Since AMD has more money now and uses TSMC and GloFo, which means they need to adapt the masks and design to a different process anyways, I thin a selerat desktop version is very likely - server tsmc, desktop glofo
Adapting a design to 2 processes is a far cry from a different design altogether. It also allows more robustness into their risk models. Same design at 2 fabs allows major failures without sinking them, even if they would still need to validate the GloFlo one, which they might slowly start to do anyhow as CYA project. AMD still is and will be for quite some time in an unstable position. Any major mistake can remove all gains made.