Microsoft provides some information to help you assess the hardware that you'll need to meet your client needs, but ultimately, this depends on several factors that are not set in stone. Your users and how they use the connection can make a lot of difference, and the best way to get a clearer picture would be to deploy one server for a pilot group and monitor the server's responses with a performance monitoring tool.
Windows has such a tool built-in, and it's a good way to keep an eye on things. The idea is to set up a pilot and monitor the server over a period of time, and then calculate the capacity for your specific organization and users based on it. With a concrete set of data, one could calculate the server's true capacity and how far it will go. In the following screenshot, for example, you can see a typical graph collected by the Performance Monitor tool, allowing us to see how performance varies along a typical work-day:
