If you have the dev resources to do that, great. But lots of other players will prefer to use off-the-shelf stuff from cloud providers.
But even with “off-the-shelf stuff”, your site reliability engineers would still be using Kubernetes as the primary language to communicate. Or I guess you could go all cutting edge and go one layer up with Istio instead. In that case then, Kubernetes is still there hiding underneath.
(Or maybe all three. They really are eating the industry, but they add value so, yeah, I guess they deserve their rent. Just sucks for anyone else in that space.)
Turtles all the way down I guess, which now I’ll have to spend all weekend on researching unfortunately.