I've got a fairly small deployment at the moment in AWS, configured exactly as you describe. I think I'm running two t3.micros for fleet, a t2.small for the aurora db, and a t2.micro for ElastiCache. With 60 hosts, I'm less than 2% CPU for the fleet instances, the db hovers around 10% CPU (which seems like pretty much the baseline without any active db connections), and ElastiCache is similarly low CPU use.
I've also increased checkin values slightly for log submissions and config refresh from the base example configs.