IBM has announced the launch of its new Power11 chips, introducing a new generation of servers designed to meet the growing demands of artificial intelligence (AI) workloads while maintaining the reliability and security that IBM Power systems are known for.
The Power11 platform is IBM’s first major update to its Power processor line since 2020. Available from the 25th of July, IBM says it will bring a comprehensive redesign that spans processor technology, hardware architecture, and virtualization software. With these changes, IBM aims to deliver, through the chip, what it calls the “six-nines of operationality” — 99.9999% uptime, which translates to approximately 30 seconds of downtime per year.
This level of availability is crucial for critical infrastructure sectors such as banking, healthcare, technology, and government, where downtimes cost more and continuous operations are a must. For this reason, IBM’s design of Power11’s autonomous processor enables features like automated patching and workload migration with no application downtime, live firmware updates while systems remain operational, and spare core technology for hardware fault recovery without service interruption.
And for IBM, the Power11 launch and strategy is focused on enterprise AI inference workloads as opposed to competing with tech giants like Nvidia on AI training. By integrating AI acceleration directly into the server infrastructure, IBM aims to reduce latency and operational complexity for enterprise users. This approach places Power11 as a strong alternative to x86 and GPU-based solutions for mission-critical AI applications.
“Power11 delivers AI-ready infrastructure with built-in, on-chip acceleration for inferencing and will be able to scale to support mission-critical AI workloads through the IBM Spyre Accelerator,” IBM said.
Scheduled for release in late 2025, Power11 will be the first IBM server to support the IBM Spyre Accelerator, the company’s AI accelerator chip that was introduced last year. This AI system-on-a chip is designed for inference workloads and offers up to ten times the performance of on-chip alternatives.
There’s also the new Vector Scalar Matrix (VSM) Engine v2, which doubles the performance of 8-bit and 16-bit integer operations and half-precision floating-point tasks compared to Power10. This engine enables over three teraflops of AI inference performance per core, allowing organizations to run both large and small language models directly on their business infrastructure.
In addition to the zero downtime autonomous features Power11 provides, IBM also integrated robust security features into the new chip. There’s the sub-minute ransomware detection with IBM Power Cyber Vault, the NIST-approved built-in quantum-safe cryptography in system firmware, and the Hardware Memory Tagging to guard against buffer overflows and side-channel attacks.
“The Power Cyber Vault solution is an integrated cyber resiliency solution following NIST cybersecurity framework to help identify, protect, detect, and automatically respond to cyber threats,” said IBM in its newsroom. “Cyber Vault provides protection against cyberattacks such as data corruption and encryption with proactive immutable snapshots that are automatically captured, stored, and tested on a custom-defined schedule.”
Some of the key performance improvements of the Power11 chip compared to other IBM Power chips include up to 55% better core performance over Power9 systems, up to 45% more capacity in entry and mid-range systems compared to Power 10, and higher clock speeds with entry systems running at 4.25 GHz and high-end systems reaching 4.4 GHz.
As organizations prepare to heavily deploy new technological applications in the coming years, IBM Power11’s focus on reliability, AI integration, ransomware threat detection, and energy efficiency positions IBM as a leader and reference for businesses seeking robust, continuous, and secure AI operations.
