Samsung outlines how operators can make existing software infrastructures AI-ready, minimising the need for new hardware and lowering costs.
Operators that adopted virtualisation early are already reaping operational savings and possess infrastructure capable of handling AI workloads today without a wholesale redesign. By layering automation and observability on top of virtualised platforms, networks can feed intelligence back into operations and evolve into proactive, self-optimising systems.
Ji-Yun Seol, Executive VP and Head of Product Strategy, Networks Business at Samsung, explains: “The transition from 5G to 5G-Advanced and ultimately 6G rests on three interconnected pillars: virtualisation to enable flexible networks, AI integration across network layers, and automation to move toward autonomous networks.”
Decoupling functions from hardware
Traditional telecoms relied on network functions that were tightly bound to dedicated hardware: routers, baseband units and security appliances were distinct physical devices. While that model delivered consistent performance, adding new capabilities typically demanded more hardware, additional rack space and higher power consumption.
Virtualisation shifts this model by separating network functions from specialised equipment and running them as software on commercial off‑the‑shelf (COTS) servers. Moving to general‑purpose compute platforms delivers cost savings, greater flexibility and readiness for AI workloads.
The industry introduced this architecture in phases. Core network virtualisation came first, enabling packet gateways and subscriber databases to operate on standard servers. Virtualised RAN (vRAN) followed, decoupling baseband processing from radio units so those functions could run as cloud‑native software.
How Samsung is consolidating infrastructure for AI capabilities
Samsung has advanced this approach through vRAN software and modern processors, using cloud‑native designs that allow multiple network functions to run on a single server. Core components, RAN processing, virtualised switch and cell site routers (vCSR), and virtualised security can coexist on shared compute resources.
Consolidating functions onto fewer servers reduces physical footprint, lowers capital and operational expenses, and decreases energy consumption. It also simplifies management by turning a cluster of specialised appliances into a cohesive, software-driven platform.
Historically, transport and cell‑site systems depended on dedicated devices, constraining options and making upgrades expensive. Disaggregating hardware and software enables operators to move these functions onto whitebox switches and commodity routers of their choice, increasing vendor flexibility and reducing lock‑in.
A virtualised cell site router provides a concrete example: containerised and software‑based routing can be co‑located with a virtualised Distributed Unit on the same server. This turns the cell site from a collection of fixed hardware into a dynamic, software‑defined node, delivering total cost of ownership savings and operational agility.
Unified data for AI-ready networks
A fully software‑driven, end‑to‑end network creates the data visibility and control that AI needs. Where legacy networks lock data into hardware silos, virtualised environments provide unified, high‑fidelity data streams that AI systems can use for orchestration, performance improvement and infrastructure optimisation.
This unified data foundation enables operators to adopt AI‑driven services more rapidly and explore new monetisation opportunities. Advances in processor performance—across CPUs, GPUs and specialised accelerators—mean core compute resources can host AI workloads directly on existing telco infrastructure.
Samsung advises that operators can integrate these processors into software‑driven networks to expand AI capabilities without waiting for lengthy hardware procurement cycles. In doing so, networks can evolve at the pace of innovation rather than being constrained by legacy deployment timelines.
Looking ahead, the progression from 5G to 6G will depend on tightly integrating virtualisation, AI and automation so networks become more flexible, intelligent and autonomous.
See also: Ericsson, FET, and OPPO validate performance-based 5G slicing
Interested in learning about the IoT from industry leaders? The IoT Tech Expo runs in Amsterdam, California and London as part of TechEx, co‑located with events such as the AI & Big Data Expo and Cyber Security Expo. The event brings together industry experts to discuss trends, best practices and real‑world deployments in IoT and adjacent technologies.
Telecoms content is provided by TechForge Media. Explore additional enterprise technology events and webinars through TechForge’s event listings.