CIQ is expanding Rocky Linux beyond its community roots with commercially supported and AI-focused versions designed for enterprise and high-performance computing workloads.
In a series of moves this spring, the company introduced RLC Pro and AI-focused variants, along with an AMD partnership aimed at simplifying enterprise AI deployments.
With RLC Pro, a GPU-optimized edition, CIQ aims to reduce the operational overhead of managing AI infrastructure. By bundling security and support as standard, the platform goes beyond binary compatibility to reduce the overhead of manual tuning, positioning the OS as a key factor in hardware ROI for production-scale AI.
CIQ argues that the enterprise Linux market has long treated features such as long-term support (LTS), compliance, and direct engineering access as add-ons, forcing organizations to choose between unsupported deployments or paying for capabilities they consider essential for production.
“AI is driving a true inflection point for enterprise infrastructure,” said Gregory Kurtzer, CEO of CIQ and co-founder of Rocky Linux. “RLC Pro brings together the infrastructure capabilities organizations need into a use-everywhere model that prioritizes efficiency, security, and usability.”
He believes long-term support, FIPS validation for cryptographic certification, and direct engineering support should be standard.
Rocky Linux Pro Targets Enterprise Gap
In late February, the company launched Rocky Linux from CIQ Pro (RLC Pro), a commercially supported enterprise Linux distribution. CIQ says it is designed to support production-grade infrastructure, including AI, HPC, and security-focused deployments, and to provide a path from community-based use to enterprise production.
CIQ says enterprises deploying Linux typically face a trade-off: rely on community distributions with limited support or pay for commercial offerings that bundle critical features like long-term support, compliance, and engineering access. RLC Pro aims to reduce that trade-off by including those capabilities in the base offering, along with vendor-backed support and faster access to fixes.
The real transformation, according to Bjorn Hovland, president of CIQ, is what this enables enterprises to do.
“Deploy in regulated environments immediately. Plan infrastructure investments with confidence. Free engineering talent to build products instead of managing operating systems. RLC Pro doesn’t just change how organizations consume Linux; it changes what they can accomplish with it,” he told LinuxInsider.
Challenging Enterprise Linux Pricing
According to Brian Dawson, director of product management for Linux at CIQ, the legacy commercial model centers on vendor revenue rather than customer outcomes. He said enterprises often face a choice between relying on open source without full support or paying additional subscription fees to access the features they need.
CIQ positions RLC Pro as an alternative to that model, with FIPS 140-3 validation, long-term support, and direct engineering fixes included as baseline features rather than add-ons. It is built on the latest upstream Linux kernel, tuned and validated with current user space and production inference frameworks. CIQ delivers this as CLK, the CIQ Linux Kernel.
“Over the next three years, we expect that to become the new floor for what enterprises demand from any Linux vendor. Organizations planning three- to five-year infrastructure horizons shouldn’t have to negotiate the table stakes of production readiness. The market will follow,” he told LinuxInsider.
Dawson argued that delivering out-of-tree support allows CLK to “close the 12-to-18-month lag,” referring to the delay that can occur between new GPU hardware releases and enterprise kernel support.
In benchmarks on identical hardware, LLM inference workloads ran up to 10% faster compared to stock Ubuntu Server, while image segmentation ran up to 32% faster. At cluster scale, that can translate into fewer nodes needed to achieve the same output target, he added.
AI Edition Expands the Platform
A few weeks after introducing RLC Pro, CIQ announced RLC Pro AI, an enterprise Linux distribution designed to improve GPU performance in production environments. It includes PyTorch, Nvidia CUDA, and OFED DOCA support, with additional hardware partners and frameworks planned.
As AI infrastructure becomes more central to enterprise operations, organizations across industries are moving GPU-accelerated workloads into production. In these environments, the operating system can become a limiting factor depending on driver support, kernel tuning, and workload configuration.
The OS, along with drivers, frameworks, and workload tuning, plays a key role in determining how much performance GPU hardware delivers, and for many enterprises, some of that performance remains untapped.
“The OS is where GPU ROI is won or lost, and the industry has ignored it for too long,” said Kurtzer. “Organizations are committing hundreds of millions of dollars to GPU infrastructure and running it on operating systems that were never designed for it.”
RLC Pro AI simplifies and de-risks AI infrastructure investments while improving performance, he added. Organizations running inference at scale can see higher throughput on the GPUs they already own from day one.
AMD Collaboration Targets Faster Deployment
On March 24, CIQ announced a collaboration with AMD to deliver optimized infrastructure for AI and HPC workloads running on AMD data center hardware. The effort begins with AMD-optimized Rocky Linux (RLC+ AMD) builds from CIQ that include validated AMD drivers, ROCm support, and day-zero deployment capability.
The two companies plan to integrate AMD optimizations throughout CIQ’s infrastructure stack. As adoption of AMD Instinct GPUs grows across AI training, inference, and HPC, enterprises are increasingly seeking software stacks that are validated to match the hardware’s performance and scalability.
“Enterprise customers expect to move from infrastructure deployment to workload execution quickly,” said Kurtzer.
According to Chuck Gilbert, senior director of system design engineering at AMD, AMD Instinct GPUs and the company’s broader data center portfolio deliver high performance for AI and HPC workloads.
“By collaborating with CIQ to optimize Rocky Linux for AMD data center solutions, we are reducing time-to-deployment, simplifying operations at scale, and strengthening the enterprise ecosystem around our AI platform,” he told LinuxInsider.
Dawson said the integration work required before a single job can run has been a major hurdle. The collaboration addresses this with validated AMD drivers, ROCm support, and day-zero deployment capability.
For customers diversifying away from a single GPU vendor, that means no custom driver assembly, no hunting for toolkit versions, and no procurement delays. Hardware performance in these environments depends heavily on that validated stack.
“That’s exactly what this partnership delivers for AMD Instinct. You get the hardware performance AMD designed into the silicon without the integration tax,” he noted.
Open Source vs. Commercial Value
Dawson said the division is straightforward. The OS, kernel work, post-quantum cryptography (PQC) engineering, and AMD driver validation all go upstream to GitHub. CIQ’s FIPS-related PQC work is also open source and available there, contributing to the broader security community.
“What’s commercial is the validation, the lifecycle management, the SLAs, and the engineering accountability that production infrastructure demands,” he clarified.
That approach provides a single-vendor, commercial foundation that runs consistently from development through production, whether the workload requires a hardened security posture, GPU-accelerated AI, or general enterprise stability.
Support Differences by Version
Long-term support, compliance validation, and engineering support were part of the original RLC Pro offering. However, company officials later clarified that while these features apply to the broader RLC Pro family, RLC Pro AI does not include LTS.
A comparison chart for the RLC and RLC Pro variants is available on the CIQ website.
Loading ...
