I think this idea is kind of nuts. Even if there is a Linux distro with 1:1 feature parity with Windows for a particular business application--which I don't think is a given--there are still massive hurdles:
- The cost of retraining users
- The cost of evaluating whether, for each use case, the Linux distro has feature parity
- The cost of transitioning each old PC
- The dissatisfaction among the employee base about having to re-learn a system, even if the necessary features are there
- The fact that most business with 10k+ users are going to have business-specific apps that barely run on Windows 10, with no equivalent on Linux
- The risk to the decision-makers if it doesn't pan out -- it's the opposite of "no one ever got fired for buying IBM"
- The fact that all of the back-end infrastructure--active directory domains, security, etc--is going to have to be transitioned too.
- The likely costs to the business itself in terms of lost opportunities, errors, and so on while the transition occurs.
The hardware itself is relatively cheap compared to the cost and logistics of getting that hardware set up and training users. My guess is that basically every business with that quantity of users would choose to just buy new hardware and keep their existing infrastructure and training over the alternative.