Hammerspace Announces AI Data Platform Innovation Solution Partnership with NVIDIA

This report was published by SiliconANGLE and covers data platform company Hammerspace’s launch of an AI data platform at Nvidia’s Global Technology Conference. The platform aims to deliver high-speed distributed data to AI systems without large-scale data migration. Built on Nvidia’s reference architecture, Hammerspace’s platform helps enterprises automate data preparation, enabling actual AI project operations. The solution simplifies complex data management processes required for AI model training by facilitating automatic data discovery, preparation, and transfer to GPUs.

Hammerspace’s AI data platform focuses on reducing the cost and time associated with manual data preparation for enterprises with data scattered across storage systems or the cloud. The platform automatically identifies data locations and moves only the data needed for AI processing to GPU resources, avoiding additional purchases of flash storage.

The system monitors changes in enterprise storage environments, automatically prepares new files for AI pipelines, and uses a model context protocol server to coordinate moving only the necessary data. Designed to eliminate bottlenecks in AI project operations, it allows data to be processed immediately after generation. While maintaining enterprise security controls and file permissions, it automates data discovery and transfer through a unified metadata layer. This is also part of efforts to address the issue of GPU utilization rates below 30%.

Hammerspace’s solution is described as an automated, unified AI data pipeline, enabling users to more easily find and use data through continuous pipeline updates.