According to a new NVIDIA blog post based on its State of AI in Telecommunications report, 89% of telecom respondents say open source models and software are important to their AI strategy. The post argues the appeal goes beyond cost: open weights let operators fine-tune models on proprietary network and customer data, inspect model behavior for governance, and deploy across clouds, private infrastructure, and edge environments.
Executives quoted in the post describe open models as a foundation for telecom-specific work. SoftBank's Rajeev Koodli says the company uses open foundations including NVIDIA Nemotron models to build its Large Telecom Model. AT&T's Andy Markus says the future is matching workloads to the right mix of performance, cost, and control, with open models essential to that approach. Indosat's Chirag Sukhadia says open models let the operator adapt AI to Indonesia's language and culture.
NVIDIA also announced the 30-billion-parameter Nemotron 3 Large Telco Model, fine-tuned by AdaptKey on open telecom datasets, and released a recipe for fine-tuning open models with NVIDIA NeMo. The post is a vendor perspective, so the benefits and adoption figures are presented by NVIDIA, but the named operator examples show how some large carriers are putting open models into practice.