What do you do when you have so many published services in ArcGIS that running even one instance of each service maxes out RAM/CPU?

We've grown to the point where we have more published services than our current server can run at one time, and expect to continue to grow considerably over the next few years. Even at the minimum one instance per service, we're running out of RAM.

Of course, doubling the RAM would get us by for probably another few months, but we will eventually need multiple servers, and I need to know how to work this out.

My understanding is that when you add a new server to an existing site, all of your published services will have at least one instance per service on all servers. Is there a way of distributing these services across multiple servers without having multiple 'sites'?

I konw that I could publish X amount of services to one server, and another batch to a separate server, but I'm assuming there's a better way to do this, and I'm hoping someone can shed some light on this.

Thanks!!!