Would an $800 per year dedicated server be sufficiently reliable in a large college with students using it daily, without some serious time and effort spent managing the configuration?
Lets look at a descent VPS solution, say £1000 a year.
You'll need a member of staff to dedicate some time to setting it up, installing Moodle, working out if it can be connected to the domain (via LDAP for automatic authentication), MIS system (for automatic enrolment), then coming up with a plan for how the system will be structured and rolled out across the organisation. If you have a member of staff who has prior knowledge of Moodle and experience in this field then great, otherwise you'll need to hire.
Then you need to look at customising your theme a little bit because it doesn't quite fit in and feedback from staff and students is not great, it doesn't do x, y and z when <insert name here> proprietary system does. So again you've got to pay for staff time,
Maybe you just racked up another £5-8k in IT staff time? So we're totalling maybe £8k ($13k) so far.
Then, as in the OP's original post, things start to break as you scale up and someone has to work out how to fix it. Now everyone in the organisation is suggesting that you ditch Moodle and move to a proprietary LMS.
This is of course not a failing of Moodle in any way, my point was just (as Richard put it above) that Moodle is not a zero cost alternative to run at any significant scale.
I see so many people blaming Moodle for being insecure, patchy, having poor performance and being buggy, when it is actually that the installation has not been properly planned and/or is under resourced.