Snowflake, in partnership with Amazon Web Services (AWS) and NVIDIA, is launching NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs on its AI Data Cloud platform in select AWS regions. This integration brings high-performance visual and generative AI compute directly alongside enterprise data, enabling accelerated full-lifecycle AI development within a secure environment.
The collaboration addresses the traditional fragmentation in enterprise AI workflows, which often require separate environments for training, inference, and serving, leading to complex data transfers and increased security risks. By embedding NVIDIA Blackwell-class GPUs into Snowflake’s architecture, organizations can build and deploy powerful AI models and applications with high-throughput processing while maintaining data governance and security.
Key features of this unified platform include Snowflake Container Runtime for GPU-powered container provisioning, enabling data scientists to perform interactive development and distributed training using frameworks like PyTorch. The Snowflake ML DataConnector supports parallelized reading of unstructured data such as images and PDFs directly from Snowflake stages, optimizing GPU utilization and minimizing I/O bottlenecks.
Deployment flexibility is provided through Snowpark Container Services for real-time inference with millisecond latency and Snowpark-optimized warehouses for large-scale batch processing. The NVIDIA Blackwell architecture also enhances enterprise AI accuracy by accelerating reasoning at the semantic layer, ensuring models understand business logic and data context.
The NVIDIA RTX PRO 6000 Blackwell GPU offers 96 GB of GDDR7 memory, supporting models with over 70 billion parameters and multimodal workflows on a single card. It features fifth-generation Tensor Cores with native FP4 support, delivering up to five times higher inference throughput than previous generations. Alongside the RTX PRO 6000, the RTX PRO 4500 Blackwell Server Edition GPU will also be available soon on Snowflake.
Use cases include insurance providers managing millions of claims annually, where the platform enables zero-copy ingestion of high-resolution images and reports, distributed multimodal training, and unified deployment for fraud detection pipelines. This integration streamlines AI workflows, reduces total cost of ownership, and accelerates time to insight for enterprise AI teams.
Read more: snowflake.com