IBM shrinks z17 mainframes into rack-mounted systems to bring enterprise AI to smaller organizations

IBM teases new rackable mainframes that 'complete' the z17 family

IBM has posted an announcement that teases the introduction of rack-mounted and single-frame versions of its z17 mainframes. Big Blue's documentation for the z17 mentions only a model called the ME1 that scales from single-rack to four-rack systems. The Tuesday announcement mentions bundles called the z17 ME2 and IBM z17 MER, respectively a single-frame and rack-mounted affair. The announcement doesn't offer any specs but says they "complete the IBM z17 product family." IBM asserts that a z17 ME2 packing a single integrated accelerator for AI on an OLTP network "matches the throughput of running inferencing on a compared remote x86 server with 21 cores." That's not enormously impressive given that x86 processors can now pack hundreds of cores, and servers that use them are generally cheaper than mainframes. Big Blue also claims the z17 ME2 can "process up to 2.5 million inference operations per second with 1 ms response time using a Credit Card Fraud Detection Deep Learning model. The announcement claims the new boxes, and numerous other enhancements, "are designed to simplify how clients adopt and scale artificial intelligence, strengthen security, support compliance efforts, and modernize operations across mission-critical enterprise workloads." IBM will start taking orders for the new machines on August 12. Earlier this year, IBM CEO Arvind Krishna told investors that the z17 is IBM's best-selling mainframe in 20 years and attributed strong sales to big iron being cheaper to operate than alternative platforms - especially for AI applications. Analyst firm Gartner recently told The Register that adopting a mainframe can be cheaper than sticking with VMware. The firm has also published guidance suggesting that more than 70 percent of projects that aim to replace mainframes by using generative AI to rewrite COBOL code are likely to fail. "By 2030, 75 percent of vendors operating in the 'mainframe exit' market will either pivot their business models or cease to exist," Gartner wrote. Big Blue had a big day on Tuesday - at least in terms of mainframe announcements - as it also published news of improved COBOL optimization and an update to its z/VM hypervisor. ®

IBM debuts compact z17 mainframes and LinuxONE servers for on-premises enterprise AI

IBM debuts compact z17 mainframes and LinuxONE servers for on-premises enterprise AI IBM Corp. today announced a massive expansion of its on-premises enterprise infrastructure lineup, with new single-frame and rack-mount versions of its iconic z17 mainframes and more compact LinuxONE 5 server platforms. The new lineup includes the IBM LinuxONE Rockhopper 5 and IBM LinuxONE 5 Express systems, which are the first to be offered in more flexible form factors at a lower price point. Meanwhile, the smaller z17 systems, which have been reduced to a standard 19-inch rack-mountable size, are part of an effort to democratize access to the venerable mainframe so that smaller organizations can take advantage of its high-performance processing and industrial-grade security. IBM Chief Product Officer Tina Tarquinio introduced the new mainframes in a blog post, saying that they're designed to address a couple of converging headaches for enterprises of all sizes: the need to run sensitive, data-intensive artificial intelligence workloads on-premises to adhere to strict regulatory requirements, and the pressure to reduce their energy costs and data center footprints. Tarquinio said the company has solved these challenges with some impressive hardware engineering feats. For instance, the new single-frame and rack-mount z17 systems squeeze massive amounts of computing power into a smaller physical space. They can support up to 82 cores and provide a hefty 18 terabytes of memory across just two processor drawers. They're powered by IBM's most advanced Telum II processor, which delivers a 10% single-threaded improvement and up to 20% greater capacity overall than the older z16 mainframe systems. Similar gains have been made with the new LinuxONE systems for Linux computing environments. IBM Fellow Marcel Mitrain said in a separate blog post that the LinuxONE Rockhopper 5 platform can perform the workload as 23 similarly sized x86 servers. They're geared toward transaction processing and AI workloads, and can reduce power consumption by up to 83% compared to standard servers. When running AI-infused online transaction processing or OLTP workloads on Red Hat OpenShift, the new systems can do the same amount of work as a standard x86 setup with four times fewer cores. As for LinuxONE 5 Express, this is a preconfigured and low-priced entry point for smaller enterprises that's designed to slot in alongside existing x86 servers without any specific cooling or power systems upgrades required. Looking beyond the impressive performance specifications, the launch of the new systems positions IBM as a key player in the rapidly maturing enterprise AI infrastructure market. As AI workloads scale beyond just pilot projects and go into full production, many businesses have realized that moving huge amounts of sensitive data to the public cloud invites substantial challenges around latency, cost and governance. IBM's z17 and LinuxONE platforms give enterprises a way to run AI workloads without moving any data to the cloud. It means they can perform real-time AI inference, fraud detection and automated decision-making on-premises, in the same facility where their transactions are processed. IBM supports this capability with the IBM Spyre Accelerator, which makes it possible to run generative AI and agentic applications right alongside their transactional data workloads, within a robust security boundary. A number of factors are converging that should make IBM's slim form-factor platforms a very compelling offering for enterprises, said Holger Mueller of Constellation Research. These include the need to bring applications closer to the data for lower-latency, data residency requirements in regulated industries, and also the need to bring AI on-premises for compliance reasons. "With the Telum processor and the Spire accelerator, IBM has the chip platform to enable this in a smaller, single-rack footprint for both mainframes and Linux platforms," the analyst said. "Senior executives will want to consider this because IBM brings nimble system performance to their on-premises data, with support for AI on both the z17 and LinuxONE platforms. They're likely to be very attractive to IBM's install base, which increasingly wants and needs AI automation." The new hardware is being launched together with an updated software stack that includes a new infrastructure management platform. It supports infrastructure-as-code and OpenTelemetry, enabling rapid configuration on the fly and deep operational visibility. Security has been boosted too, with IBM protecting the new systems with quantum-safe cryptography as a standard feature. There are also new tools for advanced cryptographic operations and CBOL application modernization built into the new systems. Net-net, IBM's mainframes and Linux systems aren't just getting smaller, they're paving the way for smaller organizations to operationalize AI at large scale more effectively.

