AI · Sep 30, 2026
The four rules in the White House AI commitment, and the missing fifth: penaltiesNvidia survey finds 89% of telecom respondents call open models key, and adds a 30B telco model
· Updated · Signal Desk newsroom
Analysis of reporting by NVIDIA Blog. How we work

Nvidia’s State of AI in Telecommunications report says 89% of respondents see open-source models and software as important to their AI strategy. Nvidia also announced a 30-billion-parameter Nemotron 3 Large Telco Model fine-tuned by AdaptKey, and a recipe for operators to adapt open models.
Key takeaways
- 89% of survey respondents said open-source models and software are important to their AI strategy.
- Nvidia announced a 30B Nemotron 3 Large Telco Model fine-tuned by AdaptKey on open telecom data.
- A full NeMo-based recipe lets operators adapt open models to their networks.
- Indosat’s Sahabat-AI and SoftBank’s telecom model build on open foundations.
- The survey, models and platform all come from Nvidia, which shapes the framing.
What Nvidia reported
According to Nvidia’s latest State of AI in Telecommunications report, 89% of respondents said open-source models and software are important to their company’s AI strategy. Nvidia argues the appeal goes beyond cost: carriers want to inspect, fine-tune and govern the models running their networks and customer care.
Alongside the report, Nvidia announced a 30-billion-parameter Nemotron 3 Large Telco Model, fine-tuned by AdaptKey on open telecom datasets, and released a full recipe for operators to adapt open models to their own networks using its NeMo libraries. Carriers are also positioning themselves as hosts for national AI efforts, fine-tuning open models for local languages, industries and data rules and offering them to enterprise and government customers.
Why control appeals to carriers
Telecom networks are critical infrastructure. An operator that can see a model’s weights, train it on its own network and customer data, and run it on private or edge hardware has more control than one renting a closed system. AT&T’s chief data and AI officer describes matching each workload to the right mix of performance, cost and control, a mature way to frame the decision.
The local-language angle stands out. Indosat Ooredoo Hutchison’s Sahabat-AI models are built to reflect Indonesian language and culture, serving a market on its own terms in a way closed models rarely prioritize; Indosat has also worked with Nvidia on local AI talent (Indonesia AI center). SoftBank describes building its own large telecom model on open foundations, including Nemotron. Nvidia’s broader push into open ecosystems includes its Hugging Face acquisition (Hugging Face deal).
Caveats
The source is Nvidia’s blog, describing an Nvidia survey and promoting Nvidia models, recipes and an end-to-end Nvidia platform. “Open” in this context sits on a stack largely controlled by one company. Operators can welcome open weights while tracking how much surrounding tooling ties them to a single vendor’s hardware and software.
The autonomous-network vision also carries risk. Fine-tuning on customer records requires serious privacy work, which Nvidia acknowledges with references to anonymization and synthetic data. Agents that configure networks or triage incidents can fail at scale in ways a chatbot cannot, so safeguards and audit trails need to keep pace with deployment. Independent results for the telco model on real operator workloads have not yet been published. More in our Nvidia coverage, enterprise coverage and business coverage.
Who feels it
- Telecom operators
- Open weights and a published recipe lower the bar for in-house network models.
- Enterprise and government customers
- Local carriers may offer locally hosted, locally adapted AI.
- Subscribers
- Better automated care is possible, but tuning on customer data raises privacy stakes.
What to watch
- Independent results for the Nemotron 3 Large Telco Model.
- More local-language model families like Sahabat-AI.
- Operator stacks that avoid single-vendor dependence.
FAQ
What share of respondents value open models?
89%, according to Nvidia’s State of AI in Telecommunications report.
What model did Nvidia announce?
A 30-billion-parameter Nemotron 3 Large Telco Model fine-tuned by AdaptKey.
Which carriers are cited?
AT&T, Indosat Ooredoo Hutchison and SoftBank.
What is the main caveat?
The survey and platform come from Nvidia, and open weights may still sit on a single-vendor stack.
Companies: NVIDIA
More on NVIDIA
AI · Oct 7, 2026
Microsoft prices its first RTX Spark PCs; the agent sandbox coming to all Windows 11 users may matter moreAI · Sep 24, 2026
Nvidia, DeepMind and EMBL-EBI release predicted structures for protein complexes from 2,800-plus virusesAI · Oct 21, 2026
Europe’s banks are training their own AI on open weights, with Nvidia underneathAI · Oct 21, 2026
Nvidia turns Cosmos into a synthetic-data factory for robots’ rarest, riskiest moments