The Linux Foundation’s networking division (LF Networking) is continuing to grow both its mandate and project roster in 2025.
At the Open Networking & Edge Summit in London, which is co-located with the Kubecon conference, LF Networking detailed an ambitious strategic roadmap that emphasizes the convergence of open source, artificial intelligence, and cloud-native technologies as the foundation for next-generation networking infrastructure.
As part of the roadmap, a pair of open-source networking projects were announced that will bring new AI capabilities. Project Salus is a responsible AI toolkit, while Essedum is an AI framework for networking applications. LF Networking also announced the CAMARA Spring25 Meta-Release advancing the open-source telecom-focused platform.
Alongside the technical components, LF Networking released its global survey on the state of open networking that reveals a series of key trends. Top findings from LF Networking’s global survey include:
- Open-source criticality: 92% of organizations view open-source projects as critical to their future.
- Cloud-native adoption: 73% of organizations are already integrating cloud-native networking into their workloads.
- AI in networking: 74% see open source as foundational to AI success in networks.
- Top AI applications: Network automation leads at 57%, followed by security at 50% and predictive maintenance at 41%.
- Implementation barriers: 38% cite skills gaps as the primary barrier to open-source adoption, followed by security concerns at 37%.
Cloud-native networking adoption is mainstream, but challenges remain
Adoption of cloud-native technologies for networking is evidently mainstream, according to the LF Networking survey. That said, the same survey also identified some clear challenges.
“Adopters indicated architectural complexity, Integration with legacy systems, and skills gap as the three main challenges,” Arpit Joshipura, general manager of networking, edge and IoT at the Linux Foundation, told Network World. “For resolving complexity, LF Networking is creating super blueprints that serve as reference implementations for cloud-native network services.”
Beyond blueprints that provide best practices and deployment guidance, there are specific innovations inside of a series of LF Networking projects. Joshipura noted that projects like Nephio simplify the deployment of cloud-native network functions with a declarative approach to service description and intent-based automation that converts operators’ desired state of the network to actual configuration tasks.
Additionally the Cloud-Native Telecom Initiative (CNTi) project creates definitions for best practices for developing and deploying cloud-native network functions (CNFs) and test frameworks that validate the proper use of the best practices.
“This helps operators staff successfully execute complex, cloud-native service deployment tasks, even if they did not initially have deep expertise in this domain and have legacy systems,” Joshipura said. “LF also has a wide variety of e-learning courses for cloud-native and Kubernetes, and we have seen quite a steep response to these for upskilling within the community.”
New AI projects address ethics and network-specific challenges
The two new AI initiatives—Salus and Essedum—represent a strategic push into domain-specific AI for networking, with both projects built on code donated by Infosys.
Joshipura noted that a lot of people in networking organizations are going to be consumers of the same data and models through different AI-enabled applications. To make sure that there is right and responsible use of AI in these applications, organizations need AI guardrail frameworks. That is the key issue Salus is addressing.
“Salus is a framework that brings in AI guardrails on top of the data and models, which ensures enhanced security, data privacy and traceability and prevents sensitive issues like biases,” explained Joshipura. “For networks, this is relevant because it’s becoming clearer that AI for networks needs centralized and uniform data and model strategy.”
The Essedum project specifically targets the unique AI challenges facing network operators. Joshipura said that it is becoming evident that AI for networks requires two tailoring aspects. The first is the need for centralized network data. The second is the need for training and customization of models for network specific use cases.
“While general-purpose AI platforms will bring unified data management, you still need a framework for the users to orchestrate between the data, models and use cases so that the custom use cases for networks can be built,” he said. “The Essedum framework will provide catalogue, work bench, assurance, and access control for building the network AI applications.”
LF Networking’s focus areas for 2025 and beyond
Looking ahead, Joshipura outlined the core areas of focus for LF Networking.
The first area of focus is domain-specific AI with frameworks. “We are witnessing AI serving as a catalyst driving innovation in many of our projects,” he said.
LF Networking projects and communities are already implementing AI to enrich the functionality of projects. Currently a lot of that activity is around translating human intent to network configurations to enable intent-based automation.
“We expect that trend to continue and AI to proliferate into all of the hosted projects, making it easier to consume the technology and making the network operate more efficiently,” he said. “Our community members are already studying the latest trends in AI, such as agentic AI, and are figuring out ways to apply AI agents to performing network automation tasks.”
Joshipura expects the introduction of the new Essedum and Salus projects to serve as an accelerator for adopting emerging AI technologies.
AI isn’t the only thing that LF Networking is focused on. The plan is to continue the group’s strategy of enabling collaboration across different geographies, markets and technologies. Joshipura said that LF Networking will also focus on staying ahead in the technologies that are important to its end users, including Open Source RAN, cloud-native deployments, edge-cloud continuum, and edge AI, and get to AI-ready 6G in open source.
Source:: Network World