📍Summary -XCHANGING IDEAS #71 –March 2026
Sasi Kiran Nallapaneni explains that real-time AI systems require continuous, ordered & timely data delivery over networks. Key challenges including network latency, uplink bottlenecks, and the uplink-heavy nature of AI workloads. While 5G is suitable for operational monitoring & predictive analytics, it cannot consistently meet the sub-4ms requirements needed for production relay purposes like IEC 61850 standards.
Hussein Al-Chami on SUSE's Telco Cloud platform -in providing a unified, secure, AI-ready foundation for modern telco networks. Insights from MWC, emphasizing 3 key market trends: AI sovereignty, edge computing, agentic AI. Collaboration with Infosys on an AI-native RAN power optimization aimed at reducing energy consumption at the edge. Industry's current inflection point in 2026, focusing on AI-native telcos & converging cloud & AI operations.
Oswaldo Conti-Bosso on BWS IoT's IoT mobile tracker, their experience with over 3.7 million devices & their recent development of a hybrid modem combining 4G & LoRa. NB-IoT mobile tracker offers extended range up to 55.4 km, better than existing 4G, with ultra-low power consumption of 0.3 milliamperes. Plans to expand into satellite communication & develop a roadmap for terrestrial & satellite integration.
Hammad Hussain on the role of Telcos as potential custodians of Sovereign AI - Six dimensions of sovereignty including data, compute, supply chain, algorithmic, model & economic sovereignty, emphasizing that sovereignty extends beyond just data residency. Telcos could evolve from communication service providers to cognitive service providers by leveraging their existing infrastructure, regulatory expertise & data assets to support sovereign AI workloads.
Jorge Valenzuela Jiménez discusses the critical shift in telecom data management as volumes outpace legacy systems. Emphasizes that the future will be defined by data utilized at unprecedented scale and speed, requiring a move from brittle, siloed architectures like to AI-ready, federated systems.
Shabeer Puthukkudi presented on edge inference & AI, explaining why AI inference is moving from centralized hyperscalers to distributed edge environments, particularly leveraging telecom RAN infrastructure. Key benefits of edge computing including lower latency, data sovereignty, reduced bandwidth costs, while the architectural shift where RAN infrastructure is becoming a compute platform.
Richard Simon presented on sovereign AI in the age of inference, explaining the shift from training to inference in Gen AI and the importance of maintaining control over AI infrastructure & data.
Jose Gil discussed the challenges of deploying GPUs at 6G base stations, emphasizing that AI workloads span a wide compute continuum & require careful consideration of factors like latency, compute requirements, data gravity, operations & economics.
https://lnkd.in/gTND8ypZ