Many telcos still underestimate gen AI’s broader potential, warned Frost & Sullivan Senior Industry Analyst Soumyadeep Roy Chowdhury
In sum – what to know:
Not magic — foundational – Gen AI can dramatically improve telecom operations, from faster network troubleshooting to fraud detection and predictive automation, but only when operators have clean data, unified systems, and strong governance in place.
Bigger than chatbots — but not plug-and-play – Telcos risk both underusing and overhyping gen AI: limiting it to chatbots misses major gains in documentation, design, and scenario simulation, yet expecting it to fix fragmented OSS/BSS or poor-quality data will lead to disappointment.
In a recent conversation with RCR Wireless News, Soumyadeep Roy Chowdhury, senior industry analyst at Frost & Sullivan, outlined how generative AI (gen AI) is reshaping telecom operations. He highlighted the faster-than-human speed with which the technology can detect and resolve network issues, as well as how gen AI is becoming a catalyst for adjacent technologies such as augmented reality and AI-powered fraud detection — ultimately creating intelligence that “extends far beyond simple optimization.”
“This convergence signals a future where telecom operations evolve from reactive manual intervention to proactive, autonomous, and predictive systems driven by AI-enabled collaboration across multiple technology domains,” he said.
The monetization outlook for gen AI is also more optimistic than, say, 5G, which Chowdhury pointed out required heavy infrastructure investment and long, uncertain payback cycles. “Generative AI is different as it delivers rapid, measurable results with far lower entry barriers, primarily driven by software, data, and process integration rather than large capital upgrades,” he explained. Additionally, by directly impacting revenue-focused business areas such as customer experience, marketing, and fraud prevention, gen AI can lead to significant operational savings.
However, he warned that many telcos still underestimate generative AI’s broader potential. “Some telcos and associated businesses are limiting GenAI investments to chatbot applications, whereas it has better value propositions in tasks such as automating documentation, RFPs, technical designs, ticket summarization, and root-cause narratives, and knowledge retrieval,” he said. These capabilities can “significantly reduce cycle times” across engineering, operations, sales, and procurement.
Another major opportunity lies in scenario simulation. According to Chowdhury, gen AI can help employees model service demand, network traffic patterns, churn, tariff impacts, or rollout sequences — offering “what-if perspectives” that support more informed decision-making.
Still, he cautioned against overestimating the technology: It won’t “magically turn things around or completely transform” a telco’s operations, he said. “We have to understand that gen AI, like any other AI solution, is a major transformative technology that acts as an enabler. It does not completely take over the ownership of a complete workstream, but indeed, can eliminate some [manual] processes with automation.”
A key challenge, he added, is that some operators assume gen AI will deliver outsized benefits even in environments with fragmented OSS/BSS, poor data quality, or siloed systems. “But realistically, without data unification, governance, and APIs, gen AI is likely to disappoint,” Chowdhury warned, adding that a completely unsupervised gen AI system can lead to hallucinations. “A fully autonomous network powered by gen AI is not something that will be achieved overnight; it is most definitely not a plug-and-play technology.”
To get there, operators must first establish high-quality data integration, governance, and APIs. Scaling gen AI will then require cloud, hybrid, or edge architectures. “Scalable compute, distributed processing, and automated model management are needed to handle massive real-time data loads, maintain performance at scale, and continuously update and monitor AI models in production,” he said.
