That's a significant over-simplification and ends up wrong in many cases. Build vs. buy is largely the same equation as rent vs. own in real estate or automobiles. Generally speaking, in the short term renting is almost always cheaper, but there's a break-even point at which buying (aka building) becomes cheaper. Owning the system also grants considerable ability to build it to be exactly what you need, instead of hacking around deficiencies and/or begging your account manager to get your feature approved and implemented.

There are plenty of situations in which the terrible business decision is to rent instead of build. The difficulty is that without knowing the future it's not always clear, so you have to use your best judgment and hope you get it right.

Edit: Also don't forget that roll-your-own doesn't necessarily mean starting something from scratch. In many cases I opted to use and self-host an open source project that sometimes is sufficient all on its own, and when not we can make changes to it. I almost never start a non-trivial project from scratch just to avoid buying, unless it's a major piece of our product or value proposition in which case you have to consider the risk of building on a foundation you don't control.

> Build vs. buy is largely the same equation as rent vs. own in real estate or automobiles.

I would say this is not a valid analogy. If you buy real estate you have an asset. However, if you build something that is not your core product you have a liability and not an asset. You're not going to get more money because you have a bunch of tech you built that you could have bought.