Mainframes are set to play a significant role in the future of AI, with many organizations already preparing for this integration. Despite being 60 years old and not originally designed for AI workloads, a Kyndryl survey reveals that 86% of business and IT leaders are either deploying or planning to deploy AI tools on their mainframes. Furthermore, 71% of respondents are currently using AI-driven insights for modernizing their mainframe systems. Although the trend of utilizing AI on mainframes is still emerging, experts like Petra Goude from Kyndryl indicate that companies are focusing on integrating their critical mainframe data with AI tools. Almost half of the surveyed plan to leverage generative AI to transform mainframe data into actionable insights. Goude emphasizes that organizations will either move data to cloud-based AI models or migrate those models to the mainframes, and anticipates a convergence of both approaches. AI also aids in modernizing mainframe strategies, whether by moving workloads to the cloud or updating outdated code.

Most respondents support keeping essential workloads on-premises while transitioning others to the cloud, suggesting a balanced hybrid IT environment rather than an all-or-nothing strategy. Lisa Dyer from Ensono notes a growing interest in using AI for modernization, particularly in translating and updating mainframe code. This aligns with Chris Dukich of Display Now, who highlights the efficiency AI brings to tackling the complexities of modernizing legacy systems. The early adoption of AI workloads on mainframes is gaining traction, with many organizations testing concepts this year. Running AI directly on mainframes housing critical data can minimize data movement risks and enhance performance due to low latency and high throughput. Both Dyer and Dukich agree that many users want to maintain their large datasets on mainframes, with AI integration seen as the next critical frontier.