Again, technically only limited by the resources your system has available. There is no logical limit in any way. If your server has ~100 CPU cores available, you could support ~100 users at the same time - assuming each user is executing their request at the same time / without pause etc.
Since users won't typically work with a system at the same time, have different patterns of usage etc., things become quite hard to estimate unless you know your specific use case quite well. I had a customer use Alfresco to provide a platform to ~70.000 users with only two servers of 8 CPU cores each - but only <= 10% of all users would access Alfresco on any single day.
