Why would one have to hire a whole new cadre of technical talent? Wouldn't it just be taking existing talent and/or ops teams and having them work with the autoscaling APIs as opposed to working with the data center teams. If the skillsets don't match then you can either train existing talent or, yes, hire new folks.
I agree the occasional phone call removed from your workflow isn't reason enough to justify a giant cloud migration... but it is one of many reasons.
I'll acknowledge that "have to" is maybe an exaggeration, but it's definitely going to be cheaper to hire new talent than reskill existing engineers (who may not even be interested in reskilling), and profit margins generally dictate the cheapest path forward in an ultracompetitive industry like logistics. And on top of all those issues, the existing workflows must be maintained while the cloud-adapted ones come online, and in a complex enough enterprise environment we're talking about a project that can take years to execute.
I agree the occasional phone call removed from your workflow isn't reason enough to justify a giant cloud migration... but it is one of many reasons.