• Nokia named A1 Group, Chunghwa Telecom, du, e&, Mobily, stc, TPG Telecom, and Zain Saudi among operators evaluating AI-RAN on NVIDIA platforms
  • The work ranges from proofs of concept to live field trials, with no commercial deployment announced for the eight operators

The fact

Nokia said on 16 September that eight operators are advancing AI-RAN evaluations using NVIDIA Aerial RAN Computer: A1 Group, Chunghwa Telecom, du, e&, Mobily, stc, TPG Telecom and Zain Saudi. The projects range from early assessments and proofs of concept to live field trials. Nokia said its wider AI-RAN engagement now extends across North America, Europe, Asia-Pacific and the Middle East.

AI-RAN combines radio-access-network processing and AI workloads on accelerated computing infrastructure. Nokia is positioning the architecture as a way for operators to add AI computing while supporting mobile-network functions on the same broader platform. The announcement does not say that any of the eight operators has committed to a commercial rollout, placed a fleet-scale order or disclosed a deployment timetable.

The assessment

AI-RAN only becomes commercially meaningful if operators can share accelerated hardware between radio processing and AI workloads without allowing one workload to disrupt the other. A lab proof can show that the stack runs, while a field trial can show that it handles live traffic. Neither establishes whether the architecture is cheaper, more resilient or easier to operate than conventional RAN over a full load cycle.

The eight named operators therefore show the breadth of evaluation rather than the scale of deployment. Trials need to test radio performance while AI demand changes, including what happens when computing resources are constrained or hardware fails. For BTW readers, the next step is an operator committing AI-RAN to a commercial network with performance, cost and operating results that can be compared with existing baseband infrastructure.

What to watch

Watch for a named operator moving from trial to commercial deployment or including AI-RAN in a network tender or capital plan. Published results on radio latency, availability, power use and accelerator utilisation under mixed RAN and AI workloads would show whether shared hardware delivers an operating advantage rather than simply technical compatibility.