AI overtakes legacy replacement in Kyndryl modernisation

Kyndryl has published its first global Modernization Report, which found that AI adoption is now a higher priority for organisations than replacing legacy systems.

The report is based on responses from 2,000 business and technology leaders across five continents and 12 industries.

The findings suggest a shift in how companies approach technology change. Rather than consolidating systems into fewer platforms, many organisations are increasing investment across both legacy and newer environments as they prepare for broader AI use.

As a result, many businesses are managing more complex estates spanning public cloud, private cloud, mainframe, edge and on-premises systems. Mission-critical applications are now spread almost evenly across those environments, suggesting that hybrid technology models remain firmly in place.

AI ranked ahead of cost reduction, legacy replacement and concerns about a shrinking pool of legacy skills as a driver of modernisation. It was the leading reason organisations are increasing use of mainframes and edge computing, the second-largest reason for expanding software-as-a-service and private cloud, and the top reason for upgrading networks and application portfolios.

Every organisation in the survey that already uses a mainframe said it plans to deploy AI on that platform. The study also found that nearly half of respondents are behind schedule on their modernisation goals.

Delays appear to be widespread. According to the research, 98% of organisations had delayed modernisation and experienced negative business effects as a result.

Operational strain

The findings suggest technical complexity remains a central obstacle. Only 9% of respondents said they fully understand dependencies across their applications, while a quarter said they have large numbers of undocumented applications.

Those gaps matter because modernisation increasingly spans multiple business functions rather than remaining within IT alone. Respondents identified nine stakeholder groups that influence modernisation decisions, including Chief Information Officers, Chief Technology Officers, boards of directors, finance teams and compliance teams.

This shift has turned modernisation into a board-level business issue. Organisations must now coordinate decision-making across more groups while maintaining momentum.

The research also highlighted the wider pressures facing technology leaders. Chief Information Officers are being asked to manage AI adoption, cyber resilience, regulatory compliance, operational efficiency and business growth at the same time.

Agentic AI

The report points to agentic AI as one area where some organisations are moving more quickly. Kyndryl defined agentic AI as systems that can autonomously plan and execute tasks.

Ten per cent of organisations said they have deployed agentic AI in production for modernisation work. They are using it for dependency mapping, code conversion and documentation generation, tasks that have often required significant time and specialist skills.

Early adopters reported stronger progress than other respondents. Among organisations using agentic AI in production for modernisation, 67% said they were ahead of or on track with their goals, compared with 51% of other organisations.

Martin Schroeter, Chairman and CEO of Kyndryl, set out the company's view of the shift in priorities.

"AI is creating new possibilities and raising expectations for the business outcomes technology should deliver," said Schroeter. "Modernization is about preparing for what comes next, and the leaders making the most progress are investing across their technology environments - strengthening resilience and putting those investments to work where they deliver the greatest value."

Sovereignty concerns

The report also found that geopolitical and regulatory factors are shaping infrastructure decisions alongside cost and performance. Sovereignty requirements featured prominently in responses about cloud and network strategy.

Some 65% of organisations said they address sovereignty selectively or mainly through a compliance lens. At the same time, 63% expect sovereign cloud usage to increase, while 55% expect sovereignty and regulatory requirements to place greater demands on network architecture.

That suggests location and governance are becoming more important in decisions about where workloads run. These choices are increasingly influenced by regulation and geopolitical considerations, as well as performance, innovation and cost.

Overall, the study shows organisations trying to modernise while adding, rather than reducing, variety across their technology environments. Nearly half are behind schedule, only a small minority say they fully understand application dependencies, and all surveyed mainframe users now plan to bring AI to those systems.