Research Finder
Find by Keyword
Does Google Cloud Partnership Cement VAST Data as the Switzerland of Hybrid AI Infrastructure?
VAST’s AI Operating System and DataSpace global namespace with Google Cloud creates a unified data foundation for hybrid AI, removing the need for replication or migration
Key Highlights:
VAST Data has introduced a fully managed version of its AI Operating System on Google Cloud.
This extends VAST’s DataSpace namespace to create a unified data fabric that spans on-premises and cloud environments.
The collaboration enables high-performance hybrid AI, providing NVMe-class throughput across global distances, universal permissions via decentralized locking, and flexible synchronization modes tailored to each workload.
The News
VAST Data announced an expanded partnership with Google Cloud to deliver the VAST AI Operating System as a managed service through the Google Cloud Marketplace. The integration connects VAST DataSpace clusters across on-premises and cloud environments, providing a single, secure namespace that supports hybrid AI training and inference.
The solution enables enterprises to run production AI workloads on Google Cloud using existing on-prem data without migrations or delays. VAST said this reduces replication and egress, freeing up budget for AI innovation while supporting high-throughput workloads in areas such as genomics, finance, and life sciences. With TPU-ready performance, DataSpace feeds Google Cloud TPU VMs through optimized NFS paths and metadata-aware I/O for consistent load times and predictable cold-start behavior.
VAST said DataSpace can be deployed in Google Cloud immediately, and that joint validation and reference guidance for establishing a VAST DataSpace spanning Google Cloud and external clusters are available to qualified customers and partners. You can read the full press release here.
Analyst Take
In my opinion, VAST and Google Cloud are turning hybrid AI into a working reality rather than an architectural ideal. By eliminating the dependence on replication or staged transfers, they have created a data fabric that moves with the compute. VAST says this approach tackles both data gravity and compute gravity, two of the hardest challenges in large-scale AI deployment, by ensuring that datasets are instantly accessible wherever capacity or accelerators are available.
The sophistication of DataSpace stands out as a potential game-changer. By decentralizing lock management down to the file, object, and table level, VAST has addressed one of the toughest problems in distributed storage: maintaining data consistency and performance across wide geographic distances. Its combination of lazy, eager and scheduled synchronization modes gives enterprises control over how and when remote sites synchronize, depending on workload type and cost priorities. If the claims are true, this results in a system that delivers local read-write performance on a global scale.
For example, VAST links clusters located thousands of kilometers apart and can still achieve near-local performance on Google Cloud TPUs. Files written to one cluster appear instantly in distant environments, so data can remain in place while compute shifts globally.
I believe this integration further elevates VAST’s position as a hybrid data platform vendor defining the control plane for distributed AI. For Google Cloud, it expands TPU adoption with a differentiated data foundation that serves enterprises seeking hybrid scale without sacrificing governance or performance. Together, they are proving that the next wave of AI infrastructure will be defined less by where data lives and more by how intelligently compute can be applied.
What Was Announced
VAST Data and Google Cloud have integrated the VAST AI Operating System directly with Google Cloud services, enabling organizations to deploy a fully managed, production-grade AI data platform in minutes. The system extends the DataSpace global namespace across on-premises and cloud clusters so that data can be read and written concurrently from multiple locations under a single, consistent policy framework.
The DataSpace fabric is engineered for performance and reliability. It decentralizes lock management, allowing each cluster to control its own file, object, or table-level transactions while maintaining global consistency. The companies said this approach removes the latency typically associated with wide-area locking and provides universal permissions across the entire namespace. In practice, that would mean that datasets appear immediately in remote environments, and GPUs or TPUs can begin accessing them without replication or restaging.
The platform introduces flexible synchronization modes designed to accommodate diverse workloads. In lazy mode, synchronization occurs based on access predictions, minimizing unnecessary transfers. Eager mode ensures immediate caching for latency-sensitive tasks, while scheduled mode integrates with orchestration systems such as RunAI, Slurm, and Kubernetes to align data movement with training or inference cycles. This flexibility allows enterprises to balance cost, bandwidth, and performance according to the needs of each project.
DataSpace also supports advanced replication and resiliency capabilities, including asynchronous Snap-to-Object copies and synchronous replication for zero-RPO requirements. Each participating cluster—whether deployed in the data center, at the edge, or in the public cloud—contributes to the same namespace, creating a unified data layer that scales beyond the capacity of any single location.
By extending this system into Google Cloud, VAST enables organizations to pair high-performance storage with elastic compute resources such as TPUs and GPUs. The result is a hybrid platform that merges the speed and control of on-prem infrastructure with the scalability of the cloud, optimized for the performance, governance, and economics of AI.
Looking Ahead
This collaboration is a major step forward in VAST Data's quest to create one smart data architecture for the AI era. By embedding the VAST AI Operating System and its DataSpace namespace directly into Google Cloud, the company is removing long-standing barriers created by data gravity and geography. The architecture lets companies work as if their entire data estate exists in one place without costly duplication or performance compromises.
VAST’s strategy is built around three guiding principles: hybrid should be seamless, data placement should be a choice rather than a constraint, and performance must scale to meet AI’s most demanding workloads. Through this partnership, I believe the company is transforming hybrid cloud from a logistical exercise into an operational foundation for global AI.
For customers, the collaboration should provide an actionable path toward hybrid AI operations where compute and data coexist without friction. The next step will be validation: testing how synchronization modes perform under real-world conditions, how governance extends across compliance zones, and how cost models shift when data streaming replaces full replication.
For the industry at-large, this announcement signifies a clear move from “cloud-first” to “data-fabric-first.” VAST and Google Cloud are defining a model in which performance, policy, and placement are unified in one architecture. As AI adoption accelerates, this kind of intelligent workload placement will soon be the basis for competitive advantage.
Don Gentile | Analyst-in-Residence -- Storage & Data Resiliency
Don Gentile brings three decades of experience turning complex enterprise technologies into clear, differentiated narratives that drive competitive relevance and market leadership. He has helped shape iconic infrastructure platforms including IBM z16 and z17 mainframes, HPE ProLiant servers, and HPE GreenLake — guiding strategies that connect technology innovation with customer needs and fast-moving market dynamics.
His current focus spans flash storage, storage area networking, hyperconverged infrastructure (HCI), software-defined storage (SDS), hybrid cloud storage, Ceph/open source, cyber resiliency, and emerging models for integrating AI workloads across storage and compute. By applying deep knowledge of infrastructure technologies with proven skills in positioning, content strategy, and thought leadership, Don helps vendors sharpen their story, differentiate their offerings, and achieve stronger competitive standing across business, media, and technical audiences.