IBM and Nvidia announced an expanded collaboration at GTC 2026 to help enterprises move AI from pilot to production, with new tools spanning data, infrastructure and cloud.
IBM and Nvidia announced an expanded collaboration on Monday aimed at helping enterprises move artificial intelligence out of the pilot phase and into production, addressing what both companies describe as the most persistent bottleneck in corporate AI adoption: not the models, but the infrastructure beneath them.
The partnership, unveiled at Nvidia’s GTC 2026 conference, advances work across data analytics, document processing, on-premises deployment, regulated industry infrastructure and enterprise consulting. The companies said the collaboration is designed to give organizations the data foundation, hardware and expertise needed to operationalize AI at scale.
“Our partnership with Nvidia goes to the heart of that challenge,” said Arvind Krishna, IBM’s chairman and chief executive. “Together, we’re giving enterprises the solutions they need to stop experimenting with AI and start running on it.”
Jensen Huang, Nvidia’s founder and chief executive, framed the announcement in terms of data rather than models. “Data is the ground truth that gives AI context and meaning,” he said. “Together with IBM, we are bringing GPU acceleration directly into the data layer — turning analytics and document processing from bottlenecks into real-time intelligence engines.”
Also Read: Amazon’s Layoffs Aren’t a Warning. They’re a Preview.
Faster Data, Proven in Production
The centerpiece of the announcement is a collaboration on GPU-accelerated data analytics. IBM’s watsonx.data SQL engine, Presto, has been integrated with Nvidia’s cuDF library to accelerate query execution on large enterprise datasets.
The two companies tested the integration in a live production environment at Nestlé, whose order-to-cash data system tracks every order, delivery and invoice across 186 countries, processing terabytes of data across 44 tables. Previously, a single data refresh took 15 minutes and ran only a handful of times each day. With the GPU-accelerated system, Nestlé reports that query runtime dropped to three minutes — an 83% reduction in cost and a 30-fold improvement in price-performance.
“Working with IBM and Nvidia, a targeted proof of concept has demonstrated the ability to refresh global operations data in a few minutes and at reduced cost,” said Chris Wright, Nestlé’s chief information and digital officer. “Our focus now is on turning this capability into tangible business impact.”
Unlocking Trapped Data
Beyond structured analytics, IBM and Nvidia are addressing a problem most large enterprises know well: vast quantities of useful information locked in unstructured formats — SharePoint sites, vendor research, internal documents — that cannot be easily extracted, standardized or trusted at the speed decisions require.
The two companies are combining IBM’s Docling document extraction tool with Nvidia’s Nemotron open models to make intelligent document processing available at enterprise scale. Docling converts documents into AI-ready formats with source-level traceability; Nemotron models accelerate the ingestion of multimedia content. Early results show significantly higher throughput than comparable open-source approaches, the companies said, while maintaining accuracy where GPU-accelerated infrastructure is available.
Also Read: Not a “Woman Leader.” Just a Leader
Infrastructure for Regulated Industries
IBM and Nvidia are also extending their collaboration to the infrastructure layer. Nvidia has selected IBM’s Storage Scale System 6000 — offering ten petabytes of high-performance storage — to serve its GPU-native analytics engines, pairing IBM’s data access layer with Nvidia’s GPU pipelines.
For enterprises and governments subject to strict data residency and regulatory requirements, the two companies are exploring integrating IBM Sovereign Core with Nvidia infrastructure and Nemotron models to run GPU-intensive AI workloads entirely within regional boundaries without compromising compliance.
Cloud and Consulting
IBM plans to offer Nvidia’s Blackwell Ultra GPUs on IBM Cloud in the second quarter of 2026, supporting large-scale training, high-throughput inference and AI reasoning. The technology will also be integrated into Red Hat AI Factory with Nvidia, and made available to IBM Consulting clients through the company’s Advantage platform — an enterprise AI system designed to help organizations prepare data, build models and deploy AI at scale.
