A survey of 500 business and IT leaders published this week reveals 86% work for organizations that are embracing artificial intelligence (AI) technologies to accelerate mainframe modernization initiatives.
Conducted by Kyndryl, a provider of IT services, the survey also finds a third of respondents indicate they expect mainframes to also emerge as a platform for running AI workloads.
Additionally, a full 86% said they either are or are planning to add generative AI tools and applications to the platform, while 71% said they are adding generative AI capabilities to the tools they rely on to surface insights into mainframe data.
More than a third (36%) also identified opportunities stemming from AI advances as a primary driver of modernization initiatives over the next 12 months including creating faster, repeatable operations (41%), followed by improving customer experience (33%).
More than six decades after the initial launch of the mainframe, many large enterprises continue to rely on the venerable platform to process and analyze massive amounts of data. That data gravity, in turn, makes the mainframe a primary focal point to apply algorithms to analyze all that data.
Moving all that data to another platform to apply AI is simply too costly, says Petra Goude, global practice leader for core enterprise and zCloud at Kyndryl. “There is no doubt the mainframe is a good platform for AI,” she says.
Most of the survey respondents (80%) said they are in the early stages of making the AI transition on mainframes. Many organizations, for example, are still trying to determine how to incorporate probabilistic answers generated by large language models (LLMs) into tasks and workflows that are deterministic in the sense that they need to be completed the same way every time.
Of course, the mainframe itself is now part of a larger hybrid approach to managing IT environments. While continuing to invest in mainframe applications, a full 96% also said they are migrating some applications off the mainframe, with on average 36% being moved off the platform, the survey finds.
However, 53% also said on average 56% of critical workloads continue to run on the mainframe, with 53% seeing an increase in overall usage in the last year. Just under half (49%) expect usage of mainframes will continue to grow in the coming months, the survey finds.
Regardless of strategy, the return on investment for modernization projects remains high. The survey finds the one-year return on investment (ROI) average ranges from 114% when remaining on the platform to 225% for moving off the platform, for a collective savings of $11.9 billion annually.
The survey, however, also noted more than three quarters of respondents (77%) are using external providers to enable mainframe modernization projects, with 25% of respondents reporting their organizations are wrestling with mainframe skill gaps.
Ultimately, organizations need to determine which types of workloads can be most optimally run on what platforms, says Goude.
The challenge, as always however, is being able to move beyond many of the biases that have long existed among IT professionals, that for one reason or another always seem to prefer one specific platform over another.
