Cisco updated its Nvidia-based AI infrastructure offering with support for workload data fabrics.
Cisco Secure AI Factory with Nvidia now offers integration with AI operating system (OS) vendor Vast Data, specifically with Cisco’s AI points of delivery (PODs).
Cisco AI PODs are included with Vast InsightEngine, Vast’s AI platform and backbone for its Vast Data AI OS, where they can leverage the Nvidia AI Data Platform reference design to grab raw data and accelerate retrieval-augmented generation (RAG) pipelines.
Announced in March of this year at the Nvidia GTC conference, the Cisco AI Factory sees the PODs combine the performance of Cisco Unified Computing System (UCS) servers with Nvidia’s AI-native RTX PRO 6000 Blackwell Server Edition GPUs. This approach uses the power of GPUs and serverless automation to remove the delays that can hinder AI development.
With the infrastructure offering now released, Cisco customers can potentially accelerate data extraction and retrieval for agentic AI models, with agents having data at their disposal at moment of need.
Cisco is reflecting the trend in enterprise deployment of AI agents, models which carry out tasks for organizations in an autonomous manner. The networking and security giant has an established pedigree with the tech, having established the Agntcy project to be a de facto listing of AI agents. The framework was donated to the Linux Foundation this year, with Cisco staying on as a formative member for its development.
Underlining its security focus, a Cisco AI Defense tool in the AI Factory embeds security guardrails to protect against token-level attacks, such as prompt injections, which could otherwise manipulate or compromise an AI model’s output. This further safeguards AI agent deployment, preventing any compromise in their proliferation; that output is expected to be large, with Vijoy Pandey, GM and SVP of Cisco’s Outshift, expecting the equivalent to “80 billion” users on the network.
Cisco, Vast, and Nvidia
Founded in 2016, Vast Data is a non-headquartered company with offices in the U.S, Israel, Asia-Pacifc, and the Middle East. Its partners include Akamai Technologies, Crusoe, GMI Cloud, and Fluidstack.
The firm was reportedly looking to raise capital in August at a $30 billion valuation. Its most recent capital raise was in 2023, which valued the company at $9 billion. At that time, Vast was reporting an annual recurring revenue of around $200 million.
Its Vast InsightEngine, now integrated with Cisco Secure AI Factory with Nvidia, is built on Nvidia’s AI Data Platform reference architecture. The platform uses Nvidia NeMo Retriever and Nvidia NIM microservices to securely connect to and retrieve a company’s confidential data.
Cisco’s AI Factory solution follows its Nexus HyperFabric AI cluster solution, co-developed with Nvidia, which aimed to simplify the deployment and management of AI infrastructure for enterprises using Ethernet networks.
The partnership between the firms reflects Cisco’s addressing of AI’s unique requirements on data center infrastructure.
Denise Lee, VP of Cisco’s engineering sustainability office, told SDxCentral last year that the partnership is “the start of an engineering relationship where we will work together to deliver solutions that help customers automate and optimize fabrics and compute for sustainable AI workloads.”