It's not the lower level primitives here that are valuable from VMware, it's the higher level abstractions built on top of them that facilitate easier management and use by (relatively) less technical people.
You can give a "good at Windows" type technician a few hours of training and they can put together a usable vCenter deployment, including things like corporate SSO and such. They can drag and drop things around and it'll mostly work. And when things go wrong and this person is confused about what to do, they can call up support for help.
Obviously, this is centered on-prem, which is indeed shrinking - but a lot of orgs have a lot of investment in stuff running on site, and there are still plenty of cases where having e.g. lots of storage and enough compute to "process" it in some way physically close to office users can speed up their workflows.
Disclaimer: I am an ex-employee of VMware.
This is something that the HN crowd repeats as if it were an obvious axiom where no scrutiny is needed. While I'm sure it's true that there's a general movement towards cloud-based infrastructure, there is a very real and non-trivial part of the IT infrastructure of any country that will not move to the cloud in the coming decade, if it ever will. This general movement is not the same thing as "everything will eventually have moved".
The hypothesis that on-site virtualization somehow is an obviously dying technology assumes that it must be true that on-site hosting will stop existing, which I have a very hard time believing. If nothing else (and there are plenty of "elses"), strict data security and privacy concerns will keep on-site hosting moated for a long time.
Which is based upon KVM - AIUI.
https://www.kovarus.com/blog/deep-dive-into-the-aws-nitro-sy...