Banking's Cloud Dilemma: Confluent's Take On Legacy Systems

With 55% of banks identifying legacy systems as their primary transformation obstacle and 70% of IT budgets consumed by maintenance activities, today financial institutions face mounting pressure to modernise their infrastructure.

Peter Pugh-Jones, EMEA Field CDO at data streaming platform Confluent, believes the traditional co-existence approach may be prolonging the inevitable. 

Confluent provides Apache Kafka-based data streaming technology that enables real-time data processing across enterprise systems.

“Let’s be honest – co-existence might buy time, but it’s not a strategy,” Peter explains. “When 55% of banks say legacy systems are holding them back, and 70% of IT budgets are swallowed by maintenance, the writing’s on the wall.”

The challenge is stark. Banks cannot innovate effectively while dedicating the majority of their resources to maintaining outdated systems. 

This creates a vicious cycle where legacy infrastructure continues to drain resources that could otherwise drive digital transformation.

Digital-native fintech companies demonstrate what becomes possible when organisations build with modern architecture from the ground up. 

These companies operate on real-time data streams and deploy cloud-native infrastructure designed for scalability.

“They operate on real-time data, and their infrastructure is cloud-native and built to scale,” Peter says. “And because they aren’t weighed down by legacy systems, they can respond quickly to changing customer needs and shifting market conditions.”

The temptation for many banks lies in adopting lift-and-shift strategies. This approach involves moving existing legacy applications directly to cloud infrastructure without fundamental re-architecture. However, this strategy may merely postpone necessary transformation work.

“The ‘lift-and-shift’ model, where legacy systems are simply ported to the cloud, may feel safer. But this often puts off the inevitable: true transformation requires re-architecture,” Peter reveals.

Without clear decommissioning plans for legacy systems, co-existence strategies risk becoming permanent comfort zones. 

These approaches can quietly undermine progress while competitors embrace comprehensive cloud modernisation and establish new industry benchmarks.

The longer banks delay meaningful migration, the more expensive and complex the eventual transformation becomes. 

Meanwhile, organisations committed to full-scale cloud modernisation continue learning and setting new standards for the industry.

The skills shortage challenge

As legacy technology expertise diminishes and demand for modern banking talent increases, banks face a critical capability gap. 

The retirement of legacy system experts coincides with surging demand for cloud and data streaming specialists, creating a perfect storm.

“You can’t leap into the future with teams or systems built for the past,” Peter explains. “That’s the core challenge of cloud transformation. As legacy tech pros retire or move on, and demand for cloud and data streaming increase, banks are left in a bind.”

The industry finds itself in an unusual position. While banking moves toward modern technology stacks, maintaining existing legacy systems becomes increasingly difficult. 

This creates operational risk alongside transformation challenges.

Specialised support becomes essential in this environment. Banks require expertise that bridges legacy systems and modern cloud architectures. 

Simply understanding cloud technology proves insufficient without practical experience integrating it with existing infrastructure.

“That might mean working with tools that simplify complex technologies like Kafka or bringing in engineers with hands-on experience in cloud migrations,” Peter says. “It’s really about people who’ve seen what works, what doesn’t, and how to avoid common missteps.”

The demand for data engineers capable of building robust, real-time systems has increased sharply. Internal teams often possess significant expertise but face competing priorities between maintaining legacy systems and designing next-generation architecture.

This responsibility creates burnout risk and can stall transformation initiatives. Asking existing teams to simultaneously maintain critical legacy infrastructure while developing modern replacements often proves counterproductive.

“Internal teams may be brilliant, but the additional responsibility of keeping mission-critical legacy infrastructure afloat can stretch them thin,” Peter continues. “Asking them to simultaneously design next-gen architecture is a fast track to burnout and stalled transformation.”

Building cloud cores for AI

The relationship between cloud migration and AI adoption presents a complex chicken-and-egg scenario. 

Clean, integrated, real-time data streams prove essential for effective AI implementation, yet legacy systems prevent exactly this data quality.

Peter argues the future trajectory remains clear, regardless of current constraints. Digital-native banks demonstrate sophistication and speed that legacy solutions cannot match, delivering agile feature deployment and personalised customer experiences.

“Digital natives offer a level of sophistication and speed that legacy solutions can’t match,” Peter explains. “Those banks are incredibly agile, able to launch new features very quickly, and often offer a more personal, curated experience than many larger players.”

Modern architectures enable these capabilities because they avoid hardware, software and design compromises inherent in legacy technology. 

Advanced systems can receive, interpret and act on data with unprecedented speed. Some can even process data while it remains in motion.

“They can receive, interpret and act on data far, far more quickly. Some systems can even do so while that data is in motion, infusing each data point with context before it’s even stored,” Peter reveals. “Legacy systems aren’t capable of this.”

However, retrofitting cloud environments with AI capabilities that help modernise legacy estates provides valuable transitional approaches. The pace of progress may be slower and costs higher, but complete rip-and-replace strategies often prove impractical.

Banks must support crucial daily operations with minimal error margins. This reality makes gradual transition strategies appealing despite their limitations. Yet stopgap solutions remain temporary by definition.

“But a stopgap is, by definition, a temporary solution,” Peter says. “The cloud-native architecture that modern AI needs to perform at its best is the ultimate destination.”

The performance gap between legacy systems and cutting-edge architectures continues expanding. As AI systems develop further, this disparity will become increasingly pronounced, making eventual migration inevitable.

Balancing innovation with regulatory compliance

Regulations including the Digital Operational Resilience Act (DORA) and cyber resilience legislation take effect in 2025, creating additional complexity for cloud transformation initiatives. 

Banks must balance rapid innovation requirements with increasingly stringent regulatory demands.

DORA specifically targets financial services operational resilience, mandating comprehensive risk management for digital systems and third-party dependencies. 

These requirements add layers of compliance complexity to cloud migration projects.

The constant introduction of new regulatory standards creates confusion and operational overhead. International debates between governments and technology companies regarding different market approaches exacerbate this complexity.

“The almost-constant introduction of new regulatory standards brings increased confusion and complexity,” Peter explains. “That’s exacerbated by international debate, whether between government or big tech players, on the validity of different approaches in different markets.”

Organisations require clarity above all else when navigating this regulatory landscape. Establishing clear ownership hierarchies for key regulatory pillars represents a basic but crucial step toward meeting compliance standards.

Demanding clarity from regulatory bodies becomes equally important. This involves defining critical elements within legislation and ensuring no ambiguity exists regarding compliance requirements. Clear parameters enable effective implementation strategies.

Once these parameters are established, automation can reduce the burden of monitoring compliance performance. 

Many regulations now mandate continuous reporting, making AI-powered monitoring solutions increasingly feasible and necessary.

“Many bills like DORA now mandate constant reporting, and the advent of AI has made that process much more feasible,” Peter says.

Ideally, monitoring platforms and dashboards should enable real-time incident response capabilities. Continuous awareness provides limited value without the ability to react immediately to cyberthreats or performance changes.

“24/7 awareness will only take you so far if you can’t react to incidents like cyberthreats, or changes in performance, as they’re taking place,” Peter says.