Nothing because you can already use KVM + oVirt to get something which is reasonably close to vSphere's user interface for all but the highest end users.
You are comparing what a single windows host to business is compared to whole management environment: Active Directory/Entra and gazillion of other things that help manage the business.
You can't really "build" VMware though. It's a proprietary hypervisor and ecosystem of interoperable products with tons of hardware vendor drivers. Building on KVM or even a new hypervisor would not get you 'VMware"
For 100M $ you can buy Proxmox, start advertising, scoop up all the lower end VMware customers and fill in the feature gaps according to customer demand over time.
The whole Oxide computer company raised less then $100 million $ as far as I know. With that money and open source they build not only service that can do live migration, but also the computer to run the whole thing on top.
Sure that's not a UI and doesn't include a lot of other stuff that VMWare does, but it does illustrate what a gigantic amount of money 100 million $ actually.
The underlying open source tools are so good, if you have the money to have real engineering team and combine them in interesting way and to costumize them for your needs, 100 million $ seems like it would go a very, very long way.
Specially when we are talking 100 M$ every year or something like that.
Live migration is a pretty hair-raising feature if you know how it is implemented internally, but nevertheless qemu/libvirt put in a vast effort to make it work and many Red Hat customers are using it routinely.