Zoom pioneers the next era of custom enterprise AI with NVIDIA
Zoom (NASDAQ: ZM) and NVIDIA announced a partnership on Oct 28, 2025 to expand Zoom’s federated AI architecture by integrating NVIDIA Nemotron open technologies to power AI Companion 3.0. The collaboration adds a hybrid model approach that routes queries between Zoom’s proprietary Small Language Models (SLMs) and a fine-tuned Large Language Model for complex reasoning, aiming to improve latency, cost efficiency, and quality for enterprise use.
The release highlights a new 49-billion-parameter LLM built with NVIDIA NeMo tools, expanded retrieval-augmented generation (RAG) capabilities, and planned interoperability with Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce, and ServiceNow. Zoom reaffirmed privacy commitments, stating customer content will not be used to train models.
Zoom (NASDAQ: ZM) e NVIDIA hanno annunciato una partnership il 28 ottobre 2025 per espandere l'architettura federata di IA di Zoom integrando le tecnologie aperte NVIDIA Nemotron per alimentare AI Companion 3.0. La collaborazione aggiunge un approccio a modello ibrido che instrada le query tra i Small Language Models (SLM) proprietari di Zoom e un Large Language Model finemente sintonizzato per ragionamenti complessi, con l'obiettivo di migliorare latenza, efficienza dei costi e qualità per l'uso aziendale.
La nota stampa evidenzia un LLM da 49 miliardi di parametri costruito con gli strumenti NVIDIA NeMo, capacità espanse di retrieval-augmented generation (RAG) e interoperabilità prevista con Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce e ServiceNow. Zoom ha ribadito l'impegno per la privacy, affermando che il contenuto dei clienti non verrà utilizzato per addestrare modelli.
Zoom (NASDAQ: ZM) y NVIDIA anunciaron una asociación el 28 de octubre de 2025 para ampliar la arquitectura federada de IA de Zoom integrando las tecnologías abiertas de NVIDIA Nemotron para impulsar AI Companion 3.0. La colaboración añade un enfoque de modelo híbrido que enruta consultas entre los SLMs propietarios de Zoom y un Modelo de Lenguaje Grande afinado para razonamiento complejo, con el objetivo de mejorar la latencia, la eficiencia de costos y la calidad para uso empresarial.
El comunicado destaca un LLM de 49 mil millones de parámetros construido con herramientas NVIDIA NeMo, capacidades ampliadas de generación con recuperación (RAG) y la interoperabilidad planificada con Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce y ServiceNow. Zoom reafirmó su compromiso con la privacidad, afirmando que el contenido de los clientes no se utilizará para entrenar modelos.
Zoom (NASDAQ: ZM)와 NVIDIA는 2025년 10월 28일에 파트너십을 발표하여 NVIDIA Nemotron 오픈 기술을 통합해 AI 동반자 3.0을 강화하는 Zoom의 연합 AI 아키텍처를 확장합니다. 협력은 질의가 Zoom의 독점 Small Language Models(SLMs)와 복잡한 추론을 위한 미세 조정된 대형 언어 모델 사이를 오가도록 하는 하이브리드 모델 접근법을 도입하여 엔터프라이즈 용도에 대한 대기 시간, 비용 효율성 및 품질을 개선하는 것을 목표로 합니다.
발표는 NVIDIA NeMo 도구로 구축된 490억 매개변수 LLM를 포함하고, 검색 강화 생성(RAG) 기능이 확장되며 Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce 및 ServiceNow와의 상호 운용성 계획이 있다고 밝힙니다. Zoom은 고객 콘텐츠가 모델 학습에 사용되지 않을 것이라고 밝히며 개인정보 보호 약속을 재확인했습니다.
Zoom (NASDAQ: ZM) et NVIDIA ont annoncé un partenariat le 28 octobre 2025 pour étendre l’architecture fédérée d’IA de Zoom en intégrant les technologies ouvertes NVIDIA Nemotron afin d’alimenter AI Companion 3.0. La collaboration introduit une approche de modèle hybride qui fait transiter les requêtes entre les Modèles de Langage Restreints propriétaires de Zoom et un Grand Modèle de Langage finement ajusté pour un raisonnement complexe, dans le but d’améliorer la latence, l’efficacité des coûts et la qualité pour un usage entreprise.
Le communiqué met en avant un LLM de 49 milliards de paramètres construit avec les outils NVIDIA NeMo, des capacités accrues de génération avec récupération (RAG), et une interopérabilité planifiée avec Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce et ServiceNow. Zoom a réaffirmé ses engagements en matière de confidentialité, déclarant que le contenu des clients ne sera pas utilisé pour entraîner des modèles.
