Gartner Warns AI-Powered Mainframe Migrations Are a Bubble Ready to Burst

Gartner has issued a stark warning about the viability of AI-powered mainframe migration projects, predicting that the current enthusiasm for using generative AI to help organizations exit their mainframe environments is creating a bubble that's set to burst.

According to a new report from the analyst firm titled "Too Big to Fail: Why Mainframe Exit Projects Are Likely to Fail in the Age of Generative AI," more than 70 percent of mainframe exit projects initiated in 2026 will fail to produce the intended benefits. The firm also forecasts that by 2030, 75 percent of vendors operating in the "mainframe exit" market will either pivot their business models or cease to exist entirely.

The report, authored by Dennis Smith, Alessandro Galimberti, and Tobi Bet, identifies several critical factors behind this pessimistic outlook. The primary issue is the fundamental complexity of mainframe systems, which serve as home to mission-critical applications and decades' worth of data. "For most large-scale enterprises, the sheer volume and interconnected complexity of this data make wholesale migration a physical and financial impossibility," the analysts wrote.

While Gartner acknowledges that generative AI is very useful for helping organizations detect and describe technical debt within their mainframe environments, the technology has "significant limitations when it comes to the automated conversion and migration of legacy code." The analysts emphasize that AI tools don't account for the unique capabilities that mainframes offer, such as ensuring that the same performance and throughput is achieved after migration.

One concerning factor identified by Gartner is the role of investor pressure in driving AI adoption for mainframe migrations. "Aggressive investor demand for AI capabilities as the sole indicator of a vendor's long-term health forcing vendors to deploy AI even where unnecessary," the report states. This pressure meets users' concerns about difficulties finding staff to operate mainframes and the burden of technical debt, creating a perfect storm where AI can sometimes feel like the answer even when it isn't.

The gap between the "marketing promise" of generative AI and its actual capabilities in code transformation is particularly problematic, according to Gartner. "The stakes of a miscalculation are immense," the analysts wrote. "Poor decision making regarding migration is not merely a budgetary overage; it is a threat to business and operational continuity."

Gartner advises organizations to avoid falling for "seemingly magical solution" migration promises and instead adopt a "platform-smart approach" that involves diligently evaluating workloads and choosing the best platform for the relevant work. The firm warns that ignoring this approach leads to massive technical debt and critical enterprise risk.

This assessment will likely be welcomed by IBM, which has seen its stock price slide sharply after Anthropic touted the COBOL-conversion powers of its Claude Code tool, sparking speculation about the mainframe's future. However, IBM's revenue, which is currently swelling due to unusually high mainframe sales, suggests that big iron still has plenty of life in it. Gartner's paper ranks the mainframe as "still the leading platform for certain mission-critical applications, even with the ongoing drive toward cloud-native architectures."

The report concludes that the drive to abandon the mainframe is diminishing as customers increasingly recognize the near-impossibility of a mainframe exit at an acceptable cost and risk. Instead of pursuing wholesale migration, Gartner suggests that most organizations should continue to look for ways to improve their existing mainframe systems rather than making a move.

This analysis serves as a cautionary tale for organizations considering AI-powered mainframe migrations and for investors pouring money into vendors promising easy exits from legacy systems. The mainframe may not be as obsolete as some tech commentators suggest, and the path to modernization may be more complex than simply applying AI tools to decades-old codebases.