For CSPs, AI delivers operational efficiencies and new monetization opportunities
Artificial intelligence (AI) is reshaping the telecommunications industry, offering communications service providers (CSPs) a pathway to both operational efficiency and new revenue streams. At Mobile World Congress, the conversation around AI in telecom was particularly vibrant, with a key theme emerging: AI’s impact extends far beyond the radio access network (RAN). According to Fujitsu Networks Business Head of Global Marketing Rich Colter, CSPs need to take an end-to-end approach to AI adoption—one that includes not only the RAN but also optical networks and AIOps methodologies.
“AI is critical for the network,” Colter said. “There’s opportunities in how you monetize the network, and also how we optimize and transform the network…Open networking is just the foundation…We’re seeing that with O-RAN even into the optical network with open line systems and Open ROADM,” an initiative to disaggregate optical network components that route data in a manner that makes network management more efficient.
Focusing on how AI is affecting the RAN, Colter broke it down as AI and RAN, AI for RAN, and AI on RAN.
– AI and RAN refers to a shared infrastructure where both AI workloads and RAN workloads run on the same hardware. “From that, you can both manage your infrastructure and monetize spare capacity,” Colter explained.
– AI for RAN focuses on using AI to enhance RAN performance. Fujitsu’s customers, he noted, are seeing performance gains of 20% to 50% across various AI-driven optimization use cases.
– AI on RAN involves leveraging the radio network to support generative AI applications, ensuring that CSPs can meet evolving end-user demands.
Regarding AI and RAN, during Mobile World Congress, Fujitsu and partners Arrcus, Eviden, Liberty Global, NVIDIA and Philips demonstrated how a range of advanced consumer applications can run on an AI-based Open RAN network alongside RAN workloads. On the AI for RAN front, Fujitsu is working with SoftBank and NVIDIA on verification testing of use cases, including uplink channel interpolation, which uses AI to boost RAN performance. Fujitsu worked with SoftBank to design and develop the AI, and Fujitsu developed the embedded interface.
Another key point Colter raised is that beyond simply buying AI solutions, CSPs need to make the appropriate organizational and workflow changes needed to successfully leverage technology investments. This gets into AIOps, “another critical area where you’re using AI and AI applications to optimize the network throughout its lifecycle,” he said. From initial deployment of network equipment to day-to-day troubleshooting, AIOps is all about doing things “in a more optimal way.”
He gave the example of using AI to accelerate and automate root cause analysis. When network operations center engineers see a surge, AI can be used to isolate the problem and quickly identify the underlying problem, thereby making resolution faster. Another area where AIOps can make a major difference, Colter said, is in retiring legacy network assets. CSPs “need to find a way to get that equipment out of the network, turn off the power that’s going to it, upgrade to new solutions that have better security and performance.”
To his last point, Fujitsu’s AI-driven network modernization (NetMod) solutions, a suite of 13 AI engineering tools, can reduce manual effort by 80%, reduce required task hours by 75%, and reduce delivery time by 67%. Specific examples include migrating like to like, migrating legacy to next-gen equipment, migrations that require circuit redesigns and network core flattening.
As Colter made clear, AI is no longer an isolated tool in telecom—it’s a fundamental enabler of efficiency and new business models. For CSPs, future-proofing the network means adopting an end-to-end AI strategy that delivers value across the entire network stack.”
