IBM (NYSE: IBM) announced an expanded collaboration with NVIDIA at GTC 2026 to help enterprises scale AI operations from pilot programs to production deployment.

The partnership focuses on GPU-native data analytics, document processing, on-premises infrastructure, and consulting services. IBM's watsonx.data SQL engine Presto will be accelerated by NVIDIA cuDF to enable faster query execution on large datasets.

In a production test with Nestlé's Order-to-Cash data mart, which processes terabytes across 44 tables in 186 countries, the GPU-accelerated IBM watsonx.data Presto engine reduced query runtime from 15 minutes to three minutes. Nestlé reported 83% cost savings and a 30X price-performance improvement compared to CPU-based processing.

"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, Chief Information and Digital Officer of Nestlé.

The companies are also developing intelligent document extraction capabilities using IBM's Docling and NVIDIA Nemotron open models to convert unstructured documents into AI-ready formats with source-level traceability.

NVIDIA has selected IBM Storage Scale System 6000 to provide 10PB of high-performance storage for GPU-native analytics engines. The storage system is certified and validated on NVIDIA DGX platforms.

IBM plans to offer NVIDIA Blackwell Ultra GPUs on IBM Cloud in early Q2 2026 for large-scale training and high-throughput inferencing. IBM Consulting will integrate Red Hat AI Factory with NVIDIA through its IBM Consulting Advantage platform to help clients build and deploy AI solutions.

The partnership also explores integration of IBM Sovereign Core with NVIDIA infrastructure for enterprises requiring data residency and regulatory compliance controls.