Analyst(s): Mitch Ashley, Nick Patience
Publication Date: May 13, 2026
Red Hat used Summit 2026 to put developers and operators at the operational center of agentic AI. CEO Matt Hicks reported that every organization at Red Hat, including teams that have never written production code, has contributed code to the company’s internal agent system, signaling a shift to an everyone-is-a-builder operating model that now shapes its developer, automation, and AI announcements.
What is Covered in This Article:
- CEO Matt Hicks framed AI as the third “what do I build on” platform inflection after Linux and Kubernetes, and reported that every organization at Red Hat has contributed code to the company’s internal agent system.
- Red Hat announced Red Hat AI 3.4, Ansible Automation Platform 2.7 with a new automation orchestrator, the general availability of Red Hat Desktop, enhancements to Red Hat Advanced Developer Suite, and a new Red Hat skills repository.
- Red Hat’s deep research agent system scaled from 10 to nearly 200 production agents, with 85 percent of calls now running on open-weight models on Red Hat infrastructure.
- CTO Chris Wright introduced a metal-to-agents stack model with four platform layers running on any accelerator hardware: AI Infrastructure, Inference Services, Model Services, and Agent Services.
- Token economics framing positions the strategic move as shifting from token consumer to token provider, with frontier per-token prices falling 75 to 90 percent per year while reasoning models and agents drive consumption up by orders of magnitude.
The News: Red Hat announced significant advancements across its developer, automation, and AI portfolios at Red Hat Summit 2026 in Atlanta on May 12. CEO Matt Hicks opened the Day 1 keynote with Red Hat’s own AI journey, reporting that the company’s internal deep research agent system has scaled from 10 to nearly 200 production agents, with 85 percent of calls now running on open-weight models hosted on Red Hat infrastructure. Hicks reported that every organization at Red Hat, including legal, inside sales, and operations teams that have never written production code, has contributed code to the agent system.
CTO Chris Wright followed with a metal-to-agents architectural model for the Red Hat AI Enterprise platform, organized into four platform layers running on any accelerator hardware: AI Infrastructure, Inference Services, Model Services, and Agent Services. The new Agent Services layer at the top is defined by bring-your-own agents, AgentOps, agent identity, agent lifecycle, MCP services and tools, and agent observability.
Companion announcements operationalize the strategy across the developer and operator tiers. Red Hat AI 3.4 delivers Model-as-a-Service through the Red Hat AI Gateway, AgentOps capabilities, prompt management, an evaluation hub, and integrated safety testing powered by Chatterbox Labs, the Garak project, and NVIDIA NeMo Guardrails. Red Hat Ansible Automation Platform 2.7 introduces a new technology preview automation orchestrator, a Model Context Protocol server for Ansible, OIDC authentication for HashiCorp Vault, and an enhanced automation portal.
Red Hat Desktop reaches general availability, providing the commercially supported Red Hat build of Podman Desktop with isolated AI agent sandboxing, access to Red Hat Hardened Images, and integration with Red Hat OpenShift Dev Spaces. Red Hat Advanced Developer Suite gains a trusted software factory in developer preview, Red Hat Trusted Libraries built on SLSA Level 3 infrastructure, AI-driven exploit intelligence developed using the NVIDIA AI blueprint for vulnerability analysis, and expanded support in OpenShift Dev Spaces for AWS Kiro alongside Microsoft Copilot, Claude CLI, Cline, Continue, and Roo.
Red Hat Brings Developers, Product, and Operations to the Center of Agentic AI
Analyst Take: Red Hat’s Summit 2026 thesis lands with unusual clarity, and the organizing insight is operational, not architectural. CEO Matt Hicks opened the keynote with Red Hat’s own AI transition, and the more consequential message was about who builds. Every organization in the company has contributed code to the internal agent system, including legal, inside sales, and operations teams that had never written a line of production code. That is the operating model Red Hat is now shipping. The product announcements are the externalized version of that shift.
Everyone Is a Builder Becomes the Operating Model
Hicks was direct about how AI affects the unit of work. Software developers are moving toward designing evaluation systems, continuous integration systems, and testing frameworks that AI runs on, becoming the “humans that build the rails that AI runs on.” Managers face pressure to decompose work and delegate to both humans and agents. Everyone else gains the ability to contribute knowledge and code into the agentic system. This reframes the developer market from a headcount problem to a craft problem, and it puts platform engineering teams at the center of operationalizing it.