Zoom (NASDAQ: ZM) und NVIDIA kündigten eine Partnerschaft am 28. Oktober 2025 an, um die föderierte KI-Architektur von Zoom zu erweitern, indem NVIDIA Nemotron Open Technologies integriert werden, um AI Companion 3.0 zu betreiben. Die Zusammenarbeit ergänzt einen Hybrid-Modell-Ansatz, der Anfragen zwischen Zooms proprietären Small Language Models (SLMs) und einem fein abgestimmten Large Language Model für komplexe Überlegungen weiterleitet, mit dem Ziel, Latenz, Kosten-Effizienz und Qualität für den Unternehmenseinsatz zu verbessern.
Die Veröffentlichung hebt ein LLM mit 49 Milliarden Parametern hervor, aufgebaut mit NVIDIA NeMo-Tools, erweiterte Retrieval-Augmented Generation (RAG)-Fähigkeiten und eine geplante Interoperabilität mit Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce und ServiceNow. Zoom bekräftigte erneut seine Datenschutzverpflichtungen und erklärte, dass Kundendaten nicht zum Trainieren von Modellen verwendet werden.
Zoom (NASDAQ: ZM) و NVIDIA أعلنتا عن شراكة في 28 أكتوبر 2025 لتوسيع بنية الذكاء الاصطناعي المجمّعة لـ Zoom من خلال دمج تقنيات NVIDIA Nemotron المفتوحة لتشغيل AI Companion 3.0. تضيف التعاون نهج نموذج هجين يوجّه الاستفسارات بين نماذج اللغة الصغيرة المملوكة لـ Zoom ونموذج لغة كبير مُعدّل لاستنتاجات معقدة، بهدف تحسين الكمون وكفاءة التكلفة والجودة للاستخدام المؤسسي.
يسلّط الإصدار الضوء على LM بسعة 49 مليار معامل مبنية باستخدام أدوات NVIDIA NeMo، وقدرات توليد معززة باسترجاع (RAG) والتشغيل البيني المخطط مع Microsoft 365، Microsoft Teams، Google Workspace، Slack، Salesforce وServiceNow. جدد Zoom تأكيد التزامه بالخصوصية، قائلاً إن محتوى العملاء لن يُستخدم لتدريب النماذج.
Zoom (NASDAQ: ZM) 与 NVIDIA 于 2025年10月28日 宣布建立伙伴关系，通过整合 NVIDIA Nemotron 开放技术来推动 AI Companion 3.0，以扩展 Zoom 的联邦AI架构。该合作引入混合模型方法，在 Zoom 的自有小型语言模型（SLMs）与用于复杂推理的微调大型语言模型之间路由查询，旨在提升企业使用的延迟、成本效率和质量。
公告强调一个由 NVIDIA NeMo 工具构建、具有 490亿参数的 LLM，扩展的基于检索的生成（RAG）能力，以及与 Microsoft 365、Microsoft Teams、Google Workspace、Slack、Salesforce 和 ServiceNow 的互操作性计划。Zoom 重申隐私承诺，表示不会使用客户内容来训练模型。
- 49-billion-parameter LLM developed with NVIDIA NeMo tools
- Federated hybrid model routing SLMs and a fine-tuned LLM
- Planned integrations with Microsoft 365, Teams, Google Workspace, Slack, Salesforce, ServiceNow
- Claims of lower latency and optimized cost for enterprise AI workflows
- None.
Insights
Zoom expands its federated AI stack with NVIDIA Nemotron to speed, customize, and scale enterprise AI.
Zoom’s federated architecture now integrates NVIDIA Nemotron open technologies and a new 49-billion-parameter LLM built with NVIDIA NeMo tools to route workloads between Small Language Models (SLMs) and a fine-tuned Large Language Model for complex reasoning.
The business mechanism ties model selection to task needs, aiming to improve latency, cost efficiency, and reasoning by dynamically routing queries to the optimal model. The announcement cites existing real-time capabilities (transcription, translation, summarization) and extends them via Nemotron-based reasoning models.
Key dependencies and risks include integration quality between SLMs, the new 49B LLM, and third-party services named (Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce, ServiceNow); performance gains depend on effective routing and fine-tuning. Security and privacy depend on Zoom’s stated commitment not to use customer content for model training.
Watch for measurable deployment milestones such as broader availability of AI Companion 3.0, performance metrics on latency and cost per task, and adoption in regulated sectors (
The partnership signals material product differentiation for Zoom by offering customizable, faster AI for enterprise workflows.
Zoom positions AI Companion 3.0 as a hybrid model stack that mixes proprietary SLMs with a fine-tuned LLM and Nemotron-based reasoning to surface task-appropriate responses and integrate with major collaboration platforms. That structure intends to balance cost, speed, and deeper reasoning.
Operational success will rely on proven cost and latency improvements, secure handling of enterprise data, and seamless connectors to listed services. The release highlights expansion into finance, healthcare, and government but states privacy is central and that Zoom does not use customer communications to train models.
Near-term monitoring items include release cadence for AI Companion 3.0, published benchmarks for accuracy/latency, and customer adoption in the named industries; these items will reveal whether the claimed efficiencies and interoperability materialize.
