Red Hat, Nvidia, Palo Alto forge AI-native telco stack

Red Hat, Nvidia and Palo Alto Networks have formed a telecoms-focused partnership that combines cloud-native software, accelerated computing infrastructure and network security in a single architecture for AI-driven network operations.

The effort targets service providers looking to embed AI into day-to-day network management across central data centres, edge sites and other network locations. The partners are positioning a common foundation where AI workloads and telecoms network functions run on the same platform, with consistent security policies across environments.

Platform Stack

The collaboration centres on Red Hat OpenShift and Red Hat OpenShift AI as the base platform. OpenShift provides the Kubernetes layer for deploying and managing workloads, while OpenShift AI supports AI development and operations.

Under this approach, telecoms network functions and AI workloads run across core and edge locations under the same operational model, using the governance and management processes service providers already apply to cloud-native infrastructure. The partners argue that a single platform reduces the need for separate tooling and isolated environments for AI projects.

Nvidia provides the compute layer for model training and inference. The partners pointed to Nvidia RTX PRO Servers for data centre and edge deployments, and the Nvidia Aerial RAN Computer family for radio access network and network-edge use cases.

They linked these systems to AI-driven tasks such as real-time inference and operational analytics, and to AI-native network services that operators may deploy closer to users at the edge.

Security Model

Palo Alto Networks is contributing security products that run on the Red Hat platform and integrate with Nvidia infrastructure components. The partners highlighted Prisma AIRS and Nvidia BlueField as key building blocks, along with Nvidia DOCA steering and Nvidia ConnectX for traffic handling and enforcement.

The design applies security controls at the infrastructure layer rather than only at the application tier, enabling consistent enforcement across core and edge deployments. The partners also said it can avoid performance impacts and added latency when security tools sit outside the main data path.

They described the result as a shift from loosely connected components to a cohesive platform that combines AI lifecycle management, AI infrastructure and security operations in a single stack.

Telco Use Cases

Telecoms operators have invested in automation for years, with AI now taking a more central role in network operations. The partners said AI is moving beyond the application layer and becoming part of how networks function, driven by rising complexity and the need for faster service delivery.

Examples include AI-driven RAN optimisation, predictive maintenance, energy efficiency and automated security enforcement. These use cases often rely on distributed data sources and require inference close to where traffic is generated, which is why the partners emphasised edge deployments.

They also framed AI-native telecoms as a requirement spanning the core network, edge environments and the wider footprint. That implies standardisation across infrastructure types and locations, which they said their combined approach addresses.

Industry Context

Telecoms operators face pressure to manage costs while maintaining service quality and preparing for new connectivity demands. AI is increasingly discussed as a tool for operational efficiency and fault management, while security remains a persistent concern as networks become more software-driven and distributed.

The partnership comes as the industry assesses infrastructure needs for future wireless standards. The partners linked their work to the longer-term path toward 6G-era networks, where increased automation and new edge services are expected to raise compute and security requirements.

They added that service providers are at different stages of adoption and that the move to AI-native operations is incremental rather than a single deployment event.

"By pairing Prisma AIRS with OpenShift and NVIDIA BlueField, we enable a powerful synergy of centralized and distributed security enforcement-leveraging NVIDIA DOCA steering for hardware-level precision-specifically optimized for the demands of AI-driven telecom workloads."

The partners said service providers can engage with them on the architecture at Mobile World Congress, with discussions expected to focus on operational transformation and service innovation using distributed AI workloads.