Dynamic Allocation of GPU Cores

I would like to see the ability to run multiple workstation virtual machines off a single cluster with a couple of high-end GPUs, each running Adobe packages to create content, with the virtualized GPU cores being dynacmically allocated depending on which user is demanding the most resources in that moment. I look after a team of creators and this would be an excellent centralized solution for us, as opposed to individual workstation upgrades. The hardware already exists too - AMD's Vega GPUs can virtualize individual cores. All we need is software support.

You can do it, Adobe!