- Zoom AI Companion adapts to users’ personal work habits through model customization and retrieval-augmented generation powered by NVIDIA technology
- Zoom’s federated AI stack expands with NVIDIA Nemotron models that make AI faster and smarter for every business
SAN JOSE, Calif., Oct. 28, 2025 (GLOBE NEWSWIRE) -- Zoom Communications, Inc. (NASDAQ: ZM) and NVIDIA are working together to deliver faster, higher-quality, and customizable AI to the enterprise. Zoom’s federated architecture expands to include NVIDIA Nemotron open technologies to support AI Companion 3.0 across industries such as finance, healthcare, and government. This framework is expanding to a next-generation hybrid language model approach, a breakthrough AI architecture that can intelligently route queries between Zoom’s proprietary Small Language Models (SLMs), optimized for low latency and quality for a particular skill or tasks, and a fine-tuned Large Language Model (LLM) for complex reasoning. The new hybrid model will accelerate enterprise productivity and collaboration experiences with optimized cost efficiency, quality, and latency.
Zoom’s AI framework uses a federated architecture to intelligently select the best AI model for each task. Through this approach, Zoom AI Companion can leverage NVIDIA’s advanced AI infrastructure, services, and software, alongside other models, delivering enhanced capabilities while optimizing cost for customers. This includes Zoom’s new 49-billion-parameter LLM, based on NVIDIA Nemotron and developed with NVIDIA NeMo tools to achieve the optimal balance of speed, cost, and accuracy. This innovative architecture allows enterprise customers and government organizations to benefit from both open and closed model innovation, delivering greater cost efficiency, faster AI workflows, deeper reasoning capabilities, and more efficient collaboration within AI Companion.
Accelerated AI innovation and interoperability
This initiative is set to power the next evolution of Zoom’s federated AI architecture, which dynamically incorporates optimal models, including the Llama Nemotron Super-based reasoning model, to deliver the ideal balance of accuracy, performance, and cost. Zoom’s patent-pending federated AI approach has already proven its ability to deliver high-quality performance for real-time transcription, translation, and summarization. This new approach extends Zoom’s exceptional performance with the NVIDIA open model innovations.
“We’ve increased our speed and enhanced lower-cost model decision making using NVIDIA GPUs and AI software stack, helping to optimize AI Companion’s core capabilities and enable faster go‑to‑market timelines,” said X.D. Huang, chief technology officer at Zoom. “With the help of NVIDIA Nemotron open technologies, we’re accelerating the development of our enterprise retrieval-augmented generation (RAG) capabilities, allowing AI Companion to work seamlessly with Microsoft 365, Microsoft Teams, Google Workspace, Slack, Salesforce, and ServiceNow. This partnership allows us to deliver powerful, security-focused, and scalable AI experiences to our customers at rapid speed.”
Through this work, NVIDIA and Zoom are further accelerating AI innovation and enterprise readiness.
“The integration of NVIDIA Nemotron into Zoom AI Companion allows enterprises like NVIDIA to experience a private, powerful, and personalized work environment for enhanced productivity," said Kari Briski, vice president of generative AI software at NVIDIA. "Our collaboration with Zoom to bring smart and efficient AI reasoning to customers is just beginning.”
A responsible AI foundation
Zoom is committed to expanding the AI frontier responsibly and plans to utilize NVIDIA Nemotron reasoning models to enhance decision-making and manage multistep tasks across meetings, chat, documents, and more. These models will empower AI Companion to reason more deeply, automate workflows, and deliver intelligent assistance throughout the enterprise.
The collaboration also lays the groundwork for future AI deployments, which will allow Zoom to further extend its AI capabilities to industries such as finance, healthcare, and government, where data privacy is paramount.
Zoom has taken a strong stance on responsible AI, with security and privacy at the core of its generative AI capabilities. Zoom does not use any customer audio, video, chat, screen sharing, attachments, or other communications like customer content (such as poll results, whiteboard, and reactions) to train Zoom’s or its third-party artificial intelligence models.
About Zoom
Zoom’s mission is to provide an AI-first work platform for human connection. Reimagine teamwork with Zoom Workplace — Zoom’s open collaboration platform with AI Companion that empowers teams to be more productive. Together with Zoom Workplace, Zoom’s Business Services for sales, marketing, and customer experience teams, including Zoom Contact Center, strengthen customer relationships throughout the customer lifecycle. Founded in 2011, Zoom is publicly traded (NASDAQ:ZM) and headquartered in San Jose, California. Get more information at zoom.com.
Zoom Public Relations
Lacretia Nichols
press@zoom.us
FAQ
What did Zoom announce with NVIDIA on Oct 28, 2025 for ZM shareholders?
How will Zoom AI Companion 3.0 change performance for enterprise users of ZM?
Does the Zoom and NVIDIA partnership support Microsoft Teams and Google Workspace integrations for ZM customers?
What is the size and purpose of Zoom’s new LLM mentioned in the ZM announcement?
Will Zoom use customer content to train these AI models after the NVIDIA deal?
