Mainframe Modernization Strategies Shift Amid AI Optimism, According to New Kyndryl Report

Hassan Zamat, global practice leader for Core Enterprise and zCloud at Kyndryl, explains the findings of the new “2025 State of Mainframe Modernization Survey Report"

Massive, wholesale modernization projects are out; agile, iterative projects are in, according to findings from Kyndryl’s “2025 State of Mainframe Modernization Survey Report,” released Sept. 9.

Of the 500 senior IT leaders polled in the survey, 80% said their organization had changed its modernization strategy over the previous year. “That is a significant development,” Hassan Zamat, global practice leader for Core Enterprise and zCloud at Kyndryl, told TechChannel.

While the mainframe is typically associated with massive scale and complexity, smaller and nimbler modernization projects are gaining favor as they replace the “big bang” approach, where the new environment is built and the cutover takes place en masse, Zamat said.

The typical time horizon for modernization projects used to be three to five years, he noted; now, it’s 12 months, a timeframe that allows organizations to repeatedly reinvest gains from one project into another.

With AI, Modernization Picks up the Pace

It’s no coincidence that this change in approach has coincided with the rise of AI. “It used to be very difficult to make changes,” Zamat said. All the necessary documentation and assessment of dependencies between applications took time.

But now, AI is accelerating that process. In addition to converting legacy programming languages to modern ones, AI is also mapping environment-wide application dependencies and making sense of code that may have been written decades ago, Zamat said. With this kind of AI-enabled speed, enhanced business agility through automation was the top desired outcome of AI deployment (37%), the report noted.

Speed is one factor organizations are considering as they project the cost savings they expect AI to bring in the next three years—a “staggering” $12.7 billion in cost savings and $19.5 billion in revenue, according to Kyndryl’s report. “The ability to apply AI on the data that is sitting on the mainframe—it’s really opening the door for increased revenue, as well as lowering your costs,” Zamat said.

Here are the top three most popular AI use cases, according to the report:

• Optimizing performance and resource allocation (33%)
• Improving fraud detection (29%)
• Supporting improved security and testing assessment (26%)

In all, 88% of survey respondents said their organizations were planning to implement AI on the mainframe or already had. Fifteen percent said they had completed AI integration or were nearing that state, up from 4% in 2024.

Growing ROI Optimism

As AI begins to propel modernization efforts, organizations are becoming increasingly optimistic about the payoff of those projects. According to Kyndryl’s survey, their expectations for return on investment (ROI) are as follows:

• Modernizing on the mainframe – 288%, up from 114% in 2024
• Integrating with the cloud – 297%, up from 145% in 2024
• Moving off the mainframe – 362%, up from 225% in 2024

“The fact that modernization ROI has jumped 2-3x in one year, that was a surprising outcome,” Zamat said.

Despite the ROI expected from moving workloads off the mainframe, 95% of organizations reported either maintaining their mainframe usage or increasing it. While 98% reported they are moving some applications off the mainframe—an increase of 2 percentage points over 2024—the average share of applications they were moving dropped from 36% to 28%.

Only one respondent out of the 500 surveyed planned to move off the mainframe entirely. “The idea of leaving the platform today doesn’t seem to be a strategy that we’re finding,” Zamat said. “ … It’s all about, ‘Okay, I’m going to take this workload, I’m going to modernize it. Part of it will stay on the mainframe; the other part will go to the cloud. And I want to be able to work on both.”

Hybridization Changes Perceptions, Heightens Skills Challenges

The hybridization of the mainframe explains a statistical paradox revealed by the survey. While 56% said they had increased their usage of the mainframe, the perceived strategic importance of the platform dropped by 11%. But that drop represents dilution rather than diminishment, Zamat said.

“It is not surprising,” he said, “because you are expecting now to think of the mainframe as part of a bigger hybrid environment. It’s no longer just this monolithic, siloed machine just running mainframe workload.”

This hybridization is exacerbating skills shortages. Seventy percent of survey respondents said they were struggling to find people with the right blend of skills to effectively modernize their mainframes. To address that shortage, 40% of organizations reported using AI to reduce dependency on specific skill sets, while 74% said they needed outside help to modernize.

Mainframers aging out of the workforce without being replaced is only a partial explanation for skills shortages. Another factor is that roles require a wider blend of skills as the mainframe becomes increasingly integrated with other platforms, according to Zamat.

For instance, it’s no longer sufficient to simply be a mainframe DBA or a cloud DBA, he explained. “Now you need to be able to describe both when you’re talking modernization,” Zamat said. “Those skills—in terms of knowing the mainframe, the hyperscalers, and also using technology such as AI—those are the skills that are very high to find.”

---