IBM Fusion, the intelligent hyperconverged platform that unifies compute, storage, and data services, now supports the NVIDIA RTX PRO 6000 Blackwell Server Edition and NVIDIA Hopper GPUs as integrated options, delivering powerful new capabilities for deploying the NVIDIA AI Data Platform and NVIDIA AI Blueprints at scale.
At NVIDIA GTC Washington last week, we showed that the NVIDIA AI Data Platform reference design can now be deployed directly on IBM Fusion, a fully integrated AI inferencing platform designed to streamline lifecycle management, unify software-defined storage, and ensure high-speed access to both modern and legacy data.
By combining NVIDIA accelerated compute and networking with high-performance AI-ready storage from IBM, Fusion helps enterprise data become more seamlessly available to agentic AI applications. This innovation will help organizations accelerate retrieval-augmented generation (RAG), large-scale training, and inferencing workloads.
NVIDIA RTX PRO Servers bring the performance and energy efficiency of the NVIDIA Blackwell architecture to the enterprise data center, providing a universal data center platform to power AI factories and accelerate demanding enterprise AI, industrial AI, and visual computing workloads.
Fusion now supports multi-modal data vectorization, enabling rapid retrieval and inference performance for text, image, and video datasets. This capability can dramatically accelerate RAG pipelines by transforming globally distributed enterprise data of all types into high-speed, AI-ready vectors, eliminating external preprocessing steps and simplifying integration with frameworks such as NVIDIA NeMo and RAG LLM architectures
In a live demonstration from IBM at GTC DC, Fusion’s content-aware storage showcased how AI-driven storage intelligence can automatically organize, label, manage and retrieve unstructured content such as documents, images, and video. Fusion uses natural language processing and vector analytics to understand the meaning inside unstructured content, enabling faster insights and smarter decisions.
Fusion’s content-aware storage capabilities leverage active file management and a massively parallel file system to seamlessly connect global data centers, clouds, and third-party storage systems without data migration or duplication.
The demo shows how Fusion integrates the NVIDIA Video Search and Summarization (VSS) blueprint, so that as videos are ingested, AI models detect objects, scenes, and events, and create descriptive metadata that makes every clip searchable by context rather than requiring manual tagging. Fusion goes beyond keywords, making it a powerful tool for helping organizations tame their large and growing collections of unstructured data.
IBM is helping enterprises to move from AI pilots to production at scale with NVIDIA. From training foundation models to operationalizing agentic AI applications, IBM Fusion provides the integrated, data-centric foundation that enterprises need to fully leverage the NVIDIA AI Data Platform.