Exactly the right question. If atoms are forever retained as a locally true-in-its-domain and at-its-level-of-description phenomena in every future theory, I think they count as real in any important sense. Even classical mechanics is true in the sense of strongly accurate and predictive at its scales of description, as an approximation of something more precisely described by QM.

One thing that gets me excited is that there's a tantalizing possibility that the 21st century might have an Einstein-level breakthrough that treats holography and some principle of informational consistency as more fundamental than QM, which is amazing, and would change everything.

But even in that hypothetical future paradigm, an "atom" would still be something true and meaningful against that backdrop, and our measurements or knowledge claims about it would still be meaningful. And our progress toward knowledge of the atom was still real progress.

It's legitimate to treat our knowledge as limited, subject to revision, or approximating. But treating that grain of truth like it implies no knowledge or progress is in hand is an abuse of the concept.