NVIDIA's State of AI in Telecommunications report shows 89% of respondents value open source models. Learn why operators like SoftBank, AT&T, and Indosat are building on open
89% of respondents reporting that open source models and software are important to their company's AI strategy
NVIDIA announced the 30-billion-parameter Nemotron 3 Large Telco Model (LTM), fine-tuned by AdaptKey
SoftBank Corp. is using open foundations extensively, including NVIDIA Nemotron models, in developing its SoftBank Large Telecom Model
Indosat Ooredoo Hutchison is building AI that reflects the country's own language and culture through its Sahabat-AI family of open source models
NVIDIA's latest State of AI in Telecommunications report indicates a strategic shift among telecom operators toward open models, with 89% of respondents reporting that open source models and software are important to their company's AI strategy. The NVIDIA blog post, published October 6, 2026, outlines five strategic values of open models for operators, including trust, control, and customization across workloads from autonomous networks to customer care. The post also announces the 30-billion-parameter Nemotron 3 Large Telco Model (LTM), fine-tuned by AdaptKey on open source telecom datasets to deliver accuracy gains for telecom-specific tasks. This model provides an open baseline that can understand telecom industry terminology and reason through operations workflows such as network configuration and customer incident triage.
For AI practitioners in telecom, the shift means open models are becoming practical building blocks for production workflows. The NVIDIA post describes how operators need pipelines that prepare and protect data for fine-tuning by anonymizing sensitive records and generating privacy-preserving synthetic datasets. NVIDIA released the full recipe for the end-to-end fine-tuning pipeline using NVIDIA NeMo open libraries, allowing operators to adapt open models to their specific networks, customers, and procedures. The post also highlights AT&T's approach, where chief data and AI officer Andy Markus states the future is about intelligently matching every workload to the right combination of performance, cost, and control. This suggests a move away from single-model strategies toward governed, autonomous agentic workflows.
Readers can verify the reported statistics and announcements directly from the NVIDIA blog post. The post states that 89% of respondents in the State of AI in Telecommunications report consider open source models and software important to their AI strategy. It also names specific operators and their initiatives: SoftBank Corp. is developing its SoftBank Large Telecom Model using open foundations including NVIDIA Nemotron models, as stated by principal fellow Rajeev Koodli. Indosat Ooredoo Hutchison is building its Sahabat-AI family of open source models to reflect Indonesia's language and culture, according to chief data and AI officer Chirag Sukhadia. The 30-billion-parameter Nemotron 3 LTM announcement is also directly verifiable in the source.
The supplied feed entry is a concise summary; consult the canonical article for full context. The NVIDIA blog post does not provide details on the performance benchmarks of the Nemotron 3 LTM beyond stating it delivers accuracy gains for telecom-specific tasks. The post also does not specify the timeline for availability of the fine-tuning recipe or the Nemotron 3 LTM, nor does it detail the specific results operators have achieved with these open models in production. The post mentions NVIDIA AI Enterprise software and NVIDIA Agent Toolkit as part of an end-to-end platform, but does not elaborate on the technical architecture or integration requirements. Additionally, the post does not quantify the cost savings or operational improvements that open models might deliver compared to proprietary alternatives.
According to the NVIDIA blog post, the Nemotron 3 Large Telco Model (LTM) is a 30-billion-parameter model fine-tuned by AdaptKey on open source telecom datasets. It is designed to deliver accuracy gains for telecom-specific tasks, understand telecom industry terminology, and reason through operations workflows such as network configuration and customer incident triage. NVIDIA released a full recipe for fine-tuning this model and other open models using NVIDIA NeMo open libraries.
The NVIDIA post explains that open models give operators a foundation to deliver AI services tailored to local languages, industries, regulations, and data governance requirements. For example, Indosat Ooredoo Hutchison is building its Sahabat-AI family of open source models to reflect Indonesia's language and culture. Operators can fine-tune open models for local needs and host them on trusted platforms for customers to consume or build upon.