Production Evidence Closes the Credibility Gap
Red Hat’s own production evidence is the strongest part of the Summit 2026 message. The internal deep research agent system scaled from 10 to nearly 200 production agents. The model substitutions happened layer by layer: document search first, then hallucination detection, then safety management, and finally planning. 85 percent of calls now run on open-weight models on Red Hat infrastructure, with Nemotron Super, Nemotron Nano, and IBM Granite as the reference stack. Hicks reported that results improved after the substitutions, not just unit costs. BNP Paribas adds the external proof point with nearly $600 million in additional value from AI across 1,000 use cases. These are not pilots. They are the credibility baseline Red Hat needed to make the operating-model argument stick.
Nick Patience on Red Hat AI
The inference governance layer is where Red Hat AI 3.4 earns its enterprise relevance. The Model-as-a-Service architecture through AI Gateway, combined with the curated MCP server catalog and MCP Gateway, positions Red Hat as the governed entry point for agent-to-tool connectivity, which matters as agent credential sprawl becomes a real operational problem at scale. The safety testing integration, drawing on Chatterbox Labs, the Garak project, and NVIDIA NeMo Guardrails, reflects a recognition that automated red teaming needs to be a platform capability rather than a pre-deployment checklist item.
Red Hat characterizes sovereignty as a horizontal attribute, integrated directly across the portfolio, evidenced here by the distribution of on-premises telemetry for data sovereignty, day-0 compliance landing zones, and code-boundary controls within OpenShift DevSpaces. This approach moves beyond separate SKUs to provide regulated industries, such as defense, financial services, and the public sector, with ‘operational sovereignty.’ This framework allows for audit-ready, demonstrable oversight of model provenance, inference boundaries, and agent credentials. The IBM Sovereign Core integration, built on OpenShift and Red Hat AI and going GA during IBM Think 2026 last week, represents the most comprehensive realization of this strategy for regulated enterprise and government clients. However, the partner use cases named at Summit – Core42 in the UAE, Sopra Steria in European regulated industries, Eurocontrol for air traffic management, NxtGen in India, confirm that the primary demand signal is outside the US.
Nick Patience on NVIDIA Partnership
The Red Hat-NVIDIA partnership announced at Summit 2026 is primarily an inference infrastructure play with a sovereignty dimension. Day-0 support for Blackwell and Vera Rubin removes a recurring friction point for enterprises deploying on new accelerator generations, and validated status within the NVIDIA Cloud Partner ecosystem gives regional and sovereign cloud operators a cleaner path to building Red Hat-based AI services. The OpenShell project – an open-source initiative for AI agents – reflects a shared interest in establishing an interoperability substrate for agentic workloads before proprietary frameworks consolidate the market. The zero-trust AI architecture component is the most interesting element for regulated industries: it connects NVIDIA’s hardware security capabilities with Red Hat’s platform governance layer, addressing the provenance and auditability requirements that are increasingly showing up in procurement criteria for AI infrastructure in financial services, defense, and the public sector.
Wright’s Stack Gives the Operating Model an Architectural Shape
CTO Chris Wright’s metal-to-agents stack is the architectural answer to Hicks’ operating-model question. Four platform layers run on any accelerator hardware, bottom to top: AI Infrastructure, Inference Services, Model Services, and Agent Services. The Agent Services layer at the top comprises six capabilities: bring-your-own agents, AgentOps, agent identity, agent lifecycle, MCP services and tools, and agent observability. This is a control plane definition, not a feature list. It gives builders and operators a shared substrate and Red Hat a coherent architectural narrative that competitors will struggle to match piece by piece.
Token Economics Reframes the Buying Decision
Wright’s token economics framing is the sharpest competitive wedge in the keynote. Per-token frontier pricing is falling 75 to 90 percent per year. Consumption is climbing by hundreds of percent per year. Reasoning models burn 10 to 20 times more tokens than standard models, and agents add another 5x multiplier as they plan, call tools, and loop. The trajectory turns API-only AI strategies into open-ended cost exposure. Wright’s prescription is to move from token consumer to token provider, owning the inference layer rather than renting it. vLLM and llm-d, with llm-d delivering 3x more output tokens and 10x faster time to first token over one year, are the operational tools for that move.
