IBM’s Power11 Chips Aim to Crush Downtime IBM unveiled Power11 servers today, promising 30-second annual outages and built-in AI inference security. Can energy-wise chips finally dethrone GPU stacks in corporate datacenters? For more AI News, follow @dylan_curious on YouTube.
IBM debuts new servers based on custom Power11 processor https://t.co/DnubaeTgT6
LAUNCH ALERT 🚀 “SambaManaged delivers high-performance AI with just 10kW of air-cooled power and minimal infrastructure changes – making rapid deployment simple for any data centre.” — Abhi Ingle Read more from @DataCentre_Mag 👇https://t.co/yNLz0vw0Kk
IBM has introduced its Power11 processor family, the first significant refresh of the company’s Power line since 2020. The new chips will underpin four servers that begin shipping on 25 July and will also be offered as virtual machines in IBM Cloud. Targeted at data-centre customers in sectors such as finance, manufacturing and healthcare, Power11 is pitched as a simpler way to deploy artificial-intelligence inference. IBM says the silicon delivers up to 55 percent higher per-core performance than Power9 while consuming less energy than rival x86 offerings from Intel and AMD. Reliability and security are the centrepieces of the launch. Systems built on Power11 are designed for 99.9999 percent availability, require no planned downtime for routine software updates and average just over 30 seconds of unplanned outages per year. Embedded ‘Cyber Vault’ features can detect and respond to a ransomware attack in under one minute. IBM plans to pair Power11 with its Spyre AI accelerator in the fourth quarter, further boosting inference speeds. By releasing high-end, mid-range and entry models simultaneously—and making the platform available in the cloud—IBM aims to broaden the appeal of its Power architecture as enterprises seek to run mission-critical workloads alongside emerging AI applications.