Red Hat's AI Platform Strategy for Enterprise AI at Scale

Red Hat AI Enterprise (RHAE): A Unified Platform

RHAE is designed to be an integrated AI platform built on RHEL and OpenShift, focusing on deploying and managing AI models, agents, and applications across various environments. It aims to bridge the gap between AI experimentation and production by providing a consistent, security-hardened environment that enterprises can manage using their existing OpenShift tools and processes. This "metal-to-agent" stack approach emphasizes unification of the AI lifecycle from infrastructure to application deployment, crucial for operationalizing AI at scale.

• High-performance inference capabilities.
• Model tuning and customization features.
• Agent deployment and management tools.
• Support for diverse models, hardware, and environments, leveraging Red Hat's Linux and Kubernetes foundation.

Key Architectural Enablers in RHAE

• OpenShift AI Catalog: Provides access to compressed, production-ready versions of various large language models (LLMs) like Mistral-Large-3, Nemotron-Nano, and Apertus-8B-Instruct.
• Models-as-a-Service: A technology preview offering self-service access to privately hosted models via an API gateway, standardizing AI consumption within organizations and promoting internal model sharing.
• Broad Hardware Support: Includes generative AI on Intel CPUs for small models, and expanded certification for NVIDIA's Blackwell Ultra GPUs and AMD's MI325X accelerators.
• GPU-as-a-Service Features: Automatic checkpointing helps prevent data loss for long-running AI jobs.
• AI Python Index: A trusted repository of hardened AI tools for improved governance and software supply chain security, including tools like Docling for unstructured document processing and SDG Hub for synthetic data generation.
• Enhanced Observability and Safety: Detailed telemetry for AI workloads and integrated NVIDIA NeMo Guardrails for policy enforcement in AI interactions.

Red Hat AI Factory with NVIDIA: Industrial-Scale AI

The Red Hat AI Factory, co-engineered with NVIDIA, is an end-to-end AI stack optimized for large-scale enterprise deployments. It combines RHAE with NVIDIA AI Enterprise, targeting enterprises that need to transition from ad-hoc AI projects to industrial-scale production systems. This partnership focuses on accelerating time-to-value, optimizing performance and cost, and strengthening enterprise security posture for AI workloads.