Ansible Becomes the Trusted Execution Layer
The new automation orchestrator in Ansible Automation Platform 2.7 is the most consequential supporting announcement. Red Hat is positioning Ansible as the bridge between AI intent and deterministic action, with a single workflow canvas spanning task-based, event-driven, and AI-driven automation. The Model Context Protocol server for Ansible, plus OIDC authentication for HashiCorp Vault, gives agents short-lived, job-scoped credentials in place of static service accounts. In regulated environments, this is the difference between an agent that can be audited and one that creates credential sprawl.
Developer Environments Become Security Control Points
Red Hat Desktop reaching general availability matters more for what it represents about the developer endpoint than for the tool alone. Local container development now ships with isolated agent sandboxing, the supported Red Hat build of Podman Desktop, and direct access to Red Hat Hardened Images. Red Hat Trusted Libraries with SLSA Level 3 origin and integrity, paired with AI-driven exploit intelligence that reasons about whether a vulnerability is reachable in the runtime, push security policy into the laptop image itself. The shift-left posture is now wired into the developer environment.
Where the Gaps Sit
The multi-coding-assistant strategy in OpenShift Dev Spaces, supporting Kiro, Copilot, Claude CLI, Cline, Continue, and Roo, preserves developer choice while expanding the governance surface area. Enterprises will need clear policies on which assistants touch which repositories and what data crosses the perimeter. The skills repository depends on partner and customer contribution velocity to deliver on its promise, and remains a chicken-and-egg problem until the catalog reaches critical mass.
What to Watch:
- Watch whether the “everyone is a builder” operating model travels outside Red Hat. The internal proof point is compelling, and the question is whether enterprise customers can replicate the cross-functional code contribution pattern without a Red Hat-shaped culture underneath it.
- Watch the token economics shift from consumer to provider in enterprise procurement conversations. CFO scrutiny of agent token spend will reach board level over the next two quarters, and Red Hat AI Gateway’s governed Model-as-a-Service is timed for that conversation.
- Watch how the Ansible automation orchestrator performs against ServiceNow Now Assist and IBM watsonx Orchestrate as the agentic execution layer for IT operations. Red Hat’s wager on deterministic execution under AI direction is a sharp competitive position, and proof will come from production deployments.
For more information, visit Red Hat’s newsroom on the company website.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other Insights From Futurum:
AI Platforms Market Hits $109.9B, More Than Tripling by 2030
Atlassian Teamwork Graph: The Secret Weapon That’s No Longer a Secret
Can IREN Build a Full-Stack AI Cloud? The Mirantis Acquisition Makes the Case
The New Rules of Digital Sovereignty: Architecture, Control, and Competitive Advantage
AWS Pushes the Agent Stack: Quick, Connect Verticals, OpenAI on Amazon Bedrock
Author Information
Mitch Ashley is VP and Practice Lead of Software Lifecycle Engineering for The Futurum Group. Mitch has over 30+ years of experience as an entrepreneur, industry analyst, product development, and IT leader, with expertise in software engineering, cybersecurity, DevOps, DevSecOps, cloud, and AI. As an entrepreneur, CTO, CIO, and head of engineering, Mitch led the creation of award-winning cybersecurity products utilized in the private and public sectors, including the U.S. Department of Defense and all military branches. Mitch also led managed PKI services for broadband, Wi-Fi, IoT, energy management and 5G industries, product certification test labs, an online SaaS (93m transactions annually), and the development of video-on-demand and Internet cable services, and a national broadband network.
Mitch shares his experiences as an analyst, keynote and conference speaker, panelist, host, moderator, and expert interviewer discussing CIO/CTO leadership, product and software development, DevOps, DevSecOps, containerization, container orchestration, AI/ML/GenAI, platform engineering, SRE, and cybersecurity. He publishes his research on futurumgroup.com and TechstrongResearch.com/resources. He hosts multiple award-winning video and podcast series, including DevOps Unbound, CISO Talk, and Techstrong Gang.
Nick Patience is VP and Practice Lead for AI Platforms at The Futurum Group. Nick is a thought leader on AI development, deployment, and adoption - an area he has researched for 25 years. Before Futurum, Nick was a Managing Analyst with S&P Global Market Intelligence, responsible for 451 Research’s coverage of Data, AI, Analytics, Information Security, and Risk. Nick became part of S&P Global through its 2019 acquisition of 451 Research, a pioneering analyst firm that Nick co-founded in 1999. He is a sought-after speaker and advisor, known for his expertise in the drivers of AI adoption, industry use cases, and the infrastructure behind its development and deployment. Nick also spent three years as a product marketing lead at Recommind (now part of OpenText), a machine learning-driven eDiscovery software company. Nick is based in London.
