Why Telecom Operators Are Building Their AI Strategy on Open Models
Telecom operators are adopting open AI models to customize systems for network operations and customer care, citing trust, control, and cost benefits over proprietary alternatives.
Telecom operators are shifting their AI strategies toward open models to gain trust, control, and customization capabilities for critical workloads such as autonomous networks and customer service. According to NVIDIA’s latest State of AI in Telecommunications report, 89% of respondents consider open source models and software essential to their AI strategies. Open models like NVIDIA’s Nemotron family provide frontier-level reasoning and speech capabilities with transparent weights, training data, and recipes, enabling operators to tailor AI to their specific needs.
SoftBank Corp. is leveraging open models to advance its telecom-specific AI capabilities, including the development of the SoftBank Large Telecom Model. Rajeev Koodli, principal fellow and senior vice president at SoftBank Corp., highlighted the ability to build on global foundation models while applying the company’s network expertise. The Nemotron 3 Large Telco Model, a 30-billion-parameter model fine-tuned on open telecom datasets, offers operators an open baseline for tasks like network configuration and customer incident triage, improving accuracy in telecom workflows.
NVIDIA provides tools to help operators customize open models with their operational data, including a full recipe for fine-tuning using NVIDIA NeMo libraries. AT&T emphasizes the importance of model choice, aligning flexibility with business priorities to ensure performance, cost, and control. The company collaborates with NVIDIA to implement a model-choice strategy that meets scale, reliability, and governance requirements for production deployment.
Operators also require data pipelines to prepare and protect sensitive information, along with platforms to deploy governed, autonomous agentic workflows. NVIDIA’s end-to-end platform, powered by NVIDIA AI Enterprise software and NVIDIA Agent Toolkit, supports data pipelines, open models, agent orchestration, secure runtimes, and simulation. Indosat Ooredoo Hutchison in Indonesia is using open models like Sahabat-AI to adapt AI to local language, culture, and regulatory needs, enabling broader adoption and local innovation.