This HPC in Norway uses Rome. Peak performance is 5.9 petaflops.
This HPC at Indiana University also uses Rome. Peak performance is also 5.9 petaflops.
It is telling that these HPCs are choosing Rome over well established Intel Xeon chips. That means Naples had laid the necessary groundwork so that these HPCs are convinced to choose Rome even before it is officially launched. No way this could happen without Naples leading the way.
On another positive note, Gartner projects cloud services to grow exponentially through 2022. AMD will gain market share in an expanding TAM!
Here Anandtech speculates that Samsung and AMD are not involved doing the usual IP licensing, but rather are collaborating in a long term relationship, which can only be a win-win for both.