The 3-2-1 solution (3 hosts, 2 switches, 1 SAN) would be very appropriate to what you want to do. I've deployed a number of them for sizes that you describe. From a pure load perspective, it might be a bit overkill, but is typically the minimum recommended.
3 hosts will give you HA and DRS features that is really needed to create a "highly available" cluster. You also need to figure in an N+1 for sizing. Assuming modestly configured host, a 3 host cluster will go an incredibly long way for you in the future.
2 switches is an absolute must for your back end storage fabric. You want to elliminate single points of failure, and a 2 switch stack (along with a properly meshed cabling) will acheive this.
1 SAN. The idea here is basically "shared storage" So whether it be block based (SAN via say, iSCSI) or a or a file based NAS (NFS), you want to have the VMs not actually living on the host that is performing compute services, so that it can float in between hosts as needed.
3 Dell R720's, 2 Dell PowerConnect 6224 iSCSI switches, and an EqualLogic array would fit your arrangement quite nicely. (It wouldn't be the only solution of course, as the magic really happens when you get vSphere up and running). You could go with Dell Direct, or a VAR/Channel partner. The latter can usually provide more guidance for you. Remember that a good design and deployment is key to making the magic happen.
As for the gentleman suggesting Nutanix, it is an incredibly good solution (an amazing solution actually), but it fits a very particular use case. It is simply not appropriate for what the OP originally stated goals were.
