Across Europe, many mainframes still handle the core operations of banks, government platforms, infrastructure providers, and large enterprises. In fact, according to IBM, 71 percent of Fortune 500 companies use mainframe systems, and the market is expected to grow in the years to come. These machines were never meant to be temporary: they were built for scale, resilience, and long-term performance and they continue to quietly prove their value.
Despite repeated predictions of their retirement, many organizations have kept mainframes in place as nothing else delivers the same balance of stability and processing power. Indeed, in some environments, they still outperform newer architectures on the tasks that matter most.
What’s changed isn’t the relevance of mainframes, but the context around them. Today’s teams face new constraints: tighter regulations, faster delivery cycles and stricter documentation requirements. The systems are still reliable, but in many cases the ability to work with them clearly and confidently is missing. Inheriting decades-old code without documentation means even minor updates can be risky and slow.
The reason for stalling mainframe modernization is not outdated technology; it is a lack of clarity. With accelerating compliance demands and a push for digital transformation at the board-level, this lack of visibility is no longer sustainable. By combining static analysis tools with generative AI, organizations can reduce the opacity within their mainframes, and turn complexity into control.
SVP of Software Engineering at Rocket.
Without clarity, progress stalls
Mainframes often operate within a complicated framework of past decisions and inherited code. Engineers who once knew every corner of the system have moved on. Teams now manage code they didn’t write, connected to decisions they didn’t make.
This creates friction. When a system works but no one fully understands it, every change feels uncertain. As clarity fades, even minor updates start to take time. Teams struggle to measure the risks, and in some cases, prefer to leave the system untouched rather than risk unintended consequences.
That opacity doesn’t just affect development: it extends to data, operations, and reporting. When teams cannot explain how numbers are calculated or how records are processed, trust begins to erode – manual checks increase, deadlines slip, and compliance becomes harder to manage.
In sectors under strict oversight, this uncertainty becomes a barrier. Without clarity, justifying outcomes is difficult. A system can be stable, yet lose credibility if its inner workings remain opaque. When that happens, even the most reliable infrastructure turns into a point of hesitation.
From visibility to compliance: rebuilding control step by step
Established static analysis tools already help teams make sense of complex, decades-old applications, delivering a detailed and deterministic understanding of existing codebases. These tools remain the foundation for efforts to modernize mainframe environments.
Generative AI, while still maturing – especially for languages like COBOL or Assembler – adds value, enabling compliance, collaboration and safer modernization. It is opening new possibilities for these environments, not by replacing the systems, but by helping teams see them more clearly.
By augmenting the information available through static analysis tooling, it simplifies and summarizes existing code, maps out dependencies, and produces documentation based on how systems work today, not just how they were designed to work decades ago.
This clarity makes action possible. Instead of relying on memory or disconnected knowledge, teams can base decisions on the organization’s real structure, identifying where risks are visible. Priorities are therefore easier to define and compliance becomes part of the system’s natural lifecycle.
In Europe, where frameworks like GDPR, NIS2, DORA and the AI Act set high expectations, this kind of visibility is essential. In France, certain organizations are designated as Operators of Vital Importance – entities responsible for infrastructure critical to the country’s security and public services.
These organizations operate under additional regulatory obligations that go beyond general compliance frameworks. That includes traceability, auditability, and the ability to explain every layer of decision-making. While in the UK, these systems are referred to as critical national infrastructure (CNI), a designation which as of last year includes data centers.
Generative AI augmented tooling supports this by making documentation adaptable. It reduces the time needed to answer regulators, eases coordination between IT, security, and business teams, and, more importantly, it makes modernization safer, because the system is no longer a black box.
Modernization becomes a method, not a disruption
Most teams don’t begin a modernization project with a roadmap, but with questions about what is still running, what might be adapted, and what should remain in place. These questions often remain unresolved simply because the system lacks transparency.
Once the structure becomes clear, informed decision-making can follow. Some systems stay in place but become better documented. Others are connected through APIs or wrapped with automation. Changes can happen gradually, while governance is tightened.
What used to feel untouchable becomes manageable. Businesses can facilitate gradual transformation that supports long-term progress and allow teams to confidently upgrade their systems without dismantling stable infrastructure and disrupting business operations.
Mainframes aren’t going anywhere. The systems are stable, proven, and essential. What’s changing is how they should be approached, with more attention paid to what they do, how they’re connected, and how they’re maintained. With clearer insight, progress becomes possible.
We’ve featured the best productivity tool.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
