NVIDIA launches NemoClaw and full-stack AI agent platform for autonomous telecom operations at DTW Ignite 2026

NVIDIA unveiled a comprehensive suite of products on June 22 designed to bring autonomous, 24/7 AI agents to telecommunications network operations, debuting the platform at DTW Ignite 2026 in Copenhagen — the flagship TM Forum industry conference.

What's new

The platform consists of four integrated layers:

• NemoClaw: A secure agent runtime that wraps AI agents in policy-based guardrails and sandboxed system access, allowing operators to define exactly what an autonomous agent can touch on live telecom infrastructure.
• NeMo Safe Synthesizer and NeMo Anonymizer: Paired synthetic-data tools that let carriers generate privacy-preserving training datasets from real network logs, addressing what NVIDIA says is the core bottleneck in telecom AI — "54% of operators cite data-related issues as their biggest barrier."
• Nemotron open models for telecom: Domain-specific language models tuned on telecom operations data and open for operators to fine-tune on their own environments.
• NVIDIA NV-Tesseract and Aerial Omniverse Digital Twin: GPU-accelerated simulation environments that validate AI agent recommendations in a virtual twin of the live network before any action touches production infrastructure.

Partner implementations demonstrated at the event included:

• AdaptKey and NTT DATA: Autonomous network self-healing agents
• Amdocs: AI-driven customer care automation
• ServiceNow: AI incident response integrated into IT operations workflows
• KDDI, Forsk, and VIAVI: RAN (radio access network) optimization and simulation pipelines

Context

NVIDIA's NeMo and Nemotron product lines have been growing for the last two years as the company moved beyond chips into full-stack enterprise AI infrastructure. The telecom industry is a natural fit: networks generate enormous volumes of structured operational data, fault patterns repeat predictably, and the cost of human-in-the-loop response at 2 a.m. is prohibitive. NVIDIA has been building toward agentic telco deployments since the Aerial Omniverse Digital Twin announcement in 2024, but this marks the first time the company has packaged secure runtime containment (NemoClaw), synthetic data (NeMo Safe Synthesizer), domain models, and simulation into a single named platform.

The push also aligns with broader telecom industry momentum: major carriers have been under pressure to cut operational expenditures while managing 5G densification, and autonomous fault-handling has been flagged by analyst groups as one of the few remaining large-scale cost levers.

Why it matters

The combination of secure agent runtime and simulation-before-deployment is the pair operators have been asking for — they want AI to act autonomously, but not on live infrastructure without a verified plan. NemoClaw's sandboxed access model, if it works as described, offers a concrete answer to the trust problem that has slowed agentic deployment in regulated industries broadly. For NVIDIA, this is also a land-and-expand play: once Nemotron models and the digital twin are embedded in a carrier's operations stack, the hardware refresh cycle for the simulation and inference infrastructure ties back to NVIDIA silicon. The partner list — ServiceNow, Amdocs, NTT DATA — represents enterprise software vendors with existing carrier relationships, meaning NVIDIA is entering telco ops through established distribution rather than direct sales.