IBM and NVIDIA expanded their AI partnership at NVIDIA’s GTC 2026 conference in San Jose, targeting enterprise-scale deployment of artificial intelligence beyond pilot projects.
The collaboration addresses a critical bottleneck in enterprise AI adoption by integrating GPU acceleration into data analytics and document processing workflows. The partnership spans four areas: GPU-native data analytics, intelligent document processing, cloud infrastructure, and consulting services.
A proof-of-concept at Nestlé demonstrated measurable performance gains. The test integrated IBM’s watsonx.data SQL engine, Presto, with NVIDIA’s cuDF library to accelerate queries on Nestlé’s Order-to-Cash data mart. The system processes terabytes of data across 44 tables tracking operations in 186 countries. CPU-based data refreshes previously required 15 minutes and ran only a few times daily. GPU acceleration reduced refresh time to three minutes, delivering 83% cost savings and 30X price-performance improvement, according to the companies.
“Our focus now is on turning this capability into tangible business impact,” said Chris Wright, Nestlé’s Chief Information and Digital Officer.
IBM will offer NVIDIA Blackwell Ultra GPUs on IBM Cloud starting early Q2 2026 for training, inferencing, and AI reasoning workloads. The technology will also deploy across Red Hat AI Factory with NVIDIA and VPC servers with compliance and data residency controls.
NVIDIA selected IBM Storage Scale System 6000 to provide 10 petabytes of high-performance storage for its GPU-native analytics engines, certified on NVIDIA DGX platforms.
The companies are integrating IBM’s open-source Docling tool with NVIDIA Nemotron models for intelligent document processing, converting unstructured documents into AI-ready formats at enterprise scale.
“In the next wave of enterprise AI, the model layer will rely on the data, infrastructure, and orchestration layers,” said IBM Chairman and CEO Arvind Krishna. “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 CEO, said the partnership brings “CUDA GPU acceleration directly into the data layer — turning analytics and document processing from bottlenecks into real-time intelligence engines.”
