Zoom Communications and NVIDIA have joined forces to advance the next era of enterprise AI through model customization and hybrid intelligence. The collaboration brings NVIDIA Nemotron open technologies into Zoom’s federated AI architecture to power AI Companion 3.0, enabling faster, smarter, and more adaptive AI experiences across industries such as finance, healthcare, and government.
Zoom’s new hybrid model intelligently routes queries between its Small Language Models (SLMs)—optimized for low latency and specialized skills—and a fine-tuned Large Language Model (LLM) for deeper reasoning. This design enhances enterprise productivity and collaboration while maintaining cost efficiency, accuracy, and performance.
The partnership integrates NVIDIA’s advanced AI infrastructure, GPUs, and NeMo tools to optimize Zoom’s new 49-billion-parameter LLM. This federated approach lets enterprises dynamically select the most efficient AI model for each task, unlocking powerful retrieval-augmented generation (RAG) workflows and seamless interoperability with tools like Microsoft 365, Google Workspace, Slack, Salesforce, and ServiceNow.
“With the help of NVIDIA Nemotron open technologies, we’re accelerating AI Companion’s core capabilities and enabling faster go-to-market timelines,” said X.D. Huang, CTO, Zoom. “Our collaboration delivers security-focused, scalable, and high-performance AI experiences to customers.”
Kari Briski, VP of Generative AI Software at NVIDIA, added, “Integrating Nemotron into Zoom AI Companion allows enterprises to experience a private, personalized AI environment built for productivity.”
Both companies emphasize a strong foundation of responsible AI, ensuring customer data privacy remains intact. Zoom confirms that it does not use customer content—such as video, audio, chat, or shared materials—to train its own or third-party AI models.
The initiative marks a milestone in federated, customizable AI—where enterprises can balance speed, privacy, and performance to deploy next-generation intelligent collaboration.