CIQ는 AMD GPU 기반 클러스터 간 워크플로우를 조정할 수 있는 권한 부여형 AI 에이전트를 지원하는 MCP 서버를 도입한 Fuzzball 4.2를 출시했다.
RENO, Nev., Sept. 3, 2026 — CIQ, the enterprise software company behind Rocky Linux, today announced Fuzzball 4.2, a release of its turnkey sovereign AI and HPC orchestration platform designed to transform training and inference jobs from fragile infrastructure experiments into reliable production workloads. The update integrates the platform into the agentic AI ecosystem, enabling direct operation by AI agents while allowing internal workloads to orchestrate additional tasks autonomously. Recognizing that long-running AI workloads demand infrastructure-aware reliability, Fuzzball 4.2 embeds that resilience directly into the platform, allowing enterprise teams to build, train, and serve proprietary models on existing hardware.
Fuzzball 4.2 introduces agent-ready infrastructure through a new MCP server that allows AI agents to inspect a Fuzzball environment, draft, submit, and monitor workflows. Writes, workload execution, and destructive actions each require explicit operator permission, ensuring every agent operates within defined boundaries. Concurrently, workflows can now interact directly with the Fuzzball API. Each job and service container automatically receives a workflow-scoped API credential and the cluster’s API address, enabling running jobs or services to submit, track, and terminate additional workflows without separate logins or long-lived shared secrets. This capability enables more complex Fuzzball-native pipelines and hosted services that broker access to managed compute for external users. Together, these updates allow an AI agent to drive the platform, while the platform can simultaneously run an agent performing identical functions.
“Enterprises have moved their AI in-house faster than they have built the operations to run it. What decides whether that works is the software layer above the hardware, and Fuzzball 4.2 is that layer. Teams build, train, and serve on one platform, on infrastructure the organization governs, with their models and their data staying where they belong,” said Gregory Kurtzer, CEO and founder of CIQ and founder of Rocky Linux.
Fuzzball’s updated scheduling engine collects host and GPU health signals, assigns a reliability score to every node, and factors that data directly into scheduling decisions. When a node fault is detected, jobs automatically restart elsewhere. The platform also automates the controlled cordoning, draining, and replacement of degraded hardware, maximizing productive output from expensive infrastructure.
Production AI infrastructure must safely serve multiple teams simultaneously. Organization-level storage isolation, default storage provisioners, and compute-policy grants enable enterprises and service providers to share large GPU environments while maintaining strict boundaries around data, infrastructure, and placement rights. These boundaries are enforced through architecture rather than policy alone; Fuzzball executes workloads in an unprivileged, rootless environment, an approach inherited from multi-user high-performance computing systems. Additionally, per-workflow accounting for compute, storage, and network egress provides operators with complete visibility into resource consumption.
These capabilities transform how organizations utilize existing accelerator hardware. A single GPU environment can now serve multiple teams concurrently, with strictly enforced boundaries around each team’s data, storage, and placement rights. Inference services operate as persistent endpoints within the same platform used for training, alongside the pipelines and agents that drive them. Models, weights, and data remain on organization-owned and governed infrastructure—whether on-premises or in any supported cloud—with consistent workflow definitions across environments and fully auditable actions.
AMD ROCm GPU support expands Fuzzball’s ability to discover and schedule workloads across both NVIDIA and AMD infrastructure. As GPU vendors increasingly acquire the schedulers their customers rely on, Fuzzball remains hardware-agnostic and independent of any single accelerator supplier. This allows enterprises to purchase hardware that best fits their workloads without altering their operational workflows.
Fuzzball 4.2 is available now. It deploys to AWS, Google Cloud, Oracle Cloud, CoreWeave, and Azure, as well as on-premises clusters built with Warewulf, VMware, or bare metal. To get started, visit ciq.com/products/fuzzball.
CIQ is the founding support and services partner for Rocky Linux and a leading provider of enterprise Linux infrastructure. The company delivers commercially supported Linux offerings, high-performance computing solutions, and AI infrastructure to enterprises, government agencies, research institutions, and supercomputing centers worldwide. CIQ’s product portfolio includes the Rocky Linux from CIQ (RLC Pro) operating system family, Ascender Pro for IT automation, Fuzzball job-based container orchestration, Warewulf cluster provisioning, and Apptainer, the leading container system for high-performance computing. For more information, visit ciq.com.