Hammerspace Launches AI Data Platform Built on NVIDIA Reference Design

Announced at NVIDIA GTC 2026 in March 2026, Hammerspace introduced its new AI Data Platform (AIDP), a turnkey architecture built on NVIDIA’s AI Data Platform reference design and aimed at addressing one of enterprise AI’s largest operational barriers: seamless access to distributed enterprise datasets.

The Shift From AI Pilots to Production

As enterprises move from AI experimentation toward production deployments, one of the industry’s biggest infrastructure challenges is no longer simply compute availability. Increasingly, the bottleneck centers on access to distributed enterprise datasets and the complexity of preparing those datasets for AI applications.

Hammerspace positions the platform as a way to operationalize enterprise AI without forcing organizations into large-scale data migrations or isolated AI storage silos.

According to the company, the platform allows enterprises to begin making existing data AI-ready using infrastructure already deployed across their environments, eliminating the need to purchase large new pools of flash storage solely for AI workloads.

“Hammerspace is the only AI Data Platform that can access data anywhere across edge devices, data centers and clouds, across high-performance file and object storage, without forcing enterprises into a copy-first AI silo,” said David Flynn, Founder and CEO of Hammerspace. “We overcome data gravity by continuously identifying the data that matters, orchestrating it efficiently to GPUs, and enabling processing where it’s most optimal, whether that’s local GPU resources near the data or centralized GPUs at scale.”

Solving the Primary Blockers to Enterprise AI Success

As enterprises scale AI initiatives beyond pilot environments, Hammerspace argues that the largest barriers are increasingly operational rather than computational. The company frames fragmented enterprise data, excessive data movement, and complex orchestration workflows as the primary obstacles preventing organizations from moving AI deployments into production at scale.

Eliminate Data Fragmentation

Hammerspace argues that one of the largest operational problems facing enterprise AI deployments is fragmented data infrastructure.

The company says enterprises often repeat the same processes — locating relevant data, enriching metadata, organizing datasets, and preparing information for AI models and agents — across disconnected teams and storage systems because the underlying data estate lacks unification.

The AI Data Platform is designed to provide a unified operational view across heterogeneous storage environments while automating the pipeline that transforms distributed enterprise data into AI-ready data streams.

Skip Costly Mass Migrations

A major architectural element of the platform is Hammerspace’s “data-in-place” approach.

Rather than requiring enterprises to consolidate data into a newly deployed AI storage environment, the platform is designed to make distributed enterprise data immediately accessible for AI workflows without large-scale migrations or copy-first data pipelines.

The company says this approach is intended to reduce operational overhead while accelerating time-to-value for enterprise AI initiatives.

Reduce Data Copies

Hammerspace also positions the platform as a way to reduce unnecessary data movement inside AI workflows.

The architecture continuously catalogs distributed data in place and uses the company’s Model Context Protocol (MCP) server to coordinate with NVIDIA and other AI applications so that only required data moves when necessary. Hammerspace says policy-driven automation governs data placement, synchronization, security, compliance, and performance across the end-to-end pipeline.

The company argues that this operational model enables enterprises to scale pilot AI deployments into production environments with greater simplicity and governance consistency.

Delivered and Validated by SHI

Hammerspace said SHI International played a significant role in the development and testing of the AI Data Platform through the company’s AI and Cyber Lab environments.

“AI data preparation shouldn’t require a costly rebuild of the data estate,” said Jack Hogan, Vice President of Advanced Solutions at SHI. “As a key development and testing partner, we used SHI’s labs to validate that the Hammerspace AI Data Platform on Cisco UCS, with its logical visibility to distributed data, can drastically reduce data complexity.”

According to SHI, the platform can be integrated into existing enterprise architectures while maintaining compatibility with preferred infrastructure and security controls.

Simple to Deploy, Turnkey Solution for Enterprise-Wide Inference and Agentic AI

Hammerspace says the platform is designed to reduce the operational sprawl surrounding enterprise AI infrastructure by consolidating data orchestration, governance, and movement into a unified environment. The company positions the architecture as a turnkey approach for organizations seeking to operationalize inference, RAG, and agentic AI workloads without assembling large stacks of disconnected tools and workflows.

All-in-One Orchestration

Hammerspace says the platform collapses as many as 15 disconnected tools used for data discovery, cataloging, classification, policy management, and data movement into a unified orchestration layer.

The company also describes the environment as a fully agentic data foundation capable of dynamically managing data placement and movement based on real-time demand.

NVIDIA Partnership

The AI Data Platform is built on NVIDIA’s reference architecture and supports NVIDIA RTX PRO 6000 and RTX PRO 4500 Blackwell Server Edition GPUs.

The platform integrates NVIDIA AI Enterprise software, including NIM microservices and NeMo Retriever, to combine data management and orchestration across heterogeneous storage systems. Hammerspace says the resulting environment is designed to simplify and automate data pipelines for inference, retrieval-augmented generation (RAG), and agentic AI deployments.

“Enterprises require a unified data foundation capable of overcoming the friction of data gravity to power today’s complex, distributed AI pipelines,” said Jason Hardy, Vice President of Storage Technologies at NVIDIA. “Built on the NVIDIA AI Data Platform, Hammerspace enables organizations to seamlessly scale from early experimentation to high-performance, production-grade AI.”

Secuvy DSPM Integration

The platform also integrates with Secuvy Data Security Posture Management (DSPM) technology to provide continuous security monitoring, compliance, and governance throughout the AI data pipeline.

“For a true end-to-end AI solution, security and governance must be native to the data platform,” said Mike Seashols, CEO of Secuvy.

Hardware Platform Flexibility

Hammerspace says its software-defined architecture supports deployment across a broad ecosystem of infrastructure vendors, including Cisco, Lenovo, and Supermicro hardware platforms. The company says the environment is designed to integrate into existing server infrastructures while allowing enterprises to maintain flexibility around infrastructure selection.

Availability

Hammerspace introduced the AI Data Platform during NVIDIA GTC 2026 and said the solution is now generally available.