Rockwell Automation to advance industrial intelligence with NVIDIA Nemotron

In collaboration with NVIDIA, Rockwell is leveraging the open-source Nemotron-Nano-9B-v2 model and NVIDIA NeMo to deliver an edge-based generative AI capability designed specifically for industrial environments. Nemotron Nano distillation techniques provide the foundation for an SLM that can run in edge environments with less space and power than a traditional data centre. By fine-tuning the model with data used by FactoryTalk Design Studio Copilot, Rockwell is creating a solution that demonstrates new potential for industrial automation professionals.

Built for use across design, development, production and maintenance workflows, the model operates seamlessly on HMI panels, appliances, desktop IDEs and server or private cloud environments. It supports both edge and air-gapped deployments, offering improved reasoning, predictability and responsiveness compared to other SLMs.

Early evaluations show the step change value of this model having new reasoning, parallel processing and key performance breakthroughs that enable it to stand out in the SLM space. Results highlight the model's strong fit for industrial edge scenarios where instant responsiveness, data security and offline operation are essential.

"Industrial automation demands AI that works reliably at the edge and in secure environments," said Tony Carrara, FactoryTalk Design Studio business manager, Rockwell Automation. "By fine-tuning the open NVIDIA Nemotron model with FactoryTalk Design Studio data, we're creating solutions that can be deployed anywhere to help our customers accelerate workflows without compromising predictability or control."

"Small language models like NVIDIA Nemotron Nano bring real-time intelligence to where decisions are made - from factory floors to power grids," said Joey Conway, senior director, generative AI software for enterprise, NVIDIA. "With NVIDIA Nemotron Nano, enterprises using Rockwell FactoryTalk Design Studio can deploy AI in environments with limited space and power, extending AI from data centres into the heart of real-world operations."