If it is necessary to throw X amounts of more hardware at a database (as the tweet says), then this usually means that the database is not up to the task.
Proper databases are designed to scale well with increasing data amounts, and, in the normal case also with changing database structures, just because they are designed for exactly that purpose, i.e. to live for quite some time, at least more time than a few hardware generations.
Well, if at some point you find out that your database is possibly not suited for much more data, or for different, upcoming hardware structures, or you find out that the recently added complexity can be handled by your database only in a suboptimal way,…
… you have the option to migrate the database to a more suited model. But this costs resources.
And you have the option to leave it where it is, which does not cost resources. But may have further implications in the future.