The pandemic did something to financial infrastructure that thirty years of digital transformation consultancy could not. It exposed which systems actually worked when nobody could go into the office.
For most of the last decade, the conversation about legacy modernisation was abstract. Banks knew their core systems were old. They knew the technical debt was real. They knew the cost of running mainframes alongside cloud-native challengers was unsustainable. But nothing forced the issue. Quarterly earnings still came in. Customers still got served, mostly. The pain of the existing system was always less than the pain of changing it, which is the precise definition of an organisation that does not yet have a real problem.
Then 2020 arrived. Branch networks went dark overnight. Call centres collapsed. Operations teams scattered to their kitchens. The institutions that had quietly invested in API-first architectures, cloud-native cores, and digital onboarding kept functioning. The ones that hadn't, didn't. The difference was not subtle. It was the difference between processing a mortgage application in fourteen days and processing it in fourteen weeks. Customers noticed. Boards noticed. Suddenly the abstract conversation about modernisation had a number attached to it.
What is interesting now, three years later, is that the urgency has not faded. Most large institutions have moved from talking about composability to actually decomposing. The monolith is no longer the default end-state of a banking platform. Microservices, event-driven architectures, and modular cores are now the conversation. Even the most conservative tier-one banks are quietly migrating workloads to architectures that would have been considered reckless a decade ago.
The technical shift is the obvious part. The harder shift is organisational. A monolith does not just describe a piece of software. It describes how decisions get made, how teams get structured, and how risk gets owned. You cannot run a modular technology stack on top of a hierarchical decision-making culture and expect either of them to work. The institutions that have moved fastest are the ones that have rebuilt the org chart alongside the architecture diagram. The ones that have only rebuilt the architecture diagram are finding that their new modular systems behave suspiciously like the old monolith, just with more YAML files.
There is also a question worth sitting with: what does this all mean for resilience? The old monolithic core was, for all its faults, a single thing you could understand. A modular cloud-native stack is a thousand small things you can never fully see at once. The failure modes are different. The dependencies are wider. The blast radius of a third-party outage is larger. The industry has traded one set of operational risks for another, and the new set is not yet well understood.
What 2020 proved is that legacy systems fail under stress. What the next stress event will prove is whether modern systems fail differently or just fail in newer ways. The post-pandemic financial stack is being built right now. We will not know how good it is until something breaks it.