Gartner Predicts 70% of 2026 AI-Led Mainframe Exit Projects Will Fail
TL;DR: Analyst firm Gartner has published a paper predicting that more than 70 percent of mainframe exit projects initiated in 2026 will fail because of overestimated generative AI capability, and that 75 percent of vendors in the “mainframe exit” market will pivot or fold by 2030. The call-out is unusually sharp for Gartner and arrives while IBM’s mainframe revenue is rising.
What the Paper Actually Says
The Gartner team — Dennis Smith, Alessandro Galimberti and Tobi Bet — argues the sheer volume and interconnected complexity of mainframe-hosted data make wholesale migration a physical and financial impossibility for most large enterprises. They accept generative AI is useful for detecting and describing technical debt, but find significant limitations when it comes to automated conversion and migration of legacy code — particularly around preserving mainframe-specific performance and throughput characteristics after a move.
The analysts attribute vendor pressure to push AI-led migrations to “aggressive investor demand for AI capabilities as the sole indicator of a vendor’s long-term health”, combined with customer anxiety about mainframe staffing gaps and technical debt. AI can feel like the answer, Gartner writes, even when it is not.
The Gap Between Marketing and Capability
Gartner’s framing — “the gap between the marketing promise of generative AI and its actual capabilities in code transformation” — is blunter than the firm usually issues publicly. The stakes justify it: “poor decision making regarding migration is not merely a budgetary overage; it is a threat to business and operational continuity”. The paper advises treating the choice of platform as a workload-by-workload decision rather than chasing a single “magical” migration tool.
Customer behaviour is already shifting, according to the analysts. Many organisations are stepping back from mainframe exit ambitions as they recognise the near-impossibility of exit at acceptable cost and risk, and instead looking for ways to improve what they already run.
The IBM Read
IBM’s stock slid earlier after Anthropic touted Claude Code’s COBOL-conversion capabilities — the exact marketing promise Gartner is now challenging. Current IBM revenue supports Gartner’s position: mainframe sales are unusually high, and the z17 has been received as a meaningful upgrade. Gartner’s own ranking retains the mainframe as “still the leading platform for certain mission-critical applications, even with the ongoing drive toward cloud-native architectures”.
Looking Forward
For UK enterprises — particularly in banking, insurance, and government — Gartner’s paper provides useful cover to delay or cancel AI-led legacy migration programmes pitched on generative AI’s code-conversion claims. UK technology leaders should scrutinise any 2026 mainframe exit business case against three tests: does the workload-by-workload analysis exist, is there a provable transformation record on COBOL or PL/I of comparable complexity, and what does the post-migration performance and throughput contract look like? If any of those is weak, the Gartner forecast applies. Expect vendor consolidation in the “AI-led code transformation” category to begin visibly in 2027.