Firstly, because the sales guys aren't technical. They are smart, but not computer smart. The value proposition of having them learn GNOME to do work would never fly with the suits. The big Cs would rather eat the capex and just give them Macs and never hear about it again. I also greatly enjoy not having to help the important ones with pressing technical issues. As far as GNOME has come, it isn't a replacement for Aqua or Explorer just yet. It's a death-by-a-thousand paper-cuts situation that still has a ways to go.

Additionally, workstation RHEL also isn't quite as bulletproof as the server variant. Such is the nature of the Linux graphics stack. We had a kiosk PC fail to boot to graphical target two weeks back because of an update that nuked dbus. It was just a Grafana kiosk so who cares really. Hasn't happened again since, but it shakes confidence you know? The servers, however, have been minimal in their issues. I think the only major issue we ran into this year was libvirt imploding on an on-prem server. The post-mortem was interesting on that one.

When your app vendors write their apps for Windows, no. You could try moving everything in Citrix or VDI, but then you're still running Windows and doing it with more costs.