PG requires a lot of expertise to keep running when you get to a billion rows or massive ingest. It can do it, but it doesn't just do it out of box running the defaults.
Unfortunately, most places go with hardware first, so the problem grows larger and larger. When they do finally hire someone who knows how to address it, it’s a monstrous undertaking that dev teams are reluctant to do, because it’s a lengthy pause on feature development (so really, PMs are reluctant), and also future velocity is typically somewhat slower – turns out proper RDBMS data modeling is rigid, and doesn’t suffer fools.
