The cheapest option is not to install any Anti-virus software on your Terminal Servers at all.
While this may sound like madness, I've implemented this solution successfully at government sites.
The trick is to make sure that all sources of data are protected, such as Exchange servers, proxies, file servers and the like.
This used to be my number one recommendation, until I started getting overruled by corporate managers who wouldn't hear of it. The only argument seems to be "We must protect all of our servers".
Here's the theory, showing how the servers can be indirectly protected;
A terminal server runs applications in user sessions that are spawned after the user authenticates to the domain.
If correctly configured, absolutely no data is stored on the terminal server, except temporary data that belongs to each session - which is isolated in a specific folder. This folder is typically purged regularly, and is subject to restrictions on the owner.
Given that data processed on a terminal server will be stored elsewhere (ie a file server containing home directories, shares, etc), it is only susceptible to viruses embedded in source data coming from outside.
This means virus-protecting all servers that have contact with "the outside world", and all client devices that can access data from outside (eg PC floppy disks, external e-mail clients).
This way, at least, it's per-device licensing, and you don't have the worry about whether it'll work with Terminal Services or not - many packages don't, or at least require significant "tweaking".
I hope this helps