Cisco says the integrated package will enable fast data extraction and retrieval to unlock agentic AI use cases for enterprises.
Cisco is using its Nvidia partnership in collaboration with VAST Data to offer customers a turnkey AI infrastructure package including compute, network, storage and data. The combined package will offer customers a blueprint for building pre-integrated AI infrastructure to support AI workload data fabrics as well as large-scale training and inference, according to the vendors.
The new offering is built around Cisco Secure AI Factory with Nvidia, which brings together Cisco security and networking technology, Nvidia DPUs, and storage options. The joint Cisco-Nvidia AI infrastructure integrates Cisco’s Hypershield and AI Defense packages to help protect the development, deployment, and use of AI models and applications. Nvidia BlueField-3 DPUs and SuperNICs are also featured. SuperNICs are Nvidia’s new class of network accelerators designed to supercharge hyperscale AI workloads in Ethernet-based clouds. The Nvidia AI Enterprise software platform, which features pretrained models and development tools for production-ready AI, is also part of the Secure AI Factory package.
For the new collaboration, VAST brings its core product, InsightEngine, to provide a data intelligence layer that plugs directly into Cisco Secure AI Factory. InsightEngine scans, organizes, and builds a catalog of everything stored in the VAST data platform, such as files, objects, tables. Customers can search and query that data instantly, according to VAST.
“The VAST Data InsightEngine automates AI-ready data pipelines from the moment new data is ingested. It provides real-time vectorization and automated inferencing, turning raw, unstructured data into operational intelligence without complexity. It’s the engine that powers the data pipelines that fuel the Cisco Secure AI Factory,” Chalon Duncan of VAST technical alliances marketing wrote in a blog post about the news. “[It’s] built on the Nvidia AI Data Platform design that leverages GPU-accelerated I/O and serverless automation to eliminate the latency and friction that have traditionally slowed AI adoption.”
VAST’s InsightEngine includes Nvidia NeMo Retriever and Nvidia NIM microservices that connect to proprietary data for secure, enterprise-grade retrieval. (VAST also has an alliance with Nvidia to integrate its products.) These are optimized containers for AI models, and InsightEngine is designed to embed and manage them natively. This approach allows AI practitioners to easily consume these models and run their RAG pipelines directly on the VAST AI OS. InsightEngine manages the lifecycle, deployment, and autoscaling of these models, delivering a new level of efficiency for AI applications, Duncan wrote.
With VAST’s technology, the idea is to simplify infrastructure complexity by converging structured, unstructured, and vector data management, enabling real-time reasoning and workflow automation at scale, Duncan wrote.
The integrated package will be offered in three versions under the Cisco Secure AI Factory with Nvidia banner:
- VAST on Cisco UCS: This foundational building block provides a scalable unified data store fully tested for compatibility through the Cisco SolutionsPlus program.
- VAST on Cisco AI PODs: This version offers pre-validated full infrastructure stack for AI and simplified ordering that improves data scientists’ and developers’ productivity. (Cisco AI Pods are preconfigured and optimized designs aimed at streamlining deployment of Cisco UCS servers, networking, and Nvidia AI Enterprise.)
- VAST InsightEngine on Cisco AI PODs: A turnkey AI Data Platform for enterprise RAG enabling Nvidia NIMs as a Service for simplified, efficient AI workflows.
With the new offering, Cisco says customers will gain fast data extraction and retrieval to unlock agentic AI use cases by reducing RAG pipeline latency from minutes to seconds for near-real-time AI responses.
Customers can experience “agentic AI at enterprise scale by enabling AI agents to operate continuously, learn dynamically and deliver contextualized business outcomes. The high throughput of data unlocks multi-step reasoning, and the architecture is designed for scale by supporting multiple agents and workloads simultaneously,” Cisco stated. In addition, the package offers role-based access control and compliance and audit readiness to secure enterprise information.
Cisco AI PODs with VAST InsightEngine, offering an Nvidia AI Data Platform solution, can be ordered from Cisco now. The AI POD designed for RAG acceleration with Nvidia and VAST is the first in a series of AI services PODs built to support the growing number of use cases in the enterprise, Cisco noted.
