Red Hat expands agentic AI strategy with new inference, automation and sovereignty capabilities

Overview

Red Hat used its Red Hat Summit in Atlanta to announce a set of products and partnerships meant to help enterprises put AI into production, modernize infrastructure and extend open-source platforms into new environments. The centerpiece is Red Hat AI 3.4, an updated enterprise AI platform built for large-scale inferencing and agentic AI across hybrid cloud. The company is also emphasizing governance, sovereignty and security features as customers move from experimentation to production.

Key Takeaways

Red Hat AI 3.4 is positioned for large-scale inference and agentic AI deployments across hybrid cloud environments.
A new model-as-a-service capability lets enterprises expose internally approved models through governed interfaces while monitoring usage and applying policy.
Red Hat is expanding distributed inferencing and adding speculative decoding to improve performance and reduce operating costs.
New agent management and observability features include tracing for inference calls and tool usage, plus support for Model Context Protocol gateways and catalogs.
AI safety testing capabilities are powered in part by Red Hat's acquisition of Chatterbox Labs.
Red Hat believes inference, not model training, will become the dominant enterprise AI workload, driven by AI agents.

Stats & Key Facts

#Red Hat AI 3.4 is the named version of the updated platform
#Speculative decoding accelerates text generation up to threefold without reducing output quality

Red Hat expands agentic AI strategy with new inference, automation and sovereignty capabilities

What Red Hat announced

The announcements target enterprises moving AI into operation.

›A broad set of product and partnership announcements made at Red Hat Summit in Atlanta.
›Extensions of Linux and container platforms into specialized environments, including software-defined vehicles and computing in space.
›Greater operational control over hybrid cloud infrastructure.
›New governance, sovereignty and security features for production AI.

Red Hat, an IBM subsidiary, framed the announcements as helping enterprises operationalize AI, modernize infrastructure and extend open-source platforms into new settings.

Red Hat AI 3.4

›Updated enterprise AI platform designed for large-scale inferencing and agentic AI across hybrid cloud.
›Introduces a model-as-a-service capability for governed access to internally approved models.
›Administrators can govern model access through a centralized gateway, track usage and apply policies.

The model-as-a-service capability lets enterprises expose internally approved AI models through governed interfaces while monitoring usage and applying policy controls.

Leadership perspective

Executives framed AI as a major inflection point.

›CEO Matt Hicks compared AI to Linux, open source and cloud computing as a technology inflection point.
›Hicks said enterprises do not want to discard existing infrastructure investments to adopt AI.

Hicks said Red Hat wants its platforms to let customers use AI to its fullest extent while improving what runs their business today.

Four AI strategy pillars

Joe Fernandes outlined the strategy.

›Scalable inference.
›Connecting enterprise data to models and agents.
›Managing agents across hybrid infrastructure.
›Providing a unified AI platform spanning hardware and cloud environments.

Fernandes, vice president and general manager of Red Hat AI, said inferencing rather than model training will become the dominant enterprise AI workload, and that AI agents will drive inference demand exponentially.

Inference performance and cost

›Expanded support for distributed inferencing.
›Speculative decoding introduced to improve performance and reduce operating costs.

Speculative decoding is described as a large language model inference optimization technique that accelerates text generation up to threefold without reducing output quality.

Agent management and governance

New tooling supports agents as managed assets.

›Agent management and observability features, including tracing for inference calls and tool usage.
›Support for Model Context Protocol gateways and catalogs.
›Prompt management tools that treat prompts as managed enterprise assets.
›An evaluation hub to assess model and agent quality, safety and accuracy.

The platform uses MLflow for experiment tracking and lifecycle management. Integrated AI safety testing capabilities are powered in part by Red Hat's acquisition of Chatterbox Labs.

Enterprise focus

›Enterprises are focused less on building foundational models and more on operationalizing existing ones.
›Pretraining models from scratch is limited to a few very large organizations.

Fernandes said enterprise customers are more focused on consuming existing models and applying them to proprietary enterprise data.

Frequently Asked Questions

What is Red Hat AI 3.4?

It is an updated version of Red Hat's enterprise AI platform designed to support large-scale inferencing and agentic AI deployments across hybrid cloud environments.

What is the model-as-a-service capability?

It lets enterprises expose internally approved AI models through governed interfaces, with administrators governing access through a centralized gateway, tracking usage and applying policies.

What is speculative decoding?

It is a large language model inference optimization technique that accelerates text generation up to threefold without reducing output quality.

Why does Red Hat expect inference to dominate?

Joe Fernandes said inferencing rather than model training will become the dominant enterprise AI workload, with AI agents driving inference demand exponentially.

How does Red Hat support AI safety testing?

Integrated AI safety testing capabilities are powered in part by Red Hat's acquisition of Chatterbox Labs.

Red Hat is positioning its updated platform and governance tooling to help enterprises move AI from experimentation into production across hybrid cloud.