As organizations grapple with the complexities of deploying AI and HPC workloads, CIQ has emerged as a transformative force in enterprise infrastructure. The company behind Rocky Linux has spent the past year addressing critical pain points—from GPU software complexity to hybrid infrastructure management—while deepening its partnership with NVIDIA to deliver full-stack AI solutions that get enterprises from installation to inference 9x faster. With innovations including the security-hardened RLC-H distribution, the federated Fuzzball orchestration platform, and Warewulf Pro’s turnkey cluster deployment, CIQ is building what founder Gregory Kurtzer calls “next-generation infrastructure that delivers both security and performance without compromise.”
In this exclusive SC25 interview, Kurtzer reveals how CIQ is tackling the hidden infrastructure challenges slowing AI innovation, from achieving FIPS 140-3 compliance for federal security standards to enabling seamless workload orchestration across on-premises and cloud environments. He discusses the company’s unique position as the only provider uniting open source, security, and performance into a comprehensive infrastructure stack, and shares his vision for 2026 and beyond, where intelligent infrastructure will adapt in real time to balance performance, energy efficiency, and security demands. For HPC organizations facing budget constraints and talent shortages, Kurtzer makes a compelling case for how CIQ’s technology stack translates directly into lower operational overhead and faster return on compute investment.
++
VMblog: If you were giving an SC25 attendee a quick overview of your company, what would you say? How would you describe your company’s evolution since SC24?
Gregory Kurtzer: Here’s the overview in a nutshell: CIQ builds the next-generation infrastructure that delivers both security and performance without compromise.
As the founding partner of Rocky Linux and creator of Fuzzball and Warewulf, CIQ has had a transformative year since SC24, delivering a steady cadence of innovations addressing the most pressing challenges facing organizations deploying performance-intensive workloads: GPU software complexity, hybrid infrastructure management, cluster deployment friction, and enterprise-grade security.
2025 highlights:
- Partnership between CIQ and NVIDIA deepens: In just two months, a collaboration to provide CUDA integration rapidly expanded to enable full-stack support with Rocky Linux. What began as collaboration to simplify GPU acceleration has evolved into a comprehensive, licensed distribution that delivers NVIDIA’s entire AI and networking stack out of the box, enabling enterprises to move from installation to inference 9x faster.
- Warewulf Pro: Released in August, this turnkey enterprise solution features an intuitive web interface, pre-built node images for common workloads, and production-ready packages. Deployments that might have taken weeks can be completed in hours.
- Cloud marketplace expansion: CIQ became Azure’s Endorsed Distribution provider for Rocky Linux, launched RLC-Hardened (RLC-H) and Fuzzball on AWS Marketplace, and made Rocky Linux 9.6 and RLC-H available on Google Cloud.
- Purpose-built Linux distributions: CIQ expanded its Rocky Linux portfolio with RLC-AI, an OS engineered for AI workloads with optimized kernel configurations and confidential computing support, and RLC-H (GA August), a security-hardened distribution now available across all three major cloud marketplaces featuring LKRG, hardened packages, and verified supply chain.
- FIPS 140-3 compliance: In April, CIQ announced that Rocky Linux 8 and Rocky Linux 9.2 achieved FIPS 140-3 compliance, officially listed under the NIST Modules in Process list. This certification provides organizations with validated cryptographic security for critical workloads, meeting stringent federal and enterprise security standards.
- Fuzzball federate: Introduced in March, Fuzzball transparently unifies clusters for seamless workload orchestration across on-premises and cloud environments. “CIQ’s Fuzzball gives us the ability to implement and scale these pipelines for 100x the throughput in a fraction of the time,” said Claire Seibold, director of software and data analytics at FYR, which uses Fuzzball to power its precision medicine platform.
At SC25, CIQ will demonstrate our secure, performant software infrastructure stack for the AI era with reference architectures combining RLC, Fuzzball, NVIDIA GPUs, and Warewulf Pro.
VMblog: What key message will SC25 attendees hear from you this year? What concrete value proposition will they take back to convince their C-suite and budget holders?
Kurtzer: At SC25, CIQ will demonstrate how next-generation infrastructure delivers both security and performance without compromise. CIQ is addressing the hidden infrastructure challenges that slow AI innovation. From hardened operating systems to federated orchestration, we’re delivering open, enterprise-ready tools that cut complexity, boost security, and reduce time-to-results. Attendees will see that investing in CIQ’s stack (Rocky Linux by CIQ, Warewulf Pro, and Fuzzball) translates directly into lower operational overhead and faster return on compute investment.
VMblog: Given the current challenges facing HPC organizations-from budget constraints to talent shortages-what specific problems does your technology solve?
Kurtzer: CIQ reduces the friction of deploying, managing, and scaling HPC and AI workloads. Warewulf Pro turns cluster setup from a weeklong process into a few hours. Rocky Linux from CIQ ensures enterprise-grade compliance and security. Fuzzball orchestrates multi-cluster and cloud resources seamlessly.
VMblog: With so many vendors claiming AI-HPC leadership, what sets your approach apart?
Kurtzer: CIQ is the only company uniting open source, security, and performance into a full-stack infrastructure model. We’re making the foundations of computing stronger and more flexible. Our partnership with NVIDIA brings validated, ready-to-run Linux images with CUDA and DOCA pre-integrated. No other Linux provider offers this level of turnkey GPU enablement for both AI and HPC workloads.
VMblog: The HPC-AI convergence has accelerated dramatically in 2025. How are your solutions enabling seamless workflows across traditional simulation, AI training, and inference?
Kurtzer: CIQ’s Fuzzball platform bridges the worlds of HPC and AI. It enables unified orchestration of compute, storage, and GPU pipelines across hybrid environments. This means the same infrastructure can handle a CFD simulation one moment and an LLM training run the next. Fuzzball Federation now supports transparent cluster unification, enabling scientists and developers to scale workloads without rewriting pipelines.
VMblog: As cyber threats targeting HPC infrastructure have intensified, what comprehensive security frameworks are you implementing without compromising performance?
Kurtzer: Security is at the core of our engineering. Rocky Linux from CIQ – Hardened (RLC-H) integrates advanced runtime defenses such as Linux Kernel Runtime Guard (LKRG) and hardened_malloc to block exploitation at the kernel and memory levels. Our FIPS 140-3 certification ensures compliance with federal and enterprise cryptographic standards, while maintaining the deterministic performance required for mission-critical workloads.
VMblog: Looking toward 2026 and beyond, what transformative trends do you see reshaping supercomputing, and how is CIQ positioning for them?
Kurtzer: We’re preparing for a world where AI is the new HPC. The next frontier is intelligent infrastructure that understands workloads and adapts in real time, balancing performance, energy efficiency, and security. CIQ is laying the groundwork with a modular stack that scales from edge to exascale. By continuing to align closely with NVIDIA and the open source community, we’re building the backbone for sustainable AI and scientific innovation.
##
