IBM (NYSE: IBM) has announced new collaborations with NVIDIA (NASDAQ: NVDA), which include future integrations based on NVIDIA’s AI data platform reference design. This initiative aims to assist businesses in leveraging their data more effectively to develop, scale, and manage generative AI workloads and agentic AI applications. As part of this announcement, IBM will launch a new content-aware storage (CAS) capability for its hybrid IBM Fusion solution, plans to expand integrations with watsonx, and introduces new consulting services focused on agentic AI in partnership with NVIDIA.
IBM Drives Performance of Unstructured Data for AI
One of the most significant announcements is the introduction of content-aware storage (CAS), which will allow businesses to extract meaningful information from their growing volumes of unstructured data for inference processes without compromising security or trust. This functionality is key for applications like retrieval-augmented generation (RAG) and agentic AI.
IBM Storage Scale will be able to respond to queries using extracted and enriched data, accelerating communication between GPUs and storage through NVIDIA BlueField-3 DPUs and NVIDIA Spectrum-X networking technology. Additionally, workflows for data extraction from multimodal documents will integrate NVIDIA NeMo Retriever microservices developed on NVIDIA NIM. IBM will incorporate this functionality in the upcoming IBM Fusion update, scheduled for the second quarter of 2025.
Advanced Integrations with watsonx to Facilitate AI Adoption
IBM also plans to integrate its watsonx platform with NVIDIA NIM, enabling organizations to utilize leading AI models across multiple cloud environments. This integration will allow developers to work with watsonx.ai to build and deploy AI models within their applications, using externally hosted models while ensuring robust monitoring through watsonx.governance. This interoperability is particularly relevant for organizations looking to implement agentic AI and advanced applications that require connectivity between models.
IBM Cloud Expands Its Accelerated Computing Offering with NVIDIA H200
With the growing adoption of generative AI and high-performance computing (HPC), IBM Cloud has announced the availability of instances based on the NVIDIA H200 Tensor Core GPU. These instances, featuring high memory capacity and elevated bandwidth, are designed to meet the demands of large foundational models and modern AI applications.
Consulting for Intelligent Processes with Agentic AI and NVIDIA
IBM Consulting introduces its new AI integration services, designed to help businesses transform and govern end-to-end business processes utilizing agentic AI and NVIDIA Blueprints. Use cases include autonomous inspection and maintenance in the manufacturing industry, and proactive video analysis and anomaly response in the energy sector.
Furthermore, IBM Consulting assists its clients in building, modernizing, and managing high-demand AI workloads across hybrid environments, leveraging Red Hat OpenShift and key NVIDIA technologies such as NVIDIA AI Foundry, NeMo, AI Enterprise, Blueprints, and NVIDIA Clara. All with a strong focus on AI governance, data security, and regulatory compliance.
A Major Step Towards AI Monetization
Hillery Hunter, CTO and General Manager of Innovation at IBM Infrastructure, stated: “IBM is focused on helping businesses build effective AI models and scale them rapidly. Together with NVIDIA, we offer the solutions and services needed to unlock, accelerate, and protect data, helping organizations overcome hidden costs and technical challenges of AI to achieve real business outcomes.”
On his part, Rob Davis, Vice President of Storage Networking Technology at NVIDIA, emphasized: “AI agents need to access, retrieve, and process data quickly and at scale. The integration of IBM’s content-aware storage with NVIDIA’s AI platform breaks down data and computing silos, providing an intelligent and scalable system that allows for near-real-time inference for agile reasoning.”