Research Finder
Find by Keyword
VAST Data and Microsoft Align on a Shared Vision for Agentic AI at Global Scale
Azure customers will gain access to the VAST AI Operating System, creating a unified data and execution environment for agentic workloads across hybrid and multi-cloud architectures.
19/11/2025
Key Highlights:
Azure customers will gain access to the VAST AI OS in the Microsoft Marketplace with Azure-integrated governance, identity, and billing.
VAST brings InsightEngine and AgentEngine to Azure, enabling high-performance vector search, RAG pipelines, and autonomous agent workflows across hybrid environments.
VAST DataSpace provides a global namespace that lets customers move from on-premises to Azure GPU infrastructure without data migration or reconfiguration.
Azure GPU and CPU clusters, including those powered by the Laos VM Series and Azure Boost, will benefit from VAST intelligent caching and metadata-optimized I/O.
VAST’s disaggregated shared-everything (DASE) architecture enables independent scaling of compute and storage, improving efficiency for large-scale AI estates.
The collaboration aligns VAST with Microsoft’s roadmap for custom silicon, heterogeneous compute, and future AI systems that require consistent, high-throughput data supply.
The News
At Microsoft Ignite, VAST Data and Microsoft announced a collaboration that will make the VAST AI Operating System available natively to Azure customers. The integration positions VAST as a service that can be deployed, billed, and governed through the Azure platform.
Microsoft and VAST are framing this as a foundational step toward enabling agentic AI at global scale by pairing VAST’s unified data and execution environment with Azure’s elasticity and reach. Both companies signaled alignment on future AI systems, including Microsoft’s custom silicon initiatives and heterogeneous compute strategy.
Analyst Take
This collaboration is a significant win for Azure customers, offering a consistent, high-performance path for advanced AI workloads without the need to re-architect the data layer or manage complex hybrid environments. Azure AI Foundry serves as the enterprise entry point for model builders and agentic teams. VAST complements this direction, supplying a unified data and execution foundation that operates with low friction across regions. Elevating VAST also materially simplifies how customers procure, deploy, and support this critical foundation.
I view this announcement as further insight into how Microsoft is approaching the next generation of AI infrastructure. As Azure is about to celebrate its sixteenth year, Microsoft is investing heavily in custom silicon, high-density GPU clusters, and heterogeneous compute architectures that will carry its customers and AI workloads into the future. These investments can achieve full value when they are consistently supplied with high-throughput data and real-time context. An AI operating system such as VAST that can keep any processor or accelerator fully utilized becomes an essential part of that vision.
From the perspective of model builders, this collaboration carries particular weight. Consider that the Laos VM Series offers tens of thousands of IOPS, hundreds of gigabits per second of network bandwidth, and high local NVMe storage per vCPU. Meanwhile, Azure Boost offloads storage and networking operations onto dedicated hardware and FPGA systems so guest VMs can focus fully on compute tasks. When those capabilities are paired with VAST’s intelligent caching, metadata-optimized I/O, and parallel data services, the result is a platform that can keep GPU and CPU clusters saturated from pilot scale through multi-region production.
It is easy to imagine this partnership extending beyond a straightforward Azure-native deployment into a co-designed platform layer. VAST can help shape how future Azure systems move and transform data, with software and hardware tuned together for vector search, RAG pipelines, and agentic workloads. Microsoft can provide telemetry, placement controls, and scheduling hooks that allow the VAST AI OS to adapt dynamically to each new chip generation. The result could be an AI stack where data pipelines are optimized at both the software and silicon levels.
There is also room for tighter feedback loops between Azure AI Foundry and VAST AI OS. Foundry could remain the unified surface for designing agents and applications, while VAST continually optimizes how data is placed, cached, and streamed to the appropriate compute layer. As new processors arrive within Azure, those agents would not need to change. The underlying VAST fabric would learn how to maximize each platform’s strengths, whether that involves high-bandwidth memory, specialized inference units, or new forms of near-data processing.
