IBM launched the latest version of its dominant mainframe platform — the z17 — at its virtual IBM Z Day event this week. The z17, which features comprehensive AI integration, is designed to bring AI workloads closer to enterprise data. Embedded AI accelerators within its microprocessors support predictive and generative AI models alongside AI-powered security features. The z17 also leverages AI for critical system functions such as tuning and management. It is scheduled to be generally available in mid-June of this year.
Last month I received an advance briefing on the z17’s capabilities during an analyst event held at IBM’s new headquarters in New York. The new rollout builds on the success of the predecessor in the Z lineup, the z16, which debuted in 2022 and delivered consistent revenue growth, showing the enduring value of IBM’s mainframe offerings half a century after the technology became widespread. “This has been one of the longest-running and most consistent programs in terms of revenue growth we’ve ever seen,” noted IBM’s CFO James Kavanaugh on the company’s Q4 2024 earnings call.
Nate Dotson, principal product manager for IBM z17, framed the strategic direction for the launch: “The z17 presents new possibilities, evolving how enterprises can utilize their core systems through AI, automation and enhanced security. Our focus is on three key areas: supporting innovation and growth with AI, enabling automation and transformation for efficiency, and enhancing the security of critical data.” AI advancements for mainframes can improve and secure business interactions, leading to faster, more reliable, personalized consumer experiences. This is particularly important because many everyday consumer transactions rely on mainframe technology, including most credit card purchases as well as other steps in financial transactions such as identity verification.
(Note: IBM is an advisory client of my firm, Moor Insights & Strategy.)
The Staying Power Of The Mainframe
Despite common misconceptions about mainframes being relics of a bygone era, they remain a significant component of the modern IT landscape. Mainframes are widely recognized for high reliability, scalability and performance, which is what allows a technology that became common during the Cold War to support today’s technologies and hybrid infrastructure models. They manage substantial volumes of critical business data and process a considerable portion of the world’s transactional workloads. The IBM Institute of Business Value reports that 79% of IT executives consider mainframes essential for enabling AI-driven innovation and value creation. Modern mainframes offer significant processing power, handling up to 30,000 transactions per second, and can accommodate substantial data with up to 40 terabytes of memory. Economically, 75% of IT executives report that mainframes are equal to or better than cloud computing in terms of total cost of ownership.
According to IBM, organizations that integrate mainframes into a hybrid-by-design strategy achieve a 3x higher ROI from their digital transformation efforts. The overwhelming majority (88%) of executives from the same study believe that modernizing mainframe applications is critical for achieving long-term success in digital transformation.
How It’s Done: Enhancing Core Operations With Integrated Intelligence
The z17 incorporates several new AI-related features, including the ability to use multi-model AI in real time by combining AI models and LLMs for every transaction. Integrating multiple AI models enables deeper analysis and understanding of transactional data, increasing predictive accuracy by uncovering patterns and trends that might otherwise go unnoticed. And real-time processing ensures that decisions such as flagging suspicious transactions or approving loans are made instantly. In the case of a financial institution, this could lead directly to faster and more accurate fraud detection, credit risk assessments and anti-money-laundering measures. This protects both customer accounts and the institution itself while also improving customer experience, for example by enabling quicker loan approvals. The expanded insights and accuracy also help meet stringent regulatory requirements by ensuring thorough transaction monitoring.
Meanwhile, retailers can explore generative AI models on the mainframe to refine customer interactions and recommendations. Airlines rely on mainframes to manage booking systems and ensure smooth customer service, while insurance companies use them to process high volumes of claims efficiently.
In industries dealing with high volumes of transactions, mainframes’ extreme scalability for rapid processing can result in better customer satisfaction and retention. The new ability in the z17 to process AI tasks on the mainframe is intended to further improve latency, potentially leading to quicker service responses. The IBM Z environment also addresses key data privacy and compliance considerations. The inherent scalability of mainframes ensures consistent service during peak demand, while the built-in resilience is designed for zero downtime.
