VAST Data Launches its Comprehensive, Robust Data Platform to Support AI and ML Workflows

VAST Data, the data platform company for the AI era, is unveiling its data computing platform, engineered to underpin the burgeoning world of AI that rivals the capabilities of other existing data platforms. Focused on uniting storage, database, and virtualized compute engine services, the VAST Data Platform delivers a scalable, robust architecture that enables federated deep learning.

With AI being the topic on everyone’s lips—both technical and non-technical professionals alike—the opportunity to innovate technology that successfully propels AI’s success has surfaced. VAST Data, identifying an existing gap between data platforms and their ability to meet the needs of deep learning applications, is introducing its own platform into the mix.

According to VAST, this era of AI can only be supported by rectifying certain data platform trade-offs—such as reducing infrastructure deployment complexity yet lacking in parallel file access or GPU-optimized performance—within a single, easy-to-manage offering.

The VAST Data Platform is designed to mitigate these trade-offs, accommodating the entire spectrum of data—including unstructured and structured data types such as video, image, free text, data streams, and instrument data. Unified in a platform that processes data against a global data corpus in real time, the platform closes the gap between event-driven and data-driven infrastructures, according to the company.

“Our goal is for people to start using [VAST Data] for the machine learning and deep learning applications that they're building in such a way that they don't have to worry about the infrastructure anymore,” explained Andy Pernsteiner, field CTO at VAST Data.

The platform delivers key capabilities that enable enterprises to maintain a comprehensive and robust foundation for their AI applications, including the access and processing of data in any private or major public cloud data center; natural data comprehension via an embedded and queryable semantic data layer; and the continuous and repeatable computation of data in real time.

“I'm excited to see what people build,” said Pernsteiner. “[The platform invites users to rethink] how they look at a data platform, how they look at doing analytics, how they look at doing machine learning, and not having to worry about all of the bits and pieces that fit things together—just being able to run their workflows and run their experiments without having to go and build separate infrastructure for each piece.”

Several technological components combine to create the VAST Data Platform, including:

  • VAST DataStore, a scalable storage architecture for unstructured data that eliminates storage tiering
  • VAST DataBase, a combination of a database, data warehouse, and data lake that limits the tradeoffs between transactions and analytics
  • VAST DataEngine, a global function execution engine that consolidates data centers and cloud regions into one global computational framework
  • VAST DataSpace, a global namespace that permits every location to store, retrieve, and process data from any location with high performance while enforcing strict consistency across every access point

Ultimately, these features amalgamate to form a data platform that addresses the needs of AI applications while simultaneously driving new discoveries and understandings of enterprise AI and LLM systems.

“There's a couple of motivations for us in extending our platform to not only do data storage and access but also to include the ability to do processing directly on it,” explained Pernsteiner. “One motivation is that we want to make sure that the code or the applications which are running on the data are run as close to where the data is as possible…and so leveraging the DataSpace, we have the ability for the DataEngine to be able to consult the DataSpace to identify where data has the most gravity, and then perform analytics or processing or functions in place.”

“Our hope is that people start to recognize the inefficiencies in how they're building and stitching together these pipelines and realize that there is a better way,” continued Pernsteiner. “And the better way starts by having a foundational layer that is able to be extended very easily, to allow for more and more functionality as time goes on.”

All these features, with the exception of the VAST DataEngine, are now generally available within the VAST Data Platform. The VAST DataEngine will be available in 2024.

To learn more about the VAST Data Platform, please visit