Legacy mainframes remain one of the most underestimated parts of the enterprise architecture of large-scale organizations. Every second, they process thousands of transactions, store data and handle operations. And there are still many businesses, from data-critical industries as well, that ignore the importance of mainframe modernization. As a result, outdated system design and technologies lead companies to high maintenance costs and make modernization seem daunting.
AI-assisted mainframe application modernization can solve this challenge. Leveraging generative AI, agentic automation and hybrid-cloud architectures, companies can turn their legacy codebases into strategic business assets.
The high cost of legacy tech
According to a 2025 BusinessWire analysis, the average global enterprise wastes over $370 million annually due to inefficiencies and technical debt rooted in legacy systems. The report breaks down this figure into more than $130 million lost through delayed modernization timelines, around $60 million through failed transformation efforts and over $50 million through ongoing maintenance and integration work.
Across industries, the scale of the problem remains striking. Deloitte estimated in 2020 that organizations spend 55% of their IT budgets maintaining outdated systems, leaving less than one-fifth available for innovation. In financial services, legacy technologies — often COBOL-based platforms — consume 70% to 75% of annual IT spending. This restricts modernization and slows the delivery of digital services that customers increasingly expect in real time.
The public sector faces similar challenges. The U.S. Government Accountability Office (GAO) reports that 10 of the most critical federal legacy systems cost around $337 million per year to operate and maintain, with some agencies allocating up to 80% of their IT budgets to keep decades-old systems running. Many of these applications were developed more than 40 years ago and rely on aging hardware and code bases that are difficult to secure or integrate with modern platforms.
Beyond direct maintenance costs, technical debt imposes a hidden tax on productivity. Research by CAST Software suggests that organizations worldwide would need an estimated 61 billion workdays to fully eliminate accumulated technical debt — and nearly 45% of enterprise code is considered fragile or high-risk to modify.
These numbers underscore a common reality: Every year that modernization is postponed, costs and risks multiply. Aging systems drive up hardware and support expenses, drain IT budgets and make talent replacement harder as experienced legacy specialists retire. The financial and operational impact is no longer theoretical, but measurable and growing.
The role of AI vs. humans in transformation
From my perspective, AI is redefining how enterprises approach mainframe modernization. Long-term processes can now be accelerated and simplified through intelligent automation. Generative AI and code-assist tools interpret legacy code, understand its structure, suggest optimized patterns and generate new modules.
I’ve seen how IBM Watsonx Code Assistant, GitHub Copilot for COBOL, AWS Mainframe Modernization Service and Google Gemini are already transforming development workflows. These tools analyze millions of lines of COBOL or PL/I code in minutes, detect interdependencies across applications and automatically translate legacy components into Java, C# or Python. They identify redundant functions, recommend optimized algorithms and flag potential performance or security issues before they surface in production.
But the real value is in combining AI with human expertise. AI handles the repetitive, high-volume analysis and code generation tasks, freeing developers to focus on strategic design, system architecture and innovation. This combination of machine efficiency and human insight has cut analysis time in modernization projects I’ve worked on by more than half and decreased the possibility of human error during refactoring.
What I find most powerful is that AI doesn’t replace engineering judgment, it enhances it. Developers can interact with these systems conversationally, asking for explanations, testing scenarios or reviewing suggested refactoring approaches. This collaborative model accelerates delivery and builds team confidence in the modernization process.
Ultimately, AI turns modernization from a technical necessity into a strategic advantage. It enables organizations to evolve their core systems continuously, creating a path toward scalable, future-ready enterprise architectures.
Automation boosts resource allocation and operational flexibility
Based on recent projects, AI-powered modernization makes a difference in human and technical resource allocation.
By automating significant portions of code analysis, generative AI solutions reduce the reliance on highly specialized legacy COBOL experts for routine or complex refactoring tasks. This allows organizations to reallocate their most experienced personnel to architectural design, strategic planning, quality assurance and innovation. Younger developers can also become productive on mainframe projects much faster, aided by AI tools.
In my experience, AI enables organizations to adapt core systems quickly, whether responding to new market demands, regulatory changes or end-users’ expectations. From my perspective, this is where AI becomes a true business driver: it shifts mainframes from rigid, slow-moving platforms into agile engines for growth.
Case study: Logistics transformation
Recently, I observed a major logistics company undertaking a large-scale modernization of its supply chain optimization platform. We decided to integrate GitHub Copilot for COBOL and our team transformed critical legacy modules into a cloud-native Java environment.
The results spoke for themselves: we reduced modernization timelines by more than 50%, which allowed us to deploy predictive delivery route optimization tools much faster. The company also gained scalability and flexibility, enabling instant booking confirmations and dynamic spot-rate calculations for shipping services.
Of course, the journey wasn’t without challenges. AI-transformed code required continuous testing and we had to implement change management to help developers trust AI outputs. Additionally, undocumented or “spaghetti” legacy code demanded careful handling.
This experience reminded me that AI is powerful, but it’s not magic. Success depends on combining AI with structured processes, rigorous testing and human oversight.
Critical success factors for AI-assisted mainframe application modernization
I believe organizations that succeed in AI-assisted modernization share several traits:
Clear vision
Modernization must align with concrete business goals, whether faster customer service, regulatory compliance or new product innovation.
Appropriate toolkit
The right AI tools must match the legacy environment, modernization goals and enterprise infrastructure. I’ve seen projects fail when tools were chosen without considering their compatibility with legacy code.
Skilled teams
Developers must be trained to validate, refine and trust AI outputs. Without this human oversight, AI’s potential is wasted.
Agile approach
I always recommend starting with high-value areas and delivering early wins, rather than attempting to reach a full-scale “big bang” modernization.
Culture of innovation
Leadership support and a willingness to embrace experimentation are critical to overcoming resistance to change.
How to address legacy challenges?
It’s been my experience that code quality, organizational inertia and skill gaps are the main obstacles to successful modernization. To address these:
- Conduct a code review to figure out challenges.
- Implement comprehensive testing to ensure the functional equivalence of AI-transformed code.
- Manage change effectively through communication, pilot programs and involving developers in AI tool selection and fine-tuning.
- Invest in upskilling the existing staff, combining mentorship, training programs and strategic hiring to bridge AI and modern development expertise.
Future trends in AI-assisted mainframe application modernization
Looking forward, I see several emerging AI capabilities transforming enterprise digital transformation over the next three to five years:
- Autonomous AI-powered legacy data and mainframe modernization
- Deep integration of AI to intelligently re-architect applications for cloud-native performance
- AI and human collaboration across all development stages
- Ensuring ethical AI with government compliance
From legacy to leadership
AI-assisted mainframe modernization is a strategic imperative. Organizations that embrace this approach gain agility and innovation and reduce costs in long-term vision. The enterprises that succeed will combine the best of AI and human expertise, turning mainframes from perceived constraints into exceptional end-user experiences.
From where I stand, the question is not whether to modernize, but how quickly organizations can leverage AI to transform legacy assets into future-ready platforms that drive growth, innovation and operational excellence. The sooner leadership acts, the sooner these companies can claim leadership in the digital economy.
This article is published as part of the Foundry Expert Contributor Network.
Want to join?
Read More from This Article: Using AI to modernize mainframes: Turning legacy tech into a strategic advantage
Source: News