The banking angle is usually less "quantum computers make money go brrr" and more two boring buckets:
- optimization/simulation problems where a useful quantum speedup would have to beat extremely good classical heuristics, not just a textbook example;
- crypto migration, where the prudent move is inventorying systems and getting post-quantum paths ready long before anyone has a scary machine.
The second one is the near-term work. Less sci-fi, more spreadsheets, HSM firmware, certificate lifetimes, vendor archaeology, and discovering that a critical batch job still speaks TLS like it was raised by wolves.
If quantum changes banking, the first visible sign may be compliance people asking surprisingly good questions.