Some say generative AI’s role is being overly overestimated in modernisation projects pursued by large organisations aiming to phase out mainframes step by step.

Techzine reported on Wednesday that Gartner said in a report titled “Too Big to Fail: Why Mainframe Exit Projects Are Likely to Fail in the GenAI Era” that many mainframe retirement initiatives launched in 2026 are expected to fall short of the results initially hoped for. It cited that expectations that AI can replace complex legacy environments quickly and reliably fall far short of reality compared with what many vendors claim.

Gartner said vendors are actively promoting AI across a range of solutions even when the practical value is limited, and that in mainframe migration this could have serious consequences. It said project costs could rise beyond plans and business process continuity could also be threatened.

Gartner warned that errors in migration decisions could lead to long-term disruption and new technical debt.

Gartner said AI is effective in identifying and documenting technical debt, but is unsuitable for automatically converting legacy code to modern platforms. It said a shortage of mainframe specialists and vast technical debt may make generative AI look like an attractive tool, but the reality is different.

Gartner said the number of vendors supporting mainframe exits will shrink sharply over the next few years, and forecast that much of the market will disappear or need to change direction as promised deliverables fail to materialise.

Mainframes remain strong. Techzine reported that despite the rise of cloud-native architectures, mainframes continue to play a central role in applications where reliability, security and high-volume transaction processing are important.