NVIDIA Drives Autonomous Networks with Agentic AI and Open Telco Models

Autonomous networks – intelligent, self-managing telecommunications operations – are rapidly transitioning from a future concept to a current priority for telecom operators. According to the latest NVIDIA State of AI in Telecommunications report, network automation is the leading AI use case for investment and return on investment. However, automation differs from true autonomy. Autonomous networks require the ability to understand operator intent, reason through tradeoffs, and independently decide on the best course of action. Reasoning models and AI agents, specifically those fine-tuned with telecom data, are crucial for enabling this shift.

The Need for Agentic Systems

Achieving fully autonomous networks necessitates an end-to-end agentic system. This system relies on key components, including telco network models and AI agents that can communicate with each other and utilize network simulation tools to validate actions before implementation.

NVIDIA’s Advancements at Mobile World Congress

Ahead of Mobile World Congress Barcelona, NVIDIA unveiled several key advancements to accelerate the path to network autonomy. These include an open, NVIDIA Nemotron-based large telco model (LTM), a comprehensive guide for building reasoning agents for network operations, and latest NVIDIA Blueprints for energy saving and network configuration with multi-agent orchestration. These resources are being released as open resources through GSMA, an organization representing the mobile communications industry, as part of GSMA’s new Open Telco AI initiative.

Open Nemotron 3 Large Telco Model

To successfully implement generative and agentic AI, telecom AI models must understand the specific language and complex workflows of the industry. NVIDIA, in collaboration with AdaptKey AI, has released an open-source, 30-billion-parameter NVIDIA Nemotron LTM. This model allows operators worldwide to build autonomous networks.

Built on the NVIDIA Nemotron 3 family of foundation models and fine-tuned by AdaptKey AI using open telecom datasets – including industry standards and synthetic logs – the LTM is optimized to understand telecom terminology and reason through tasks like fault isolation, remediation planning, and change validation.

As an open model, the Nemotron LTM provides full transparency into its training data and processes, enabling secure and rapid on-premises deployment. This allows telcos to build and run agents directly within their networks, while maintaining control over data and security. They can also safely adapt and extend the model with their own network and operational data.

Teaching AI Agents to Reason Like Network Engineers

NVIDIA and Tech Mahindra have jointly published an open-source guide detailing how telecom operators can fine-tune domain-specific reasoning models and build agents capable of safely executing network operations center (NOC) workflows.

The guide outlines a framework for teaching models to reason like NOC engineers by focusing on high-impact, frequent incident categories. It emphasizes translating expert resolutions into step-by-step procedures and converting those into structured reasoning traces that capture each action, tool call, outcome, and decision. These traces serve as “thinking examples” for the model, enabling it to understand not only what to do but why a particular sequence of checks and fixes is safe and effective.

Using the NVIDIA NeMo-Skills pipeline, operators can fine-tune a reasoning model on these traces, creating a foundation for telco-specialized AI agents that can reason and solve problems with the expertise of a network engineer.

Optimizing Energy Efficiency with Intent-Driven Blueprints

Autonomous networks rely on closed-loop operation – models that understand the network, agents that act on intent, and simulation that validates and refines decisions. NVIDIA’s new Blueprint for intent-driven RAN energy efficiency brings these elements together, helping operators systematically reduce power consumption in 5G radio access networks (RAN) while maintaining quality of service.

The blueprint integrates network test and measurement leader VIAVI’s TeraVM AI RAN Scenario Generator (AI RSG) platform. This platform generates synthetic network data – including cell utilization, user throughput, and traffic patterns – and converts it into a queryable format.

An energy planning agent then reasons over this synthetic data to generate energy-saving policies that can be simulated in AI RSG, allowing operators to safely validate these policies in a closed loop, meeting their intent without impacting live configurations or subscribers.

Blueprint Adoption by Telecom Operators

The NVIDIA Blueprint for telco network configuration is gaining traction among operators globally.

• Cassava Technologies is utilizing the blueprint to build Cassava Autonomous Network, an agentic platform designed to optimize Africa’s diverse, multi-vendor mobile network environment. The platform incorporates three agents: one for network monitoring and configuration recommendations, one for applying changes with documentation and governance, and one for assessing impact and safely rolling back changes if needed.
• NTT DATA is implementing the blueprint to enhance traffic regulation, managing surges when users reconnect after outages, and is deploying it with a tier 1 operator in Japan. An AI agent analyzes real-time demand and dynamically adjusts user admission on specific cells, creating a data-driven optimization cycle for more resilient mobile networks.

Multi-Agent Orchestration for Evolving Networks

To help telcos design, observe, and optimize complex agentic workflows across the RAN, NVIDIA and BubbleRAN are enhancing the NVIDIA Blueprint for telco network configuration with NVIDIA NeMo Agent Toolkit (NAT) and BubbleRAN Agentic Toolkit (BAT). These complementary frameworks facilitate multi-agent orchestration.

BubbleRAN is integrating NAT and BAT into its Opti-Sphere platform to manage network monitoring, configuration, and validation agents more flexibly across containers and workloads. This integration connects agents to tools that report network metrics and traffic status, enabling continuous proposal and validation of configuration changes.

Telenor Group will be the first telco to adopt the blueprint with BubbleRAN to enhance its 5G network for Telenor Maritime, its global connectivity provider at sea.

Further details on these advancements in agentic AI for telecommunications will be available at Mobile World Congress, taking place in Barcelona from March 2-5.