IBM Announces New z17 And LinuxONE 5 Configurations With Single Frame And Rack Mount Systems

IBM announced new IBM z17 and IBM LinuxONE 5 configurations, marking the first time IBM is offering rack mount alongside single frame systems across its full Z and LinuxONE portfolio. The expanded IBM z17 and LinuxONE 5 portfolios now offer a wide range of deployment options, engineered with the same flagship performance, security, and ecosystem standards. New single frame and rack mount options provide additional ways for organizations to position their infrastructure where it fits best for their business needs, helping support flexibility and operational efficiency. Organizations processing highly sensitive workloads at scale are facing record-low data center vacancy and rental rates exceeding $400 per kW/month, according to CBRE's 2026 Global Data Center Trend Report. Enterprises can use IBM z17 and LinuxONE 5 rack mount and single frame systems to address these challenges, optimizing their data center real estate to meet today's realities. The new IBM z17 and IBM LinuxONE 5 configurations support up to 82 cores and 18 TB of memory across two processor drawers, representing about a 20% increase in core count and 12% increase in memory capacity. Single processor capacity of IBM z17 ME2 provides full speed IBM z/OS configurations including 10% greater throughput per core than IBM z16 A02 with some variation based on workload and configuration. Clients have the flexibility to co-locate IBM and non-IBM equipment to achieve the best fit-for-purpose installation in their data center. Each system is designed to help organizations reclaim space, improve energy efficiency, and integrate seamlessly into existing environments: IBM z17 single frame is a fully packaged solution in an IBM rack and intelligent power distribution units (iPDUs), delivered as a complete enclosed unit ready to deploy, now with the added flexibility for clients to co-locate other technologies within the frame. IBM z17 rack mount allows clients to install IBM Z components directly into their own industry-standard rack, with built-in flexibility for co-location with other technologies. IBM LinuxONE Rockhopper 5 is the scalable, multi-drawer LinuxONE system for high-density workloads, with on-chip AI acceleration, confidential computing, and post-quantum cryptography available in both single frame and rack mount configurations. IBM LinuxONE Rockhopper 5 rack mount and Express offerings deliver enterprise-grade Linux, confidential computing, and on-chip AI acceleration in a compact 18U configuration. Designed for organizations supporting a smaller set of workloads, the offering provides a cost-efficient entry point that can scale as business grows, while prioritizing security, resiliency, and performance. As with the rest of the IBM z17 and LinuxONE 5 portfolio announced last year, the single frame and rack mount systems deliver advanced multi-model AI inferencing through the IBM Telum II processor, Red Hat OpenShift AI and the IBM Spyre Accelerator to deliver in-transaction predictive AI and generative AI. IBM is announcing new software and management capabilities designed to help clients simplify infrastructure operations, reduce the skills required to run the platform, and get more value from the workloads already running their business. IBM Infrastructure Management for Z and LinuxONE brings together provisioning, configuration, and operations together. Enterprises can now leverage Terraform and widely adopted Infrastructure-as-Code that are engineered to automate infrastructure deployments, and orchestrate configurations in a unified user interface with a simple visual I/O topology and configuration while addressing the number of specialists required. IBM COBOL Elevate for z/OS is built to simplify modernization and optimize performance for COBOL applications running on IBM z17, helping clients get more value from the applications they depend on with no rewrites or specialized skills required, with availability beginning September 18. Post-quantum cryptography security is now standard on z17 and LinuxONE Rockhopper 5 systems, leveraging post-quantum cryptography, confidential computing, and enterprise-wide secrets management. New IBM Crypto Discovery & Inventory capabilities are engineered to simplify security operations by giving security teams a consolidated view of their cryptographic posture across the enterprise, helping them prepare for post-quantum standards with end-to-end visibility. The new z17 single frame and rack mount configurations, IBM LinuxONE Rockhopper 5, and IBM LinuxONE 5 Express will all be generally available August 12, 2026. IBM Infrastructure Management for IBM Z and IBM LinuxONE will be generally available August 14, 2026. IBM COBOL Elevate for z/OS will be generally available September 18, 2026.