Armonk, N.Y. — October 7, 2025 — Leads & Copy — IBM (NYSE: IBM) has announced the upcoming general availability of the IBM Spyre Accelerator, an AI accelerator designed for low-latency inferencing to support generative and agentic AI use cases while maintaining the security and resilience of core workloads. The Spyre Accelerator will be generally available on October 28 for IBM z17 and LinuxONE 5 systems, and in early December for Power11 servers.
The IT landscape is evolving towards agentic AI inferencing, necessitating low-latency inference and real-time system responsiveness. IBM’s Spyre Accelerator addresses the demand for mainframes and servers to run AI models alongside enterprise workloads without compromising throughput. The accelerator supports generative and agentic AI while maintaining the security and resilience of data, transactions, and applications.
Originating from the IBM Research AI Hardware Center, the IBM Spyre Accelerator prototype has evolved into an enterprise-grade product for IBM Z, LinuxONE, and Power systems. The commercial system-on-a-chip features 32 accelerator cores and 25.6 billion transistors, manufactured using 5nm node technology. Each Spyre is mounted on a 75-watt PCIe card, allowing for clustering in IBM Z, LinuxONE, and Power systems to scale AI capabilities.
Barry Baker, COO, IBM Infrastructure & GM, IBM Systems, stated that the Spyre Accelerator extends the capabilities of IBM systems to support multi-model AI, enabling clients to scale their AI-enabled workloads with security, resilience, and efficiency.
Mukesh Khare, GM of IBM Semiconductors and VP of Hybrid Cloud, IBM, said the first chip from the IBM Research AI Hardware Center is entering commercialization to deliver improved performance and productivity to IBM’s mainframe and server clients.
Spyre Accelerators offer secured processing with on-prem AI acceleration, allowing businesses to leverage AI at scale while keeping data on IBM Z, LinuxONE, and Power systems. Coupled with the Telum II processor, Spyre enhances security, low latency, and high transaction rate processing power. On IBM Power-based servers, Spyre customers can leverage a catalog of AI services, enabling end-to-end AI for enterprise workflows.
Additional resources include the IBM LinuxONE, Power, Research and Z blogs.
Media Contacts:
Willa Hahn, willa.hahn@ibm.com
Chase Skinner, Chase.Skinner@ibm.com
Source: IBM
