I disagree with almost every single point of this article except that buying can be better than building (note this still isn't always true, especially if you are in a very large company that would get charged usurious rates based on usage).

The other points really only hold true for things that are allowed to break. As someone who writes code that should never, ever break, and gets support tickets even for one-in-a-million breakages, testing is absolutely paramount. I can see how certain products, like a web-app, could follow these principles and benefit from not being over-tested, but it's definitely not an absolute.

Of course, my code does still break. The author is right that this is still an absolute. But the blast radius of the breakage and the frequency are both quite low because of the measures we take to prevent them.

Regarding the point about ditching QA and testing in production: both are possible. You can make QA a shard of your production environment (there are many different ways to implement this based on your tech). That is the ideal way to structure it so that it's actually pretty representative of a real environment without requiring lots of duct tape.