Red Hat AI Factory with NVIDIA Accelerates Agentic AI Security Capabilities Across the Open Hybrid Cloud

Red Hat, the world's leading provider of open source solutions, today announced expanded updates in its ongoing collaboration with NVIDIA to accelerate security-focused, enterprise-grade agentic AI across the open hybrid cloud using Red Hat AI Factory with NVIDIA. By combining agent security capabilities, optimized model deployment architectures and deep hardware enablement, Red Hat and NVIDIA are empowering organizations to build, deploy and scale autonomous AI agents with trusted security and control.

AI security and safety have emerged as top requirements for enterprise deployment as organizations transition to autonomous agentic workflows. Autonomous agents require a security-enhanced agent runtime environment, scalable model execution and robust hardware acceleration to reason and act effectively in complex IT environments. The latest joint advancements from Red Hat and NVIDIA address these challenges head-on by delivering a robust, security-forward architecture for enterprise agentic AI.

Furthering agent security across the AI frontier

To establish safety and control at the core of enterprise agentic AI deployments, Red Hat is working with NVIDIA on the OpenShell Reference Architecture and supporting it through integrations with Red Hat AI Factory with NVIDIA.

This joint initiative leverages open source community innovation through OpenShell, an agent sandbox runtime and open source project for which Red Hat and NVIDIA are leading contributors, delivering standardized guardrails, operational visibility and runtime safety.

Optimizing models and open weights inference

Agents drive significant inference demand, requiring optimized execution layers and more efficient model access. To help organizations meet this demand, Red Hat and NVIDIA have developed a joint reference design built on Red Hat AI Factory with NVIDIA that incorporates Model-as-a-Service (MaaS) access control, AI Gateway and high-performance inference through vLLM and llm-d, alongside NVIDIA Dynamo, NVIDIA NIM microservices and GPU orchestration. This enterprise-ready reference implementation is based on the NVIDIA Inference Hub reference design and provides organizations with self-service model discovery, fine-grained access policy, rate limiting and subscription-based entitlement tracking. By decoupling access governance from execution, the combined solution delivers high throughput and predictable low latency across hybrid cloud environments while maintaining strict operational control, quota management, and cost attribution.

In addition, the latest NVIDIA Nemotron models, including NVIDIA Nemotron 3.5 Super VL, are fully supported in vLLM and Red Hat AI Factory with NVIDIA to give enterprise teams enhanced reasoning tools tailored to specific industry needs. By leveraging quantization techniques through Red Hat's LLM compressor, as demonstrated with the optimized FP8 and DSpark speculative decoding checkpoints for Nemotron 3.5 Super VL, organizations can benefit from significantly reduced inference latency while retaining accuracy in vLLM and full transparency into checkpoint generation.

Day 0 hardware enablement for NVIDIA Vera Rubin and beyond

To help enterprise infrastructure keep pace with demanding agent workloads, Red Hat continues its deep hardware alignment with NVIDIA to deliver Day 0 enablement for its upcoming architectures, including NVIDIA Vera Rubin. This effort empowers organizations with seamless operating system support and platform compatibility for Red Hat's open hybrid cloud portfolio, including Red Hat AI Factory with NVIDIA, at the moment of availability for NVIDIA's next-generation hardware.

Red Hat Enterprise Linux for NVIDIA - a specialized, partner-optimized edition of the world's leading enterprise Linux platform - now also provides support for NVIDIA Vera CPUs, enabling optimized performance for agent processing and computational execution. NVIDIA Vera is the CPU for agents, delivering 2X faster agentic CPU performance for longer-thinking agents, faster-learning RL systems and data-driven AI factories. As agentic AI shifts compute demands by requiring continuous reasoning loops, tool orchestration and sandbox execution, Vera CPUs address these bottlenecks by delivering faster sandbox performance and higher memory bandwidth compared to traditional processors.

Availability

All joint use cases, reference designs, and agentic capabilities are available through Red Hat's broad ecosystem of partners, including Cisco, Dell Technologies and Lenovo.

Red Hat is actively collaborating with NVIDIA and the community to integrate the OpenShell Reference Architecture into Red Hat AI and Red Hat AI Factory with NVIDIA as a native platform capability, enabling organizations to standardize agent security as a default practice.

Supporting Quotes

Joe Fernandes, vice president and general manager, AI Business Unit, Red Hat

"Enhancing the security of agentic AI will fuel the next major evolution in enterprise computing, but scaling autonomous agents requires trust, safety and seamless infrastructure integration. By deepening our collaboration with NVIDIA across open source communities like OpenShell, delivering optimized inference architectures and driving day-zero hardware enablement, we are giving IT decision-makers the trusted foundation they need to confidently deploy AI agents at scale."

John Fanelli, vice president, Enterprise Software, NVIDIA

"An agent shouldn't be enforcing its own boundaries - those controls belong outside the agent, in the runtime and the infrastructure beneath it. NVIDIA's expanded work with Red Hat combines a secure agent runtime, governed inference and deep hardware support to give enterprises a trusted, high-performance blueprint for agentic AI across the open hybrid cloud."