Why Open AI Models Are Becoming Strategic for Telecommunications Carriers

Telecommunications companies are increasingly building AI strategies around open models—and the reason goes beyond cost.

Open models give carriers greater trust, control and flexibility to customize AI for critical workloads, from autonomous networks to customer care.

NVIDIA’s latest AI in communications report reflects this shift: 89% of respondents said open-source models and software are important to their company’s AI strategy.

For carriers, the strategic value of open models can be summarized in five areas.

1. Expand access to frontier-level intelligence.
   Operators can use open models at a lower cost while reserving closed models for workloads that generate the most value. Independent benchmarks, including the Artificial Analysis Intelligence Index v4.3.2, show that leading open models are gaining ground across demanding inference, coding, scientific and agent workloads.
2. Customize AI for telecom operations.
   Open weights and training recipes allow carriers to fine-tune models for their own operations using network, customer and industry data.
3. Build more trustworthy AI.
   Greater visibility into model artifacts and behavior helps carriers evaluate, adapt and manage models in line with regulations and business policies.
4. Deploy AI flexibly and securely.
   Teams can size and optimize open models for public clouds, private infrastructure and edge environments.
5. Create new AI services.
   Telcos can host and fine-tune open models to deliver locally adapted AI services to business and government customers.

An Open AI Foundation for Communications Operations

The NVIDIA Nemotron open model family provides frontier-level inference performance for agent workflows and voice applications, supported by open weights, training data and recipes.

SoftBank Corp. is demonstrating how carriers can use open models as the foundation for developing and continuously improving telecom-specific AI capabilities.

“The open model allows SoftBank Corp. to build on the rapid advances in the global infrastructure model while applying the network knowledge and operational expertise we have accumulated over many years,” said Rajeev Khoudri, principal fellow at SoftBank Corp. and senior vice president at SB Telecom America. “We have extensively used open infrastructure, including the NVIDIA Nemotron model, in the development of the SoftBank large-scale communications model, and are continually evolving the model for communications-specific use cases such as network operations, design and overall management.”

NVIDIA is also working with partners to transform open models into practical building blocks for communications AI.

NVIDIA announced the 30-billion-parameter Nemotron 3 Large Carrier Model (LTM), fine-tuned by AdaptKey using open-source communications datasets. The model, available through AdaptKey, is designed to improve accuracy for communications-specific tasks.

Nemotron 3 LTM provides an open baseline for carriers to understand telecom terminology and reason through communications workflows, including network configuration and customer incident triage.

To help operators customize Nemotron 3 LTM and other open models with their own operational data, NVIDIA provides a complete recipe for running an end-to-end fine-tuning pipeline. The process helps adapt open models to carrier-specific networks, customers and procedures using the NVIDIA NeMo open-source library.

Scale Open Models Into Production Telecom Workflows

Open models are a critical building block, but models alone are not enough to bring autonomous communications operations safely into production.

For AT&T, the value of model selection lies in the flexibility to align AI with business priorities.

“At AT&T, we believe the future of AI is not about choosing a single model, but about intelligently matching any workload with the right combination of performance, cost and control,” said Andy Markus, chief data and AI officer at AT&T. “An open model is essential to that approach, and NVIDIA allows us to bring our model selection strategy into production with the scale, reliability and governance our business requires.”

Operators need data pipelines that prepare and secure information for model fine-tuning by anonymizing sensitive records and generating privacy-preserving synthetic datasets. They also need platforms that can turn open models into managed, autonomous agent workflows.

NVIDIA offers an end-to-end platform built on NVIDIA AI Enterprise software, the NVIDIA Agent Toolkit, data pipelines, open models, agent orchestration, secure runtimes and simulations.

Supported by a broad partner ecosystem, this platform gives carriers a path to convert the benefits of open models into production-ready AI workflows.

Build a Platform for Local AI Innovation

As telecommunications carriers build AI infrastructure aligned with national AI strategies, open models provide a foundation for delivering services tailored to local language, industry, regulatory and data-governance requirements.

National carriers can fine-tune open models for local language and industry needs, then host them on trusted platforms that customers can use directly or build on with their own data and applications.

For Indosat Ooredoo Hutchison, one of Indonesia’s largest telecom operators, this means developing AI that reflects Indonesia’s language and culture through the Sahabat-AI family of open-source models.

“For a country like Indonesia, the value of an open model goes beyond access to powerful AI; it’s about adapting that intelligence to our own language, culture, data and real-world needs,” said Chirag Sukhadia, chief data and AI officer at Indosat Ooredoo Hutchison. “Sahabat-AI puts this into practice, using an open model as a foundation to understand Indonesia and build AI that can be developed for local applications. This will allow us to not only deploy AI, but also build capabilities around AI, allowing more Indonesians to create with AI on their own terms.”

Learn more about NVIDIA technology for communications.