Volume 343, Issue 3 · 15 articles
Red Hat CTO for AI Brian Stevens reflects on vLLM and llm-d reaching dependable, production-grade open source AI serving.
Red Hat's Brian Stevens, SVP and CTO for AI, recalls that when open weight models first emerged, the industry lacked an open inference serving system that could run them efficiently.
His team redirected its energy to the vLLM project, which grew with the community into the de facto inference server and is now part of the PyTorch Foundation.
He describes how Red Hat continued that work toward distributed, production-scale serving and argues open source AI serving has now crossed the threshold to dependability.
Red Hat Enterprise Linux is now generally available as a native offering in the Oracle Cloud Infrastructure Console.
Red Hat says organizations running mission-critical workloads want OCI's performance and economics without sacrificing the stability, security and multicloud consistency of enterprise operations.
Red Hat Enterprise Linux is now generally available as a native offering in the OCI Console, which Red Hat calls a significant milestone in its two-decade collaboration with Oracle.
The post explains how the offering supports hybrid cloud consistency and simplifies migration of RHEL workloads to OCI.
Red Hat reviews Linux Foundation research on open source development in the generative AI era and what it means for AI.
Red Hat says the technologies that shape computing do not have to be built behind closed doors, and that AI has the chance to follow the open source path of operating systems and cloud infrastructure.
The post discusses the Linux Foundation report Navigating Open Source Software Development in the Generative AI Era, sponsored in part by Red Hat.
It highlights the report's evidence on both the opportunities and challenges of open source AI and what an open foundation for AI should look like.
Red Hat's weekly Friday Five roundup covers its AI Factory with NVIDIA update, Project Lightwell coverage and other news.
Red Hat's Friday Five roundup for October 23, 2026 collects the week's notable stories.
It highlights the expanded Red Hat AI Factory with NVIDIA, spanning agent security through OpenShell, optimized open weights inference and Day 0 enablement for NVIDIA Vera Rubin infrastructure.
It also points to BizTech Magazine coverage of how IBM and Red Hat's Project Lightwell uses AI and a trusted clearinghouse to validate and deliver open source security fixes ahead of public disclosure, along with other items from the week.
Red Hat demos how an integrated DevSecOps platform on OpenShift reduces tool sprawl and speeds application delivery.
Red Hat says platform engineers often assemble toolchains for CI/CD, monitoring, security scanning, GitOps and networking, then spend more time maintaining integrations than helping developers ship code.
It describes the real cost of tool sprawl as not just licensing fees but extra vendor relationships, constant integration work and fragmented dashboards that turn quick fixes into scavenger hunts.
The five-minute demo shows how DevSecOps capabilities built into Red Hat OpenShift offer a better starting point for faster application delivery.
Red Hat introduces RHEL for NVIDIA 26.03, a partner-optimized RHEL edition for the latest NVIDIA hardware including Vera Rubin.
Red Hat says new NVIDIA GPUs usually arrive before a standard enterprise Linux release can carry every needed driver, kernel change and virtualization patch, forcing teams to wait or run on an OS they do not plan to keep.
Red Hat Enterprise Linux for NVIDIA 26.03 is a partner-optimized edition of RHEL for the latest NVIDIA hardware, included with a RHEL for ARM 64 subscription.
Red Hat positions it as the way to start on enterprise Linux now and as the edition ready to support the NVIDIA Vera Rubin platform as part of Red Hat AI Factory with NVIDIA.
Red Hat offers IT managers guidance on building an automation-driven culture as AI makes manual, repetitive work obsolete.
Red Hat argues that relying on manual work is no longer viable in modern IT engineering organizations, and that AI has added new complexity and urgency by making repetitive execution obsolete.
It notes a persistent paradox: automation and AI tools are more powerful and accessible than ever, yet adoption often stalls.
The post says the bottleneck in becoming an automation-driven organization is rarely technology but people and culture, and offers guidance for IT managers leading that change.
Red Hat OpenShift Virtualization 4.22 is now an NVIDIA-Certified Hypervisor for NVIDIA HGX systems.
Red Hat says OpenShift Virtualization 4.22 has achieved NVIDIA-Certified Hypervisor status for NVIDIA HGX systems, validating near bare-metal GPU performance for representative AI and accelerated computing workloads.
