"forgettable infrastructure component": this is what runs their entire IT. We build both the hypervisor and the backup/orchestration for it. Our stack could kill their entire operations if it's down because $whatever. 4000 virtual machines running isn't just the print server or the coffee machine.
> 4000 virtual machines
At that point I would have created some scripts to randomly reboot or fuck with their VMs. How long will you accept this? They won't pay ever.
No they run their entire IT. Not you.
They can easily move to the hundreds of alternative platforms which do exactly the same thing.
I'm not sure you are aware about the cost of migrating from one virtualization platform to another, especially when you have 4000 VMs. I can tell you it's not exactly easy, and that's even our business now (migrating from VMware to our stack).
It's not like changing a light bulb.