Nvidia and Nokia are urging operators to think bigger about networks. Their message is clear. Telcos should turn towers, fiber, offices, and data centers into distributed AI infrastructure.
That pitch gained attention at GTC in Berlin. The companies highlighted AI-RAN trials with T-Mobile, BT, Deutsche Telekom, Elisa, and Orange. These tests show how radio networks could support AI processing closer to users.
The idea reaches beyond smarter base stations. Operators already own valuable assets near enterprises and consumers. They have sites, power, fiber routes, switching points, and regional facilities. These locations could host AI computing for factories, robots, vehicles, and machines.
In simple terms, telcos want to move AI closer to the action. This reduces delays between a device and the computing system. That matters for robotics, industrial automation, safety systems, and vehicle coordination.
Several EMEA operators are now expanding Nvidia-based infrastructure. CRA, Etisalat, Silknet, and TIM are adding new deployments. Deutsche Telekom, Fastweb, Vodafone, Kazakhtelecom, and Orion Telecom are expanding existing platforms.
Still, the commercial model remains unclear. The latest announcements offered few details on capacity, investment, or actual workloads. That makes the market opportunity difficult to measure. Operators need more than strong technology demonstrations.
Deutsche Telekom offers one of the clearer examples. It plans to expand its sovereign “industrial AI” cloud. That platform will host open AI models for enterprise customers. Agile Robots in Germany appears among the early users.
A GTC demo with Agile Robots showed a practical architecture. Robots handled some computing at the device edge. Network-edge systems supported “real-time perception and action” and “fine-tuning and coordination”. Larger training jobs stayed in central cloud infrastructure.
This model could suit many industrial environments. It may help enterprises keep sensitive data closer to home. It may also reduce reliance on distant hyperscale cloud regions. However, many deployments still look experimental.
Meanwhile, AT&T reminded the market where telcos earn today. Its latest results showed strong service revenue growth. The company added more than one million “advanced connectivity” customers again.
Fiber, fixed wireless, and mobile services continue to support its momentum. AT&T also raised its earnings outlook. At the same time, it reduced longer-term capital investment expectations.
This contrast matters for telecom leaders. AI infrastructure may become a valuable new revenue stream. Yet connectivity still pays the bills today. Customers continue buying broadband, mobile, and bundled services.
For now, telcos face a balancing act. They must explore distributed AI without ignoring core services. The winners will likely combine both strategies. They will sell reliable connectivity today, while testing edge AI for tomorrow.
