The problem with that model is support. How many devs should they have writing patches for their 13 year old OS? How many techs on phone or web support should be trained in their 13 year old OS? How much downward compatibility should the new software have with their 13 year old OS?
Supporting multiple versions of the same software is incredibly expensive, each additional version adds time to training, complexity to your dev environment (gotta keep able to build the old stuff), and you’re sapping resources that could be used to add features to the next version. Dropping XP drops 2 OSes from their support list (because there’s actually two XPs out there the 32 bit original and the 64 bit “upgrade”), that’s millions of dollars saved. There’s no way your model would keep XP profitable. It’s ancient, way past time to kill it, anybody that still has a business need to keep it needs to hoard hardware for it, or make the switch to VMs.
I doubt that any of those would ever be an issue. Security issues would be handled by the newer OS that the VM is running on.
"There’s no way your model would keep XP profitable. It’s ancient, way past time to kill it, anybody that still has a business need to keep it needs to hoard hardware for it, or make the switch to VMs."
I never said it would be profitable......but it would let customers who have VERY EXPENSIVE HARDWARE (like that $1MM++ superconducting NMR) that cannot run with newer OS's keep that hardware up and running, and keep those customers buying NEW COMPUTERS with the NEW OS as customers.