IBM has officially launched the IBM z17, a groundbreaking mainframe engineered from the ground up to support artificial intelligence at scale.
Designed to seamlessly integrate AI across hardware, software, and systems operations, this next-generation platform is powered by the advanced IBM Telum® II processor. The z17 is poised to revolutionize enterprise AI with its real-time transaction processing and unmatched inferencing capabilities — delivering over 450 billion AI inferencing operations per day with a latency of just one millisecond.
AI Built Into the Core
The IBM z17 represents five years of intensive development, informed by feedback from over 100 enterprise clients. Its AI accelerator, embedded directly into the Telum® II chip, boosts processing power and cache by 40%, enabling a 50% increase in daily inferencing over its predecessor, the z16.
To further expand AI capabilities, IBM plans to release the Spyre™ Accelerator in Q4 2025. This PCIe-based accelerator will complement Telum® II by supporting multi-model AI workloads, including generative AI, assistants, and chatbots — all running natively on the mainframe without moving sensitive data off-platform.
AI That Enhances Human Productivity
IBM’s z17 isn’t just about raw power — it’s about user empowerment. Developers and IT teams will benefit from AI agents such as IBM watsonx Code Assistant for Z and watsonx Assistant for Z. These tools will now integrate with Z Operations Unite, enabling chat-based incident detection and real-time resolution using live system data.
Unified System Management with Operational Intelligence
IBM also previewed z/OS 3.2, its upcoming operating system update tailored for AI workloads. Expected in Q3 2025, z/OS 3.2 supports hardware-accelerated AI, NoSQL databases, hybrid cloud data processing, and real-time operational insights. Together with IBM Z Operations Unite, which aggregates key metrics using OpenTelemetry, enterprises can expect faster incident detection, impact isolation, and resolution.
Securing the Future with AI-Powered Cyber Defense
Security is foundational to the z17. IBM has introduced a suite of advanced protections, including:
- IBM Vault: A secrets management solution from HashiCorp that standardizes sensitive data protection across hybrid environments.
- AI-Powered Threat Detection: Leveraging NLP and the Telum® II processor, the z17 can identify and classify sensitive data and detect anomalies indicative of cyber threats.
Integrated Storage for Mission-Critical Data
The z17 works in tandem with the latest IBM Storage DS8000 (10th Gen), providing high-performance, secure, and modular storage designed to handle evolving enterprise workloads. This synergy ensures data is not only accessible but also optimized for AI-driven insights.
AI-Optimized Support and Lifecycle Services
To help enterprises extract maximum value, IBM Technology Lifecycle Services now offers AI-enhanced support for the z17. These services use IBM watsonx to streamline incident handling and improve resolution times, ensuring business continuity in critical environments.
Availability and Future Outlook
The IBM z17 will be generally available starting June 18, 2025. The Spyre™ Accelerator is scheduled for release in Q4 2025. With more than 250 AI use cases already in action — from financial risk mitigation to healthcare diagnostics — the z17 is set to become the heart of enterprise AI transformation.
As global competition in artificial intelligence intensifies, especially between the U.S. and China, the launch of IBM z17 aligns with the industry’s push to bolster domestic innovation and reduce technological reliance — themes also echoed in this analysis of global AI competition.
To learn more about the IBM z17 and its capabilities, visit IBM.com/z17.
