Legacy IT Systems Get New Life as AI Investment Grows

What happened

A new report highlighted in The Register finds that as enterprises ramp up investment in artificial intelligence, many are turning back to legacy IT infrastructure — including mainframes — rather than replacing it. The reasoning is practical: these older systems still hold the core business data and decision logic that AI services need to function, making them harder to retire than previously assumed.

Rather than treating legacy platforms as pure technical debt to be phased out, organisations appear to be finding ways to extend their use, connecting them to newer AI and analytics layers so that data and business rules built up over decades can feed emerging AI applications.

Why it matters

This shifts the common narrative around AI adoption, which often assumes a clean break from old infrastructure toward modern, cloud-native stacks. In practice, the report suggests many enterprises are pursuing a hybrid path: keeping trusted legacy systems in place for their data and logic, while layering AI capability on top. For transformation leaders, this changes the calculus — the question is not simply "modernise or don't," but how to responsibly extract value from systems that already encode years of business knowledge.

For technology and operations teams, it also reframes what "AI-ready" infrastructure looks like. Instead of a wholesale rip-and-replace strategy, the more common path may involve integration work — exposing mainframe data and rules through modern interfaces so AI tools can draw on them safely and reliably.

The Renascence take

There's a tendency in transformation programmes to treat legacy technology as a liability to be eliminated as fast as possible. This finding is a useful corrective: the value often isn't in the hardware, it's in the decades of accumulated business logic and data trapped inside it — logic that, if lost or poorly translated, can quietly degrade the very AI experiences organisations are racing to build.

The instinct to "modernise everything" can be as risky as the instinct to change nothing at all. Before any AI initiative, leaders should ask what tacit business knowledge lives inside their oldest systems — and make sure that knowledge is preserved and understood, not just wrapped in a new interface. The organisations that get AI experience right won't necessarily be the ones with the newest infrastructure; they'll be the ones that know exactly what their legacy systems know.