The Expert View: Re-imagining Mainframes - Unlocking value to propel your digital transformation agenda

Executives across many sectors face the same challenge: how to modernise critical infrastructure when the mainframe systems at the heart of your enterprise are running decades-old software, tightly integrated with a web of tools, and too mission-critical to fail.

 

At a recent Business Reporter dinner at the House of Lords, hosted by DXC Technology and IBM, IT leaders discussed making digital transformation a reality by rethinking the mainframe: whether to stay on it and modernise around it, or migrate away from it.

 

Attendees heard that many organisations are caught between needing to innovate and remaining what is operationally efficient. The challenge is how to reinvent infrastructure in a world where the choice around the mainframe is expanding, not shrinking.

 

That tension is driving a wave of intelligent operations, AI-assisted modernisation, and hybrid cloud strategies. The discussion made clear that the question is not mainframe versus cloud, but how to unlock the value trapped across an ever-more complex estate.

 

Mainframes: still central

Despite predictions of their demise, mainframes remain critical in many sectors, including global finance, government and telecoms, the group agreed. Attendees, all senior executives in a range of industries, were told about a public sector example where staff payroll depended on an undocumented COBOL application no one on staff understood. “It couldn’t fail,” an attendee said. “We used AI to reverse-engineer the code so it could be rewritten.”

 

The situation isn’t unusual, attendees said. Many enterprises are running 60s-era software on cutting-edge hardware, blurring the distinction between legacy platforms and legacy code. As one attendee put it, “You can treat anything as legacy, and it will become legacy.”

 

Vast amounts of revenue still flow through mainframes, said one executive, admitting that they avoid placing new workloads on them. Replacing them wholesale is too costly. The complexity is not the mainframe itself but the “spider web” of bespoke applications around it. Modernising those integrations, not just the core platform, is often the hardest part.

 

Still, AI has a pivotal role. AI can automate around 80 per cent of mainframe code analysis and transformation, said one expert. The remaining 20 per cent may require manual intervention or more specialised tools, but this level of automation was impossible even three years ago.

 

Data: the new faultline

A second theme was data movement and sovereignty, issues that are particularly pressing for organisations relying on cloud environments. The upcoming DORA regulatory requirements are forcing many financial institutions to consider where and how data moves.

 

Participants noted a shift toward private, sovereign cloud environments, particularly in Europe, where concerns about reliance on too few foreign providers is growing. “Private cloud addresses the sovereignty issue,” one guest said, while others described multi-cloud plus private cloud as the emerging default architecture.

 

But the cloud has its own risks: outages, vendor lock-in, and geopolitical uncertainty.

 

While cloud services offer flexibility and scalability, they ultimately rely on physical data centres, which can be affected by outages or other disruptions.

 

One attendee highlighted Poland’s requirement that backup data centres be separated by a minimum distance, showing that governments view resilience as a national-level priority.

 

For organisations still reliant on mainframes, new patterns are emerging: pulling data before it enters the mainframe and applying AI analytics “above” the platform, no longer treating the mainframe as the centre of all enterprise data.

 

The human challenge

While the technological challenges are real, leaders agreed that transformation succeeds or fails on culture and capability. Many organisations are accumulating technical debt alarmingly fast, especially those growing through acquisition.

 

This also creates cyber sprawl: overlapping tools, duplicated capabilities and tools few people fully understand. Rationalising this estate is as much an organisational issue as a technical one.

 

“We depend on data and tech, but our C-suite are largely focused on strategy rather than technical expertise ,” one participant said. The key, said others, is to identify “toxic” tech debt, then define what a harmonised infrastructure should look like.

 

Security and resilience

Meanwhile, the cyber threat landscape has surpassed human capacity to manage it, attendees said. AI anomaly detection, agentic security operations centres and automated threat correlation are becoming mandatory. Today’s cybercrime often begins with a reconnaissance phase; behaviour AI can detect faster and more accurately than humans. As one attendee said, “Correlating multiple suspicious events is something AI excels at.”

 

AI is also becoming deeply embedded in modern infrastructure strategies, including on the mainframe itself. Putting AI capability into physical hardware is enabling new business capabilities such as fraud detection and AML (Anti-Money Laundering) analysis.

 

Furthermore, IBM has pledged to take liability for model decisions, meaning they will assume responsibility for the outcomes generated by their AI models. Organisations may embrace AI, but accountability must remain clear. This issue will grow as agentic systems take on more autonomous decision-making across infrastructure, operations and security.

 

Reinventing infrastructure

Closing the discussion, the focus was on key questions, such as: “How do you turn technical debt into value? How do you transform the running of your data centre? How do you use AI to elevate staff and model infrastructure through digital twins?”

 

This is complicated by the fact that terms like “mainframe” and “cloud” are often marketing shorthand. Modernisation doesn’t always mean retiring the mainframe.

 

In many cases, it remains the most reliable, secure, and cost-efficient place for certain workloads.

 

The consensus from the evening was pragmatic and forward-looking: unlocking enterprise value might depend on replacing the mainframe, but equally it might be the right decision for some organisations to retain the mainframe and modernise around it. The important thing is to make a well-reasoned decision.

 

---

To learn more, please visit: www.dxc.com and www.ibm.com