> It's unclear to me what the author thinks OOP is, and what he thinks we are replacing it with.

The author is complaining about bloat.

The thing is, in this case, the bloat has highly tangible costs: Spreading an application across multiple computers unnecessarily adds both operation costs and development costs.

Unnecessarily is the key word here. Sometimes, although admittedly rarely, the benefits do outweigh the costs, and it is necessary.

The problem is that if you grow your monolith to the point that it becomes necessary to split it up, now you generally have a long, expensive and error-prone migration on your hands. If you architect you system with this eventual transition in mind from the start, though, it can be much less painful.