The company says customers no longer have to choose between the flexibility of VMs and the scale of Kubernetes for AI pipelines.
GPU-accelerated virtual machines and containerized AI workloads can run on the same cluster, under the same API and managed by the same team, with the certification confirming accurate exposure of hardware topology.
BMW Group migrated its HPC environment to Red Hat Ansible Automation Platform, simplifying management of 15,000 servers.
Red Hat says BMW Group has migrated its High Performance Computing environment to Red Hat Ansible Automation Platform.
Standardizing its automation strategy let BMW modernize a decades-old production system, simplify management of 15,000 servers and speed IT service delivery for 30,000 internal users.
The HPC environment is critical to vehicle development, supporting performance-intensive applications such as 3D modeling, crash test simulations and new car design.
BMW says close collaboration with Red Hat and partners delivered the complex migration on time and within budget.
The llm-d project, founded by Red Hat and others and now under CNCF, reaches 1.0 for scalable, cost-efficient LLM inference.
Red Hat says llm-d project contributors have released llm-d 1.0, a major milestone for the open source project designed to help organizations deploy and scale large language model inference efficiently across modern infrastructure.
Founded in 2025 by Red Hat, IBM, Google Cloud, CoreWeave and NVIDIA and now governed by the Cloud Native Computing Foundation, llm-d is part of the open foundation for enterprises running AI in production.
Red Hat says that as organizations move to production, the bottleneck shifts to the economics, reliability and infrastructure demands of agentic workloads, which llm-d 1.0 aims to address.
Red Hat Enterprise Linux becomes a first-party offering in the OCI Console, with more Red Hat subscriptions on Oracle Marketplace.
Red Hat announced expanded availability of its open hybrid cloud portfolio on Oracle Cloud Infrastructure.
Red Hat Enterprise Linux is now a first-party offering in the OCI Console, and subscriptions for Red Hat Ansible Automation Platform, RHEL, Red Hat OpenShift and OpenShift Virtualization are available through Oracle Marketplace.
Red Hat says this helps organizations optimize existing cloud investments and build a consistent foundation for next-generation workloads, including enterprise AI workloads on RHEL for OCI.
Red Hat expands Red Hat AI Factory with NVIDIA with agent security, optimized model deployment and deep hardware enablement.
At NVIDIA GTC Berlin, Red Hat announced expanded updates to its collaboration with NVIDIA to accelerate security-focused, enterprise-grade agentic AI across the open hybrid cloud using Red Hat AI Factory with NVIDIA.
The updates combine agent security capabilities, optimized model deployment architectures and deep hardware enablement.
Red Hat says agent boundaries should be enforced outside the agent, in the runtime and infrastructure, and that the combination of a secure agent runtime, governed inference and hardware support gives enterprises a trusted blueprint for agentic AI.
Red Hat outlines five enterprise AI agent use cases already running in production on Kubernetes clusters.
Red Hat notes that AI agents have moved from experiments into business operations, citing a KPMG Q1 2026 AI Pulse finding that 54% of organizations now deploy agents across core systems.
It says the concept of agents as language models that can take actions is widely understood, but the concrete actions that save enterprises time and money are less tangible.
The post describes five enterprise use cases for AI agents with reasonably predictable results, all drawn from production deployments running on Kubernetes clusters.
Red Hat explains how organizations gain more value from open source by participating in the communities behind the software they use.
Red Hat says many organizations first encounter open source by adopting a library, operating system, database or application, but robust engagement goes much further than using software.
The post argues organizations should understand how to participate in the communities behind the software they depend on.
That participation gives them a voice in the direction of key technologies, builds relationships with maintainers and lets them share solutions to common problems.
Red Hat stresses that open source is a development and operating model, not a business model.
Red Hat explains how NVIDIA BlueField and DOCA Argus on OpenShift help keep autonomous AI agents within bounds and away from attackers.
Red Hat says autonomous agents are powerful because they act, but ambiguous instructions, unexpected tool behavior or complex tasks force them to make judgment calls, creating a new category of infrastructure risk.
As enterprises scale agentic AI, two failure modes emerge: agents that overstep their operational bounds and agents weaponized by external attackers.
The post describes how the NVIDIA OpenShell Reference Architecture, NVIDIA BlueField and DOCA Argus running with Red Hat OpenShift address both failure modes in the enterprise AI factory.