The long-term customer benefit is not limited to performance. A jointly engineered platform could provide consistent guarantees for governance, lineage, and resilience as hardware evolves. That stability matters for organizations planning for multiple generations of AI infrastructure while navigating growing regulatory expectations. An AI operating system that abstracts the complexity of changing silicon while preserving policy and audit behavior over time will appeal to enterprises that want predictable modernization paths.
What Was Announced
Microsoft and VAST introduced a fully Azure-native implementation of the VAST AI OS that customers can purchase and operate in the Microsoft commercial marketplace. This elevates VAST from a partner-integrated solution to a cloud-operated offering that appears directly in the Azure portal and aligns with Microsoft billing, identity, governance, and support models. Customers will be able to consume VAST AI OS using their existing Azure agreements, consolidate reporting under Microsoft cloud spend, and manage the platform with familiar tools.
The Azure deployment will include VAST’s InsightEngine for stateless high-performance compute, vector acceleration, RAG pipelines, and data preparation. AgentEngine will provide the runtime for autonomous agents operating on real-time data. These services will run directly on Azure infrastructure and benefit from Azure Boost, Laos VM Series networking, and the performance characteristics required for large-scale GPU and CPU clusters.
VAST’s DataSpace provides the unifying layer across hybrid and multi-cloud environments. Customers will gain a global namespace that spans on-premises systems, Azure regions, and external clouds without any changes to application logic. This enables GPU burst-to-cloud patterns, multi-region training, and hybrid RAG pipelines without ETL, replication, or dataset reshaping. The VAST DataStore supports file, object, and block protocols in one environment, while the VAST DataBase combines transactional performance, warehouse speed, and lake economics.
The platform’s DASE architecture allows compute and storage to scale independently within Azure, giving customers precise control over cost and performance. Similarity Reduction further reduces footprint for data-intensive AI workloads. The combination aligns with Microsoft’s goal of providing customers with efficient and predictable ways to scale model training, inference, and agentic systems.
Microsoft emphasized that many advanced model builders already standardize on VAST for throughput and scale. VAST will extend those capabilities into Azure’s operating model and can align more closely with Microsoft’s silicon roadmap. This creates a path for pairing future Azure processors, GPUs, and accelerators with VAST data and execution services as a unified AI stack that minimizes integration overhead and accelerates deployment.
Looking Ahead
From my perspective, Microsoft’s acceleration toward custom silicon, expanded GPU infrastructure, and heterogeneous compute creates an opening for VAST to become a strategic data substrate that helps maximize the value of these platforms. Several areas stand out as this collaboration deepens.
As previously stated, Microsoft’s custom processors will require predictable, high-throughput data pipelines that keep diverse compute types saturated. VAST’s metadata-optimized I/O and stateless compute services can help ensure that future Azure processors operate at full utilization. As Azure introduces new chip families across CPUs, GPUs, NPUs, and Microsoft-designed accelerators, these systems will need a unifying data environment that supports consistent agent behavior and model execution. VAST’s DataSpace can deliver that continuity across regions and environments.
Agentic workflows, large-scale RAG systems, and continuous learning pipelines will demand a hybrid data plane that behaves the same from edge to core to cloud. The combination of Azure AI Foundry and the VAST AI OS gives customers a clear path to production for these scenarios while reducing friction and operational overhead.
I also see room for tighter integration between VAST and Microsoft on solution patterns for regulated industries, marketplace offerings, and emerging compute services. Both companies have an opportunity to define a blueprint for cloud-scale agentic AI that spans data, execution, governance, and silicon-aware optimization.
From my perspective, this announcement places VAST within Microsoft’s broader AI computing strategy in a very meaningful way. The next phases of this collaboration will be measured by how deeply the two companies align on silicon-informed optimizations, reference architectures, and solution patterns for data-intensive and regulated industries. I will be watching how this partnership shapes the next decade of AI advancement on Azure.
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.