VAST Looks Inward, Outward for An AI Edge

There’s no single best way to respond to the explosion of data and AI. Sometimes you need to bring everything into your own unified platform. Other times, you lean on friends and neighbors to chart a way forward. Those are the two seemingly disparate paths that VAST Data chose this week to tackle the burgeoning challenge and opportunity that is turning raw data into an AI-powered advantage.

First, the unified platform story. VAST Data on Tuesday announced a new solution called VAST InsightEngine with Nvidia, which is the first application workflow designed to run atop the VAST Data Platform. As the company explains, the new solution is aimed at providing customers building GenAI apps with a soup-to-nuts platform that “can handle all of the data functions natively to simplify and deliver real-time AI-powered insights at scale.”

With this solution, VAST Data is providing the majority of the components needed for an AI retrieval pipeline–except for the GPU processing, which of course is provided by Nvidia via use of the Nvidia Inference Microservices (NIMs).

The VAST Data Platform here is composed of the Vast DataBase and the Vast DataEngine. As new data is input into the platform, the VAST DataEngine triggers the Nvidia embedding agent via the NIM API to process the data against the trillions of vector and graph embeddings that are stored in the VAST DataBase. The response is then incorporated into a retrieval augmented generation (RAG)-powered GenAI workflow.

VAST says that keeping all structured and unstructured data–including files, objects, tables, and streams–in its platform eliminates the need for additional data lakes, external SaaS platforms, and ETL workloads, thereby speeding up the workload, reducing complexity, and reducing cost.

“By unifying all elements of the AI retrieval pipeline into an enterprise data foundation, VAST Data InsightEngine with NVIDIA is the industry’s first solution to provide a universal view into all of an enterprise’s structured and unstructured data to achieve advanced AI-enabled decision-making,” VAST Data co-founder Jeff Denworth says in a press release.

The approach is part and parcel of the all-in approach that Denworth and his other VAST co-founders, Renen Hallak and Shachar Fienblit, have taken since they started the company back in 2016. VAST’s big goal has always been to rethink distributed systems architecture storage, first with a scale-out unstructured data storage offering called Universal Storage built atop its DASE (Disaggregated and Shared Everything) technology.

VAST InsightEngine

In 2021, VAST started selling storage hardware via the subscription model, but two years later, it pivoted to its current full-stack approach as embodied in the VAST Data Platform, which is composed of DataBase, DataStore, a universal storage offering; DataSpace, a global name space; and DataEngine, its serverless compute engine.

VAST’s hope is to deliver a single, unified system that brings exabyte-level storage together with the accelerated computing needed to build and run AI applications atop that data. The New York City company laid out its ambitious plans to schedule compute “across a globally distributed set of AI supercomputers” in its white paper, “The Rise of the Deep Learning Data Platform.”

While tomorrow’s exabyte-scale AI projects may benefit from a unified architectural approach, VAST Data definitely realizes that it can’t go it alone. To that end, the company yesterday unveiled Cosmos, a new industry group dedicated to advancing AI “by creating a comprehensive and supportive environment to nurture innovation, collaboration, growth, and economic prosperity.”

The company says Cosmos will bolster AI adoption by helping users communicate and share information by providing labs, vendor showcases, and research news. Founding members include VAST Data, NVIDIA, xAI, Supermicro, Deloitte, WWT, Cisco, CoreWeave, Core42, NEA, Impetus, Run:AI, and Dremio.

The Cosmos project marks an important step forward in the journey to define the future of data and AI, said VAST Data Co-Founder and CEO Renen Hallak.

“Since our inception, VAST Data has been dedicated to breaking down the barriers that have long confined the potential of data,” said Hallack, who also is a BigDATAwire Person to Watch for 2024. “With the VAST Data Platform at the center of this comprehensive, interconnected AI ecosystem of technology leaders and AI practitioners, Cosmos will help accelerate discovery, empowering innovation, and enabling the transformation of entire industries.”

VAST Data co-founders reportedly took inspiration from other community groups for Cosmos, including the Open Compute Project and the RSA conference. You can access Cosmos here.

Related Items:

The VAST Potential for Hosting GenAI Workloads, Data

Datanami Unveils 2024 People to Watch

GenAI Show Us What’s Most Important, MinIO Creator Says: Our Data