NVIDIA Unveils Autonomous AI Agent Stack for Telecom Operations at DTW Ignite 2026

At TM Forum's DTW Ignite 2026 in Copenhagen in June 2026, NVIDIA and its partners introduced an AI agent stack that lets telecom operators build autonomous network operations. The company showcased a combination of data generation, domain-specific models, simulation and a secure runtime to enable "trusted," round-the-clock agents (NVIDIA Blog).

The core of the announcement is support for moving from "task-based automation" that merely speeds up predefined steps toward end-to-end "Autonomous Networks" in which agents detect and coordinate issues across network, IT and business systems. NVIDIA structures this into four layers. At the data layer, it offers NeMo Safe Synthesizer and NeMo Anonymizer to generate privacy-preserving synthetic data for operators that cannot feed sensitive network or customer data directly into AI. At the model layer, it provides telecom-domain open models including Nemotron. At the runtime layer, it pairs NemoClaw, a reference stack for long-running agents, with OpenShell, a secure runtime featuring policy-based guardrails and sandboxing. At the simulation layer, it handles digital twins and radio access network (RAN) simulation using RTX PRO 6000 Blackwell Server Edition GPUs.

This design responds to a long-standing data barrier in the industry. Some 54% of operators cite data-related issues as their biggest challenge, as sensitive data cannot be handed to AI as-is, holding back autonomy efforts. NVIDIA positions the bundle of synthetic data, domain models, secure runtime and simulation as a "telecom autonomy platform" that delivers intent understanding, safe operation and human-in-the-loop policy control.

Multiple partners demonstrated concrete use cases on site.