Fast-tracking AI will still require mainframes to lend a hand

The most critical industries all run on the same foundation - mainframes.

With their unmatched uptime, robust security, and modular scalability, mainframes remain the backbone of mission-critical operations. They process 68% of the world's IT production workloads, acting as the digital core for enterprises navigating complex, high-volume environments.

Today's most advanced mainframe, the IBM z17, extends this role even further — capable of processing up to 24 trillion operations per second, while maintaining 99.999999% availability, or the equivalent of just 315 milliseconds of downtime per year.

At the same time, AI has evolved from a peripheral tool into a strategic necessity. No longer just a nice-to-have', AI is now redefining how organizations make decisions, modernize applications, and manage infrastructure. While current-generation mainframes support AI models, they tend to focus on specific use cases. The emergence of platforms built expressly for AI introduces an important shift – one that invites a broader conversation about how these systems coexist.

Are they meant to replace existing infrastructure, or to extend its value?

Current-generation mainframes can and will remain essential to enterprise IT – not just for their legacy, but for their ongoing ability to deliver at scale. From processing 90% of all credit card transactions to safeguarding public-sector data, these systems underpin the digital services millions rely on daily.

Their proven resilience and scalability make them ideally suited to support not only current business operations but also long-term transformation agendas. The challenge is not about replacement, but about integration – enabling mainframes to meet the evolving demands of insight-driven and increasingly automated workloads.

Advances in AI and automation are transforming how businesses operate. Predictive analytics, intelligent automation, and real-time anomaly detection are now central to everything from customer service to risk management.

In sectors like healthcare and financial services, where speed and accuracy are paramount, AI is driving faster diagnoses, fraud detection, and operational decision-making. AI-powered tools like IBM watsonx Code Assistant for Z are helping developers modernize legacy applications faster and with greater accuracy, bringing new agility to long-established systems.

As AI models grow more complex and data-intensive, organizations require infrastructure that can generate insights securely and with minimal delay – often at the point of transaction. This is where next-generation mainframes are making their mark.

An AI-capable mainframe is more than just faster hardware. It reflects a shift in architecture and design, supporting a broad range of AI workloads – including generative and agentic AI – with low latency and high efficiency. Intelligence is now being embedded throughout the stack: from hardware-accelerated inferencing to AI-powered assistants that help developers and operators work more effectively.

These platforms also support flexible deployment across hybrid environments, reducing friction as workloads shift between cloud and on-premises systems.

To fully leverage these capabilities, organizations must ensure their teams and tools evolve with their infrastructure. This includes upskilling IT talent, investing in AI literacy across business units, and adopting agile operational models. As AI becomes integral to enterprise strategy, success depends on how well teams trust and work with AI-supported systems.

Next-generation systems must also support trust and transparency. As AI becomes more deeply embedded in enterprise operations, organizations will demand infrastructure that maintains data privacy, delivers explainable results, and upholds compliance.

Meeting these expectations requires coordinated advances across compute, software, and systems management. It also requires close collaboration between technology providers, business leaders, and regulators to establish responsible AI frameworks.

For enterprises, the future lies in augmentation, not replacement. Integrating emerging technologies with core systems requires careful planning and a clear long-term strategy. Mainframes have always evolved to meet new demands, and with AI reshaping enterprise needs, they are poised to adapt once again.

This shift isn't only about technology – it also requires alignment between IT and leadership, and a commitment to upskilling teams.

Modernization should be seen as a series of targeted steps that build on what already works. By fostering collaboration and investing in skills, organizations can ensure their infrastructure continues to be a source of competitive advantage – today and into the future.