The leadership disconnect paralyzing enterprise modernization

COBOL still powers almost all ATMs, 90% of the largest insurance companies and 92% of the largest retailers in the U.S. As a result, many headlines have pointed to the aging workforce as a primary concern for enterprise technology stability. But the real threat isn’t just about retiring programmers; it’s the profound leadership disconnect and competing incentives that are creating paralysis in the technology function. Veteran technologists who’ve spent decades managing mainframe systems and have been promoted to seats of power in their organizations see a fundamentally different landscape than their younger counterparts. These systems have operated reliably for 30, 40, even 50 years, weathering market crashes, regulatory changes and countless business transformations. From their perspective, modernization represents unnecessary risk to systems that demonstrably work. This isn’t just conservative thinking. It’s rational decision-making based on experience. Traditional modernization projects carry failure rates of over 70%, with notorious multibillion-dollar write-offs making headlines. For senior leaders nearing retirement, taking on a transformation project that could define their legacy becomes an unacceptable gamble. The risk isn’t to just their careers; it’s also the very real possibility of business disruption and reputational damage from failed modernization attempts. Meanwhile, incoming technologists have choices: Join digitally native companies where they’ll work with cutting-edge technologies, constantly deploy new features and capabilities, and build products reaching millions of users, or they can join legacy institutions where they’ll spend months learning obsolete programming languages just to make minor changes to old systems. Becoming caretakers of long-established systems holds limited appeal for developers who entered the field to create and innovate. The talent flight creates a vicious cycle: Organizations can’t modernize because they lack talent willing to shoulder the career risk that historically comes with large-scale modernization efforts, and they can’t attract talent because they haven’t modernized. This standoff carries significant financial consequences. Recent incidents underscore the urgency. Major U.K. financial institutions experienced more than 800 hours of IT outages between January 2023 and February 2025, with many attributed to aging mainframe systems. Airlines faced similar disruptions, affecting the global flow of people and products.   New modernization approaches enabled by generative AI are emerging, however, and fundamentally alter the risk-reward equation. These methods combine proven software engineering practices, such as continuous integration, continuous delivery and test-driven development, with modern tools that make incremental rewriting safer and more efficient than traditional “big bang” approaches.

System behavior over code translation

Instead of translating existing code directly, these methodologies focus on system behavior, capturing inputs and outputs from running systems to create behavioral specifications that serve as comprehensive tests to ensure behavioral parity between old and new systems. Modern approaches capture actual runtime behavior of existing systems, creating comprehensive test suites based on real production data flows. They draw institutional knowledge and actual system behavior from the code itself. This behavioral replication means we no longer need to rely on best guesses. We can see concrete evidence that every input produces the expected output. While this was conceptually possible before, it was not achievable on the necessary scale and might have taken decades. Now, with the rapid advancements we’ve seen in AI, we can reduce that timeline to a scale well within the tenure of an ambitious IT leader.

Incremental over Big Bang

The second key mechanism for de-risking modernization efforts is working incrementally and demonstrating each incremental slice working in production alongside the legacy system. For veteran technologists, this addresses their core concern and the reason many Big Bang projects fail: verification. Historically, organizations attempted to leap from old systems directly to an entirely reimagined “future state” in one massive effort. It would be the equivalent of ripping out old pipes while simultaneously reconfiguring the entire house. We’ve learned from the astronomically high failure rate of this approach that it is simply not tenable. Today, organizations can modernize individual components while the overall system continues operating. If a modernized component fails, the original can be restored without disrupting business systems. Together, focusing on system behavior and working incrementally delivers clean, well-tested, maintainable systems that organizations can evolve over time. This approach resolves the generational standoff that has paralyzed modernization decisions by showing success early, often and concretely. All team members can now see systems thoroughly validated against actual data flows, literally running new code in tandem with old code to demonstrate behavioral parity. The incremental nature means they’re never betting their careers on a single high-risk deployment. Each small step builds confidence and proves the methodology works. For incoming technologists, this creates an entirely different value proposition: a future-ready foundation to continually deploy cutting-edge projects into production, all while building their careers at companies with deep market presence and brand recognition — not unlike the high-velocity innovation of digital natives.