Nvidia targets fully autonomous AI for telecom networks

Nvidia just made its loudest play yet for the telecom industry’s AI future. At the DTW Ignite conference on June 23, the company showcased agentic AI systems designed to run network operations around the clock, without human intervention, pushing the industry toward what’s known as Level 4-5 network autonomy.

From copilot to autopilot

The telecom industry uses a framework from the TM Forum to measure how autonomous a network is, on a scale from Level 0 (fully manual) to Level 5 (fully autonomous). Most operators today sit somewhere between Levels 1 and 3, meaning humans are still very much in the loop for critical decisions.

Nvidia’s pitch is to accelerate the jump to Levels 4 and 5, where AI agents handle everything from energy optimization to network configuration without waiting for a human to approve each step.

The foundation for this push was laid back in March 2026, when Nvidia released a 30-billion-parameter large telco model (LTM) built on its open-source Nemotron platform. That release was part of the GSMA’s Open Telco AI initiative and included agent blueprints specifically designed for energy efficiency and network configuration tasks.

Nvidia has also been working with Tech Mahindra to enhance reasoning agents, the kind of AI that doesn’t just follow scripts but actually thinks through problems.

Telcos as AI factories

Nvidia is framing telecom operators not as mere connectivity providers, but as “AI factories” or “AI grids.” The idea is that the massive infrastructure telcos already own, towers, data centers, edge computing nodes, can be repurposed to run AI workloads at scale.

SK Telecom and T-Mobile US are among those investigating AI grid deployments with Nvidia’s technology stack, which includes its NeMo framework and NIM microservices platform.

The concept of token-metered AI services has emerged from these discussions. To be clear, this has nothing to do with crypto tokens. It refers to AI inference tokens, essentially units of compute that telcos could sell as a service.

The ROI question

Nvidia’s own State of AI report, based on surveys conducted in February and March 2026, found that 50% of respondents viewed autonomous networks as delivering the highest return on investment among all AI use cases in telecom.

What this means for investors

The industry is simultaneously preparing for AI-RAN deployments, where AI handles radio access network optimization in real time, and laying groundwork for 6G technologies. Both trends feed directly into demand for Nvidia’s hardware and software.

Sovereign AI initiatives add another layer. Countries around the world are investing in domestic AI infrastructure, and telecom operators are natural partners for these deployments.

Nvidia’s NeMo and NIM platforms give it a software moat on top of its hardware dominance, making it harder for telecom operators to mix and match vendors. Once a network is built on Nvidia’s LTM and agent blueprints, switching costs become significant.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.