LF Broadband presented at the BBF BASe Summit in Milan on June 4-5, 2026, bringing the arguments from the recently published white paper “Open, Observable, Intelligent: Building the Foundation for AI in Fixed Broadband Access” to an audience of operators, vendors, and standards community members. The session made a case that has become central to LF Broadband’s 2026 work: the practical path to AI in broadband networks runs through architecture, not models.
The session featured Francisco De Carvalho of Radisys, LF Broadband Governing Board, and Mahir Gunyel, Director of Technology and Innovation at Netsia.
The Core Argument: Foundation Before AI
Mahir Gunyel, Director of Technology and Innovation at Netsia, anchored the technical presentation. His central argument was direct: AI in fixed broadband does not start with selecting a model. It starts with whether the network is open enough to expose meaningful telemetry, observable enough for operators to trust the data coming out of it, and programmable enough to support bounded automated actions through deterministic APIs with proper rollback and audit capability.
The problem Gunyel described is one most operators recognize. Fixed broadband networks are complex, multi-vendor environments running a mix of GPON, XGS-PON, and other access technologies alongside in-home Wi-Fi systems, OSS/BSS stacks, and multiple management frameworks that were not designed to interoperate cleanly. The result is fragmentation; too many signals, too many tools, and data that is difficult to correlate across domains. The gap between the data operators collect and what their teams can interpret and act on quickly enough is the operational problem AI is being asked to solve. But AI systems face the same fragmentation that human operators do. They cannot generate consistency from inconsistent inputs.
This is where the VOLTHA and SEBA architecture becomes relevant to the AI conversation. VOLTHA provides a vendor-agnostic abstraction layer for PON access equipment, normalizing the southbound interface across hardware from different manufacturers. The SEBA framework sits above it, exposing normalized APIs for policy, topology, telemetry, and control. The practical effect is that AI systems interacting with the access network through this stack encounter clean, reusable, technology-agnostic interfaces rather than a different integration problem for every vendor environment. Use cases like predictive fault detection, service assurance scoring, and network co-pilots all depend on reliable telemetry and control context; and that reliability has to be built into the infrastructure before the AI layer is added.
Gunyel also highlighted VOLTHA’s role in supporting open hardware at the edge. As AI workloads move from central data centers toward access network edges and onto OLT devices themselves, open hardware interfaces become an AI readiness issue, not just a network architecture preference.
The Role of Open Source
The presentation gave specific attention to why open source matters for AI in broadband beyond the general arguments for transparency and flexibility. Vendor neutrality is particularly important here because analytics and automation need to be reusable. If every vendor environment requires a bespoke AI integration, the operational benefits of AI will not scale across a heterogeneous production network. Open source enables operators, vendors, and implementers to turn shared needs into working, validated code; and to do so in realistic deployment conditions rather than controlled test environments. That validation process is what builds the trust that AI-assisted operations require before operators are willing to extend autonomy to automated systems.
Open source also supports the auditability that AI in operations demands. As AI moves closer to decisions that affect network state, operators need to inspect system behavior, validate implementations, and trace how inputs translate into recommended or automated actions. Open platforms make that possible in a way that opaque, proprietary AI operations stacks do not.
The Standards Connection
A recurring theme in the session was the relationship between standards bodies and open source communities. Gunyel described the two as complementary rather than parallel: standards bodies like the Broadband Forum define requirements, architecture, interfaces, and data models; open source communities build deployable implementations of those requirements. The result is a faster path from concept to production, lower integration risk for operators, and AI systems that can operate consistently across vendors because they are working from the same underlying data models. LF Broadband’s collaboration with the Broadband Forum, where BBF defines requirements and LF Broadband ships production-grade code that implements them, is a direct expression of this model.
A Practical Path to Autonomy
The session closed with a maturity model for AI adoption in broadband operations that deliberately avoided overstating where the industry currently stands. The path runs from manual operations through AI-assisted analysis and insight generation, to recommended remediation workflows that operators approve, to policy-bounded automation with strong guard rails, and eventually toward higher automation with deterministic interfaces, rollback capability, and audit trails. The human role at advanced stages is not eliminated; it shifts from manually interpreting every signal to supervising, approving, and improving bounded automation. The argument is that this is how broadband AI earns operator confidence: incrementally, with each step demonstrating value before the next level of autonomy is granted.
The Q&A that followed the presentation extended the discussion into how VOLTHA and SEBA need to evolve to support AI workloads: deeper observability, tighter integration with northbound orchestration systems, and continued alignment with standards bodies to ensure the interfaces AI systems need are clean, open, and reusable.
Recordings and Resources
The White Paper
The full white paper presented at the BBF BASe Summit is available for download at lfbroadband.org. It covers the technical architecture requirements for AI-ready access networks, the near-term use cases with the strongest operator value, and LF Broadband’s role in building the open foundation that makes those use cases deployable in production.
What Is Next
The Q&A that followed the presentation extended the discussion into how VOLTHA and SEBA need to evolve to support AI workloads: deeper observability, tighter integration with northbound orchestration systems, and continued alignment with standards bodies to ensure the interfaces AI systems need are clean, open, and reusable.
About the Project
LF Broadband supports a collection of projects that transformed broadband networks and the Passive Optical Network (PON) industry, including the SEBA reference design for building open broadband networks, and the VOLTHA open source project for virtualizing multi-vendor PON systems. These projects are in deployment with Deutsche Telekom, Turk Telekom, and more.
About The Linux Foundation
The Linux Foundation provides a neutral, trusted hub for developers to code, manage, and scale open technology projects.
Last updated June 29, 2026
