Telecom Operators Build AI Strategies on Open Models for Control and Customization

Telecom operators are increasingly anchoring their AI strategies on open models, driven by the need for trust, control, and customization across critical workloads like autonomous networks and customer care. NVIDIA’s latest State of AI in Telecommunications report highlights this shift, noting that 89% of respondents consider open source models and software important to their company’s AI strategy.

Strategic Advantages of Open Models

The strategic value for operators is fivefold, extending beyond simple cost savings:

• Access and Cost: Open models provide frontier-level intelligence at lower costs, allowing operators to reserve closed models for high-value workloads. Benchmarks like the Artificial Analysis Intelligence Index v4.3.2 show leading open models are becoming competitive in reasoning, coding, and agentic tasks.
• Customization: Open weights and training recipes enable fine-tuning with proprietary network, customer, and industry data.
• Trust and Governance: Greater visibility into model artifacts allows for evaluation and governance aligned with regulations.
• Flexible Deployment: Models can be optimized for public clouds, private infrastructure, and edge environments.
• New Revenue Streams: Operators can host and fine-tune models to deliver locally adapted AI services to enterprise and government clients.

Industry Implementations and Tools

SoftBank Corp. is using open foundations, including NVIDIA Nemotron models, to develop its SoftBank Large Telecom Model for network operations and management. To support this ecosystem, NVIDIA announced the 30-billion-parameter Nemotron 3 Large Telco Model (LTM), fine-tuned by AdaptKey on open source telecom datasets to improve accuracy for tasks like network configuration and incident triage. NVIDIA also released a full fine-tuning recipe using NVIDIA NeMo open libraries to help operators adapt models to their specific networks.

Production and Local Innovation

Moving from models to production requires robust pipelines for data protection and agent orchestration. AT&T emphasizes matching workloads to the right combination of performance, cost, and control, leveraging NVIDIA AI Enterprise and the NVIDIA Agent Toolkit to scale these strategies securely. Meanwhile, operators like Indosat Ooredoo Hutchison are using open models to build Sahabat-AI, a family of open source models tailored to Indonesia’s language and culture, enabling local AI innovation aligned with national strategies.