Databricks and NVIDIA: Powering the Next Generation of Industry AI

Industry Use Case Transformation with AI

As we head to Las Vegas for Amazon Web Services (AWS) re:Invent, one trend is unmistakable: enterprises are moving beyond generic GenAI pilots and are now building domain-specific, production-ready AI systems that demand both high-performance computing and deep industry expertise.

Together, Databricks and NVIDIA are enabling this shift. By combining the Databricks Data Intelligence Platform with NVIDIA accelerated computing and AI software stack, customers can solve their most complex challenges—from clinical research and drug discovery to global logistics and manufacturing.

While this joint platform powers solutions across nearly every vertical—including real-time fraud detection and personalized media recommendations—three areas are seeing breakthrough momentum today:

1. Medical Imaging
2. Drug Discovery and Life Sciences R&D
3. Route Optimization and Supply Chain AI

By running NVIDIA SDKs, frameworks, and CUDA-X libraries directly within Databricks on AWS, enterprises can keep sensitive data securely inside their AWS environment while leveraging state-of-the-art GPU acceleration.

Advancing Medical Imaging with Databricks Pixels and NVIDIA MONAI

Healthcare organizations face an enormous data challenge: nearly 97% of medical data is unstructured, with imaging locked inside proprietary formats such as DICOM. Radiologists often struggle to index, query, and prepare these datasets for AI pipelines.

Databricks Pixels solves this by ingesting millions of DICOM files directly into Delta Lake, extracting metadata for fast querying while managing pixel data natively. NVIDIA MONAI, the open-source, GPU-accelerated medical imaging framework, brings advanced AI capabilities directly to this curated data.

Together, organizations can build efficient workflows for:

• 3D image segmentation
• Lesion and anomaly detection
• Automated organ labeling and classification
• Multi-modal imaging analytics

Running MONAI on Databricks enables:

• Clinically-aligned automated workflows
• Faster diagnosis support
• Stronger compliance with healthcare data governance requirements

Learn more at re:Invent at NVIDIA’s Booth #1022—Wednesday, December 3, at 10:30AM. Or start building with the Pixels GitHub repo. 

Accelerating Drug Discovery with Genesis Workbench and NVIDIA BioNeMo

Modern drug discovery requires processing massive biological datasets—protein structures, molecular interactions, genomic profiles—and running iteratively over them at scale. This can take years and billions in R&D investment. Generative AI is transforming this pipeline, enabling researchers to model protein structures, design novel molecules, and analyze cell behavior with unprecedented speed.

Genesis Workbench, Databricks open-source Solution Accelerator, makes advanced biological AI accessible with strong data governance and simplified deployment. 

Combined with NVIDIA accelerated computing on Databricks Serverless GPU Compute, researchers can seamlessly integrate:

• NVIDIA BioNeMo Framework for generative chemistry and protein modeling
• RAPIDS single-cell for scalable single-cell transcriptomics
• Parabricks for GPU-accelerated genomic workflows 

This unified platform allows researchers to:

• Fine-tune and deploy domain-specific generative models
• Conduct virtual compound screening at scale
• Reduce time-to-insight for therapeutic discovery
• Accelerate R&D cycles across protein science, genomics, and cell biology

See it live at re:Invent at Databricks Booth #1420— Wednesday, December 3 at 10:00AM. And start building now, using Genesis Workbench on GitHub. 

Solving Complex Logistics with GPU-Accelerated Route Optimization

Manufacturing, retail, and logistics organizations face one of the hardest mathematical problems in operations: the Vehicle Routing Problem (VRP). On CPUs, large real-world VRP workloads can take hours to compute, often requiring manual pre-clustering that limits solution quality.

With Databricks Serverless GPUs and NVIDIA cuOpt, organizations can now run routing optimization at massive scale and in real-time. NVIDIA cuOpt is a GPU-accelerated optimization engine capable of solving the largest routing workloads with:

• Faster solve time
• Higher-quality routes
• Lower operating costs
• Dynamic re-routing in seconds

By feeding real-time fleet positions, package destinations, traffic, and weather data from Delta Lake into cuOpt, enterprises can:

• Reduce fuel consumption
• Improve delivery window accuracy
• Optimize thousands of routes simultaneously
• Respond instantly to real-world disruptions

 Start building today with Route Optimization on GitHub. 

Join Databricks and NVIDIA at AWS re:Invent

These use cases are just the beginning. If you are looking to build production-grade industry use cases, we invite you to explore what’s possible.

Meet us in person at re:Invent:

• Databricks Booth #1420
• NVIDIA Booth #1022

Attend our sessions:

• Genesis Workbench x BioNeMo: Databricks Booth #1420—Dec 3, starting at 10:00AM.
• Pixels x MONAI:  NVIDIA Booth #1022—Dec 3, starting at 10:30AM.

Reserve your spot and join the party:

• Tuesday, Dec 2, 2025 - 7:00 PM - 10:00 PM PST | Grand Lux Cafe, The Venetian