NVIDIA Unveils Open Nemotron Telco Model and Agentic Blueprints for Autonomous Network Operations

At Mobile World Congress Barcelona 2026, NVIDIA introduced the Nemotron-based Large Telco Model (LTM) and a series of Agentic AI Blueprints. These tools are designed to transform telecommunications from simple automated workflows into fully autonomous networks capable of complex reasoning. This initiative is part of the GSMA Open Telco AI program, aiming to standardize intelligent network management globally. The new 30-billion-parameter Nemotron LTM was developed in collaboration with AdaptKey AI. It is specifically fine-tuned on telecom industry standards and synthetic logs to understand domain-specific language. By offering this as an open model, NVIDIA allows operators to deploy reasoning capabilities on-premises while maintaining full control over their proprietary data.

Beyond the foundation model, NVIDIA partnered with Tech Mahindra to release an open-source guide for building reasoning agents. This framework teaches AI to mimic network engineers by capturing thinking examples or reasoning traces. These traces help agents understand the logic behind fault isolation and remediation planning rather than just following scripts.

A primary application of this technology is the new Blueprint for intent-driven RAN energy efficiency. By integrating with VIAVI Solutions simulation tools, the system allows agents to test power-saving policies in a virtual environment. This closed-loop approach ensures that energy reductions do not negatively impact subscriber quality of service. Global operators are already implementing these blueprints to solve regional challenges. Cassava Technologies is deploying an agentic platform to optimize diverse mobile environments across Africa, while NTT DATA in Japan uses AI to manage network surges during service restoration. These deployments demonstrate the practical scalability of agentic orchestration in high-stakes environments.