I'd suggest that you stop worrying about losing your job to something new and instead pioneer that something new. Relational databases aren't going anywhere, but they are no longer the only game in town. Non-relational databases have a lot of strengths and weaknesses that allow the to complement the strengths and weaknesses of the RDBMS.
While you don't want your banking to be done with eventual consistency, it's most likely not being done with ACID transactions. Huge banks are typically going to dump your daily transactions in some kind of queue for processing at some later time. It's just not possible to easily and cheaply get the kind of throughput that we demand when you're dealing with transactions from millions of individual customers plus mortgages and commercial customers.
Besides, with a properly designed eventually consistent system it's probably going to be faster and more error proof than a single server set up. Eventually consistent systems are, typically, "eventually" consistent over the period of several milliseconds, not minutes as many DBAs like to think. Werner Vogels wrote an excellent article about what eventual consistency and database durability mean: http://www.allthingsdistributed.com/2007/12/eventually_consistent.html