Telecom Operators Shift to Open AI Models, NVIDIA Report Finds

Telecommunications operators are shifting their artificial intelligence strategies toward open models to gain deeper control and customization over critical workloads like autonomous networks and customer care, according to NVIDIA’s latest State of AI in Telecommunications report. The survey data shows that 89% of respondents consider open source models and software important to their company’s AI strategy.

Expanding Frontier Access and Customization for Telecom Workloads

Open models allow operators to access frontier-level intelligence while keeping expenses down, reserving closed models exclusively for tasks where they drive maximum value. Independent benchmarks, including the Artificial Analysis Intelligence Index v4.3.2, demonstrate that leading open models are increasingly competitive in demanding reasoning, coding, scientific, and agentic workloads. Beyond cost savings, open weights and training recipes let operators fine-tune models using their own network, customer, and industry data.

This visibility into model artifacts helps companies evaluate, adapt, and govern behavior in strict alignment with business policies and regulations. Teams can also size and optimize these models to run smoothly across public clouds, private infrastructure, and edge environments. This foundation creates opportunities for operators to host and fine-tune models locally to deliver tailored AI services directly to enterprise and government customers.

SoftBank and AT&T Scale Open Models in Production

Operators are already putting these capabilities into practice. SoftBank Corp. is utilizing open foundations, including NVIDIA Nemotron models, to build its proprietary SoftBank Large Telecom Model for network operations, design, and overall management. Open models allow SoftBank Corp. to build on the rapid progress of global foundation models while applying the network knowledge and operational expertise we have accumulated over many years, said Rajeev Koodli, principal fellow of SoftBank Corp. and senior vice president of SB Telecom America. Koodli added that the company is continuously advancing the model for telecom-specific use cases.

To accelerate these efforts, NVIDIA released the 30-billion-parameter Nemotron 3 Large Telco Model (LTM), which was fine-tuned by AdaptKey on open source telecom datasets. The model provides an open baseline designed to understand industry terminology and reason through complex workflows like network configuration and customer incident triage. Meanwhile, AT&T is integrating model choice directly into its business operations. At AT&T, we believe the future of AI is not about choosing a single model; it’s about intelligently matching every workload to the right combination of performance, cost and control, said Andy Markus, chief data and AI officer of AT&T. Markus noted that open models are essential to that strategy, allowing the company to bring model choice into production alongside the required scale, reliability, and governance.

Building Local AI Infrastructure in Indonesia

Beyond internal operations, open models provide operators with a foundation to align infrastructure with national AI strategies and local data governance requirements. Indosat Ooredoo Hutchison, one of Indonesia’s largest operators, is developing the Sahabat-AI family of open source models to reflect the nation’s specific language, culture, and real-world needs. For countries like Indonesia, the value of open models goes beyond access to powerful AI. It is about adapting that intelligence to our own language, culture, data and real-world needs, said Chirag Sukhadia, chief data and AI officer of Indosat Ooredoo Hutchison. Sukhadia stated that the initiative enables Indonesians to create with artificial intelligence on their own terms by building capabilities directly around trusted local foundations.

More on this story: The Agentic Shift: Scaling Enterprise AI Through Operating Models and Sovereign Data