
With density of flights in USA, intra-ATC Comms are non trivial part of safety - passing flight information on structured messages rather than by hand between areas is... More than important.