well, that is not trivial question to answer
First, note that there're 2 different flavors of vc ops - the vApp and the Standalone, and there're some important differences between them.
More info on that as well as the different scalability specifications are available in the product documentation: VMware vCenter Operations Manager Documentation
vc ops scaling is mostly limited by the number of metric it collects and manages, and not by the number of hosts/servers.
A host is just a resource as any other (vm, datastore, SAN) with a bunch of metrics attached to it.
Thru the custom-ui, it is possible to explicitly select which metrics should be collected for a specific resource.
Thus you can limit your lab monitoring to a set of 10-20 crucial metrics-per-server, and squeeze within the limits of the product.
Of course the feasibility of doing that of that depends on your usecase.
If you need the full-scale monitoring (300+ metrics per host or per VM), you have to do the monitoring "per cluster" or "per datacenter" - basically use some logical division where number-of-objects-per-group is within the scope of a single vc ops instance. And then use multiple instances to monitor each group.