Whenever a technology "obviously" should win and doesn't, I've learned to follow the money. In regulated finance it leads somewhere uncomfortable: the friction everyone complains about is also somebody's revenue.
Think about what tokenization removes reconciliation, settlement delay, layers of intermediation. Every one of those is a line on someone's income statement. Custodians, clearing brokers, transfer agents, back-office teams: a huge share of the system's "cost" is another institution's paycheck. So "we make settlement instant and cut out layers" sounds like savings from one seat and like disintermediation from another.
That's the real adoption barrier, and it's rarely said plainly. The players with the balance sheets, licenses and client relationships to actually bring assets on-chain are often the same ones whose fees depend on the old process. Asking them to lead is asking them to shrink their own business not irrational resistance, just rational self-interest.
This is the honest test for a chain like Dusk. Routing around incumbents hits the cold-start and licensing wall; partnering with them Dusk's actual path is more realistic but lopsided: they'll adopt what cuts their back-office cost and quietly slow-walk what threatens their fee. The design that works gives them a reason to say yes defend your franchise with cheaper rails, don't disappear.
Who moves first: incumbents who see more upside in cheaper rails than in a shrinking fee. What makes it fail: asking the gatekeepers to remove themselves. They won't they'll wait it out or water it down.