VAST Data The Google Cloud AI partnership aims to solve one of the biggest challenges for companies working with advanced AI environments. Organizations want to run AI where performance is best, but data is often spread across multiple locations. This creates long migration times, high costs, and complex management. Through a common AI operating system, VAST creates Data and Google Cloud a flexible and powerful hybrid cloud solution that makes data quickly available where AI workloads run.

A fully managed AI operating system in Google Cloud

The expanded partnership enables enterprises to use the VAST AI operating system as a fully managed service on Google Cloud Marketplace. Through VAST DataSpace, organizations can link both on-premises resources and Google Cloud into a global namespace. This eliminates the need for time-consuming migrations and enables data can be used directly by AI workflows without being duplicated or moved.

VAST showcased this capability in a demonstration where clusters in USA and Japan were connected over more than ten thousand kilometers. Inference could be run on TPU resources in the US and GPU resources in Japan without copying datasets between environments. The result was near-real-time access to the same dataset in both locations.

Advertisement

AI where it performs best through intelligent data placement

Companies that build extensive AI systems often have to deal with very large data sets. Migrations take time, cost money, and create unnecessary bottlenecks. VAST Data and Google Cloud aims to solve this by making data placement an optional step rather than a requirement. Through intelligent streaming, only the parts of a dataset that the model needs are sent, saving time, reducing costs, and enabling faster AI operations.

Jeff Denworth, co-founder of VAST Data, says customers can launch a full-scale AI OS in minutes. The Marketplace integrates governance, scaling, and payment into a simple process that is suitable for hybrid cloud deployments.

Seamless performance with Google Cloud TPU resources

Tests show that the VAST AI operating system delivers model startup and data access with speeds that are close to local NVMe performance. This was also true during cold starts when the system was tested with Meta's Llama 3.1 8B Instruct model. The combination of VAST Data and Google Cloud This means that AI accelerators can be kept constantly active.

Subramanian Kartik at VAST emphasizes that the system acts as a performance engine that keeps AI pipelines moving and eliminates bottlenecks. More time is spent running models and less time is spent waiting for data.

The benefits of VAST Data Google Cloud AI in hybrid environments

Organizations that want to build modern AI systems gain access to several key benefits:

Implement AI in minutes

Companies can start running production systems in Google Cloud against existing on-premises datasets without planning migrations or delays. A global namespace ensures that data looks the same no matter where it resides.

Reduce data migration costs

Only the data volumes needed for the model are streamed. This reduces network costs, shortens lead times and allows companies are investing more resources in AI development.

Flexible data placement in Google Cloud

Organizations can choose what to keep on-premises, replicate, cache, or migrate to Google Cloud. Unified access controls provide consistent governance across all environments.

TPU optimized data access

VAST delivers validated paths that feed Google Cloud TPU instances with models and datasets without bottlenecks. Warm starts occur with near NVMe performance.

A unified platform for the entire data lifecycle

The VAST AI operating system brings together DataStore, DataBase, InsightEngine, AgentEngine and DataSpace in a single platform. This makes it easy to work with preparation, training, inference and analysis without separate systems.

Available today for Google Cloud customers

VAST can be implemented directly in Google Cloud. Common validation and reference architectures are already available for qualified organizations looking to create a cohesive data space across hybrid clouds. The VAST Data Google Cloud AI partnership clearly demonstrates how to build next-generation AI environments. They are flexible, fast, scalable, and free from constraints on where data resides.