I think that it is. The article does raise some good points to consider, but at this point in my experience with it I believe that a lot of the complexity of Kubernetes is not something it introduces but rather something it exposes: it was already there, just maybe not explicitly. Likewise with a lot of the operational costs and so forth.
OTOH, I do think that now that it has shown the way, there is room for both improvement and replacement.
What is the alternative they’re suggesting?
No. Like OKRs, Kube is something Google released on the world to kneecap potential competitors.
Is running your own data center worth it? Running ESXi? Using AWS? Rewriting everything in rust?
There are tradeoffs to be considered
I think it's more compelling to run/use k8s when you don't already have a mature infrastructure strategy that handles DNS/traffic/service discovery/secrets/persistent storage/auto healing/rolling deploys.