Well, everyone is speaking about what they have encountered. But I see your point. Companies should look ahead to when they reach a bigger scale.
You can be a small player and still get fucked over when your sole vendor increases prices/stops running a service/changes a service drastically/has large amounts of downtime/etc
If you use standard components that you deploy yourself, you are beholden to no single provider, and you can even split your hosting between DCs operated by different companies, to reduce the chances of 'whole-of-org' issues.
Honestly a server vendor should be treated how most people want phone/internet companies to act (dumb pipes): a dumb VM/Physical Host provider.