Positive News From The Kyndryl Mainframe Modernization Report
September 15, 2025 Alex Woodie
After a week of bad news, here’s something good to start your week: The costs of mainframe modernization projects are dropping while the return on investment (ROI) is rising, in large part due to AI. That’s according to Kyndryl’s 2025 State of Mainframe Modernization Survey, which documented “a degree of agility” the company wasn’t expecting in the massive projects.
As it has for the past two years, Kyndryl commissioned Coleman Parkes Research to conduct a global survey of about 500 senior leaders at mainframe-using enterprises. By mainframe, the former IBM consulting business refers to System Z systems, IBM i midrange systems, Fujitsu mainframes, and large systems from other manufacturers. The New York company is one of the biggest operators of IBM i-based private clouds in the world, and has a substantial System Z business too.
Some of the findings from this year’s report are surprising. For starters, the cost of mainframe modernization projects is going down, which seems to run counter to the prevailing environment of persistent inflation in the United States and around the world. The average cost of a mainframe modernization project dropped from $9.1 million in 2024 to $7.2 million in 2025, a decline of 19.8 percent.
High-level takeaways from Kyndryl’s report. (Source: Kyndryl)
Meanwhile, the ROI of mainframe modernization doubled, Kyndryl said. In 2024, the ROI from three forms of modernization – modernizing directly on the mainframe, integrating with cloud, and migrating off the mainframe –generated returns of 114 percent, 145 percent, and 225 percent, respectively. In 2025, those three projects are now expected to generate returns of 288 percent, 297 percent, and 362 percent, respectively.
Eighty percent of survey respondents say their modernization strategies have shifted over the past year. Kyndryl says the “big bang” modernization efforts are falling by the wayside and being replaced by “a pragmatic approach to transform in manageable phases and course correct as needed.” (This could also explain why the costs of mainframe modernization are down.)
The reasons for the project shifts are varied: 21 percent said geopolitical, macroeconomic, and regulatory developments have impacted their plans, while 16 percent said they changed gears due to budget and skills issues or due to “failed approaches.” And 11 percent cited a reallocation of budget to emerging technologies like AI, while 10 percent said they are accelerating mainframe modernization projects due to early success. All told, the changes spoke to “a degree of agility not commonly associated with technology projects of such significant size, scope, and complexity,” Kyndryl wrote.
So, what specific changes are these organizations making with their projects? Kyndryl says 43 percent are “placing more focus on modernizing directly on the mainframe and 50 percent are moving more towards a hybrid strategy.” Of the hybrid cohort, 34 percent are prioritizing integration with cloud platforms while 16 percent are increasing the pace of moving applications off the mainframe. Whole-hog migrations off the mainframe are rare; only one of our 500 respondents indicated they are trying to entirely get off Big Iron.
The ROI for different types of modernization projects. (Source: Kyndryl)
While AI may be sucking some budget from the mainframe, it’s also contributing to it. In fact, AI and GenAI projects appear to be proliferating on the mainframe (or at least with external apps that connect to mainframe data). The 500 organizations that Kyndryl surveyed said they expect to save an average of $12.7 billion in costs and increase revenue by $19.5 billion over the next three years as a result of using AI and GenAI with the mainframe.
Enterprises are turning to AI and GenAI for a wide variety of use cases, Kyndryl said. “A third of respondents (33 percent) report that they are using AI to optimize performance and resource allocation. Twenty-nine percent are using AI to improve fraud detection, and a quarter of respondents (26 percent) are using it to support better security testing and assessment,” the firm stated in its report.
The report found that 83 percent of mainframe shops have implemented or plan to implement large language models (LLMs), 82 percent are integrating development, security and operations practices (DevSecOps) into mainframes, and 78 percent are adopting agentic AI to develop, test, and manage business applications.
But AI isn’t for everyone. About 12 percent of mainframe shops report they are not yet planning to use AI or GenAI, with the biggest barriers cited including mainframe data security (47 percent), lack of AI maturity (41 percent), and technical barriers on the mainframe (40 percent).
“I believe that mainframes weren’t built with this type of workload in mind, and forcing AI workloads into that ecosystem could lead to performance issues or integration headaches that just aren’t worth it,” one CTO at a Brazilian manufacturer said, according to Kyndryl’s report. “We are keeping those [mainframe] environments stable and focusing our AI efforts on platforms that are built to handle that kind of processing and experimentation.”
Nearly every organization surveyed (98 percent) indicated they are moving some of their applications off the mainframe. The average shop is moving 28 percent of their apps off the mainframe, according to Kyndryl.
With those stats in mind, it’s not surprising that the number of organizations that consider the mainframe “very or extremely important to their business” dropped by 11 percent this year, and now sits at 78 percent. However, those same shops increased the number of applications on the mainframe by 8 percent. It turns out that what people say and what people do don’t always jibe.
“This underscores a key insight: while strategic attitudes may be evolving,” Kyndryl states, “operational reliance on the mainframe continues to grow.”
You can download a copy of the report here.
RELATED STORIES
GenAI Interest ‘Exploding’ for Modernization on IBM i and Z, Kyndryl Says
The Many Faces Of IT Modernization
Kyndryl Inspects The Modernization Plans Of IBM i And Mainframe Shops
Is That What You Think Modernization Means?