The AI infrastructure within the z17 aims to improve efficiency in AI workloads and offer enhanced throughput compared to some distributed GPU models. IBM has enhanced the z17 with the Tellum II Processor AI Accelerator, a second-generation accelerator integrated into the processor for complex computations associated with LLMs. Its architecture supports relevant computational primitives, facilitates remote AI access across the system and optimizes compute capacity through new data types. The Spire Accelerator, a dedicated AI accelerator available on a PCIe card, will be available as an add-on for z17 in Q4 2025. Designed for generative AI tasks, it has 32 cores and 128GB of LPDDR5 memory. Its architecture allows for clustering, which increases processing capacity. The accelerator supports both 4-bit and 8-bit processing for efficient execution of LLMs, reducing the need to move sensitive data off-platform.
Tina Tarquinio, vice president of product management at IBM, emphasized the strategic imperative behind this AI integration. “IBM isn’t just adding AI for technology’s sake, but is purposefully engineering AI capabilities to enhance the most critical business operations,” she said. “90% of the world’s credit card transactions run through the mainframe. 70% of the world’s [financial] transactions, by volume, run through the mainframe. We need the AI where it matters most, and for all of our clients, it needs to happen fast, it needs to happen reliably and it needs to happen securely. And this is what we’ve built the stack for.”
Security And Efficiency In Intelligent Computing
Because mainframes process such vast amounts of transactions, data security is paramount. IBM Z has historically provided a robust environment for sensitive data, and the integration of AI should bolster this capability. By processing AI workloads on-chip or within the Spire accelerator, the z17 enhances security by keeping sensitive data within the mainframe environment. The mainframe’s encryption capabilities extend to AI operations, making the environment especially suited for highly regulated industries such as financial services and healthcare.
The z17’s local processing could also lower costs by minimizing reliance on external computing resources. Mainframes have a history of providing value for enterprises by enabling them to process enormous amounts of data while keeping costs remarkably low. One indicator of this: despite handling 72% of global transaction workloads, mainframes account for only 8% of IT costs.
Cost is only one consideration. IBM designed the AI features in z17 to focus on efficient energy utilization, which is becoming increasingly critical as AI strains energy resources and impacts carbon footprints. The new infrastructure is intended to lower energy consumption for AI workloads compared to some traditional methods — and to the z16 — which could help organizations better manage operational costs and address environmental considerations.
The Ongoing Evolution Of Mainframe Technology For Enterprise Computing
IBM views AI integration into the IBM Z platform as a crucial step in the ongoing evolution of enterprise computing. In this context, IBM plans to further enhance AI capabilities for the Z platform while continuing to integrate it with hybrid cloud environments — and to explore quantum computing.
To support the adoption of these AI capabilities on the mainframe, IBM is actively developing and offering upskilling and training programs as well as strategic partnerships. The company’s Z Mainframe Skills Depot provides different learning paths and training for various roles in the mainframe ecosystem. Currently, it covers roles including system admin and application developer, with plans to expand to include a security-focused learning path in the future.
Because a significant portion of the mainframe workforce is reaching retirement age, upskilling and training programs are critical for building a new generation of professionals. By 2035, many experienced workers will have retired, creating a pressing need to transfer knowledge and expertise to younger talent. These initiatives are essential for maintaining operational continuity and enabling businesses to fully leverage AI capabilities on the mainframe. IBM’s professional development and community programs are similar to Broadcom’s Vitality Residency Program, which my colleague Matt Kimball wrote about last year. This trend of supporting mainframe talent development is a testament to the staying power of the technology and its importance in data processing.
The Future Of Experience Built On Intelligent Infrastructure
Although consumers may not directly engage with mainframes or fully comprehend their operational role, their influence remains significant. As the unrivaled dominant player in the mainframe market, IBM and its Z platform play outsized roles in the ongoing relevance and innovation of mainframes. Ultimately, these systems facilitate rapid transaction processing at great scale, personalized service delivery and secure business interactions, while also contributing to sustainable operations. For many businesses prioritizing good customer experience alongside effective risk and cost management, IBM Z represents a critical infrastructure component.
As IBM continues to integrate intelligent features that can influence the future of customer experience, it’s essential to recognize evolving perceptions of the mainframe. In the AI era, it may well become a foundational element for modern customer experience and business operations.
Moor Insights & Strategy provides or has provided paid services to technology companies, like all tech industry research and analyst firms. These services include research, analysis, advising, consulting, benchmarking, acquisition matchmaking and video and speaking sponsorships. Of the companies mentioned in this article, Moor Insights & Strategy currently has (or has had) a paid business relationship with Broadcom and IBM.
