Telecom operators weigh distributed AI infrastructure as revenue case remains uncertain
Nvidia and Nokia are encouraging telecom operators to use network sites and infrastructure for distributed AI workloads, though the business case is still unclear. Several EMEA operators are building or expanding Nvidia-based infrastructure, and Deutsche Telekom is extending an industrial AI cloud for customers including a robotics firm. AT&T's results are presented as evidence that connectivity remains a solid revenue source amid the AI hype.
At GTC Berlin, Nvidia and Nokia showcased AI-RAN trials involving T-Mobile, BT, Deutsche Telekom, Elisa, Orange and others. Nine operators across EMEA are adding or enlarging Nvidia-based systems; CRA, Etisilat, Silknet and TIM are installing new capacity, while Deutsche Telekom, Fastweb/Vodafone, Kazakhtelecom and Orion are extending existing setups.
Deutsche Telekom’s sovereign industrial AI cloud will host open models such as Nemotron and Soofi for customers including Agile Robots. A demonstration splits processing between robot-edge and network-edge for real-time perception and fleet coordination, while larger training remains centralized. AT&T, meanwhile, reported four-year-high service revenue growth and over one million advanced connectivity additions for a second quarter.
If telcos convert cell sites and network assets into distributed AI capacity, enterprises using robots, vehicles, and machines could gain lower-latency processing and new managed services. Operators may find fresh revenue beyond connectivity, while customers and communities could see changes in infrastructure investment and service pricing. The impact remains uncertain because capacity, investment, and demand details are scarce; AT&T’s results suggest conventional connectivity may stay the more reliable near-term social and economic benefit.