BROADBAND FORUM AI OPTIMISATION OF NETWORKS
STANDARDISING THE AI OPTIMISATION OF BROADBAND NETWORKS While some people may be tired of hearing the term “Artificial Intelligence (AI)”, its significant impact on human society is undeniable. This is certainly true in terms of its effect on broadband infrastructure, with AI fundamentally transforming how entire networks behave and the outcomes they are expected to deliver, writes Broadband Forum Service Requirements Work Area Directors Manuel Paul and Tony Zeng .
As of 2025, 97% of service providers identified AI as a key tool for analysing customer usage patterns. At the same time, 33% of telecommunications executives confirmed AI was already being deployed within their broadband services, with a further 56% expected to use it within 12 months. Such widespread adoption signals a clear turning point within the industry, highlighting the technology as a defining factor in how broadband networks will evolve. THE CORE VALUES OF AI AI continues to transform both the volume and the character of broadband traffic. Generative and interactive AI applications increase upstream and downstream data flows, and depending on the use case, they can place greater demands on network capacity, latency, availability and performance consistency. Concurrently, AI is becoming an enabler inside the network itself, driving higher-quality service delivery and more efficient operations. As these shifts redefine what the underlying infrastructure is being asked to support, the industry is being forced to rethink where true differentiation now comes from. While AI continues to evolve the online experience, the internet of the future will go much further. For example, AI agents will permit natural and comprehensive human-computer
dynamic interactions, while supporting self-optimising networks will fix their own faults, deliver predictive maintenance, automatically move capacity to where it is required, and make crucial energy savings across the network. In fact, three fundamental areas in which AI can create value have already begun to emerge. This includes AI agents as new services, guaranteed Quality on Demand, and self-optimising networks, which are addressed by standards development organisation Broadband Forum’s recently published vision for AI in Broadband Networks (MR-529). AI AGENTS AS NEW SERVICES AI agents now represent a fundamental shift in how broadband networks deliver value, enabling the transition from offering simple connectivity to intelligent, more interactive service layers. These agents hold the potential to enable natural and comprehensive human-computer dynamic interactions, ensuring networks can support richer, multimodal exchanges instead of solely transferring data. This evolution pushes service providers to think of AI agents as core service offerings that sit alongside – and optimise – traditional broadband packages. To best support these AI-driven experiences, networks are expected to deliver significantly higher throughput,
lower latency and greater stability. This can be achieved if computing resources can be integrated into existing telecom infrastructure through coordinated edge-to-cloud architecture. As a result, AI workloads can be distributed effectively across customer premises, access networks, central office cloud, gateways, and edge computing nodes, rather than existing exclusively within the cloud. Consequently, service providers are empowered to deliver more responsive, resilient and personalised service environments. As such, new business models can be accessed. For example, AI agents can act as customer-facing assistants or in-premises controllers, which are monetizable as premium services. Connection-oriented models will then become ones in which AI-enabled services become a true differentiator, supporting users through bandwidth upgrades and deeper engagement. GUARANTEED QUALITY ON DEMAND Of course, AI traffic means rising network complexity, with elements like training, inference and AI generated extended reality (XR) workloads all requiring consistent performance. To overcome these issues, Quality on Demand (QoD) has also emerged as a critical value for AI usage. Traditional best-effort broadband cannot support these growing demands,
22
www.opticalconnectionsnews.com
| ISSUE 44 | Q3 2026
Made with FlippingBook Annual report maker