VAST Data Accelerates Agentic AI at VAST Forward 2026
At VAST Forward 2026, VAST Data expanded the view of its ambitions for its AI operating system. The company dropped five major announcements in a single day: new agentic AI capabilities, a fully accelerated hardware stack built with NVIDIA, a global infrastructure control plane, a video intelligence partnership, and a formalized partner ecosystem. Taken together, they represent a vision of what enterprise AI infrastructure looks like when it stops being a collection of components and starts behaving like a unified, intelligent system.
Building the Thinking Machine: PolicyEngine and TuningEngine
The marquee announcement was the unveiling of two new capabilities of the VAST AI Operating System, VAST Data PolicyEngine and VAST Data TuningEngine. These two new services, slated for release by end of 2026, are designed to work in tandem inside the VAST DataEngine to create what the company calls a “thinking machine” — a system that doesn’t just execute AI pipelines but governs them, evaluates them, and improves on them automatically.
PolicyEngine functions as an inline enforcement layer for agentic workflows, applying fine-grained, tamper-proof controls on what agents can access, what tools they can invoke, and how they communicate with other agents. TuningEngine captures outcomes from those workflows and feeds them into fine-tuning pipelines using methods like LoRA, supervised fine tuning, and reinforcement learning, automatically generating candidate models for evaluation and deployment.
The result is a closed operational loop: observe, act, evaluate, improve. VAST co-founder Jeff Denworth framed it plainly: “Just as people are always learning, so should tomorrow’s applications.” For enterprises trying to deploy AI in regulated or mission-critical environments, the combination of zero-trust governance and automated model improvement in a single platform is a major step toward trusted, autonomous systems.
NVIDIA Collaboration Puts Acceleration Inside the Stack
The software story gets hardware muscle through an expanding collaboration with NVIDIA with an end-to-end, fully CUDA-accelerated AI data stack. The companies introduced CNode-X, a new class of NVIDIA-Certified servers that run the VAST AI Operating System directly on GPU-powered infrastructure. The deeper integration embeds NVIDIA libraries—cuVS for vector search, cuDF-based SQL acceleration via an engine called Sirius, and support for NVIDIA’s Context Memory Storage (CMX) platform—directly into VAST’s core data services.
The goal is to eliminate the fragmented stack problem: separate storage, database, and AI compute tiers that slow enterprise AI pipelines from pilot to production and add operational complexity at every seam. In doing so, it clears the path for agentic AI-enabled workloads to fulfill their promised potential. “CNode-X is CUDA-accelerated at every layer to give AI agents persistent memory so they can work on complex problems over days or weeks, and eventually years, without forgetting—opening the world to the next frontier of AI,” said NVIDIA Founder and CEO Jensen Huang.
CNode-X will come to market through OEM partners, including Cisco and Supermicro, giving enterprises a path to GPU-accelerated VAST infrastructure through vendors they already buy from.
Polaris Solves the Multi-Cloud Management Mess
VAST also announced Polaris, a global control plane purpose-built to provision, operate, and orchestrate distributed AI infrastructure across public cloud, neocloud, and on-premises environments. As AI pipelines span regions and providers between data collection, training, and inference, Polaris offers a centralized service delivery that converts disparate infrastructure instances into one operational environment.
Polaris is built on a Kubernetes-based architecture with a lightweight agent on every VAST node and operates as an intent-driven management layer. Administrators define the desired state of infrastructure, and Polaris coordinates the cloud-native services and VAST software to get there and keep it there. It supports cloud service provider partners, sovereign deployments, and multi-site, multi-cluster configurations under centralized management. It is available as part of VAST cloud deployments.
Video Intelligence Gets a Sovereign Deployment Path with TwelveLabs
VAST’s announced partnership with TwelveLabs, which develops video foundation models, introduced a partnership to help organizations extend video intelligence beyond public cloud deployments. Through the collaboration, the companies will support demand for deploying advanced video intelligence closer to where the data originates and is governed. The pitch is squarely at enterprises and government agencies sitting on massive video archives that can’t or won’t push data to a hyperscaler: media companies, financial services firms running surveillance-based fraud detection, and public sector agencies where data sovereignty is non-negotiable. TwelveLabs gains a deployment path into on-prem and neocloud environments, and VAST gains a compelling vertical use case to anchor its platform story.
Cosmos Gets a Real Partner Program
Finally, VAST formalized its Cosmos partner ecosystem into a unified global partner program, consolidating resellers, system integrators, independent software vendors, cloud providers, and advisory partners under a single framework with structured onboarding, tiered benefits, deal registration, and a centralized partner portal. Cosmos offers a clear engagement model for each partner type, from hardware platform partners validating architectures to consulting firms running deployment practices. H2O.ai and WWT are among the early participants.
The TechArena Take
Today’s announcements represent the most concrete evidence yet that VAST is building out its “thinking machine” vision a coherent, layered way, and is doing so with the right partners. The strategic logic is sound: if AI pipelines are becoming continuous, always-on systems, then the infrastructure layer needs to behave like an operating system—governing, learning, and adapting in real time. PolicyEngine, TuningEngine, and Polaris are all designed to elevate the company’s AI OS position.
Meanwhile, its work with TwelveLabs, the formalization of its partner ecosystem, and the CNode-X collaboration with NVIDIA demonstrate that VAST is assembling a coalition to maximize its reach and impact. Each partnership extends VAST’s reach into a different part of the enterprise buying process, including technical validation, vertical use cases, and channel distribution. Together, they suggest a company that understands the AI OS can’t succeed alone.
At VAST Forward 2026, VAST Data expanded the view of its ambitions for its AI operating system. The company dropped five major announcements in a single day: new agentic AI capabilities, a fully accelerated hardware stack built with NVIDIA, a global infrastructure control plane, a video intelligence partnership, and a formalized partner ecosystem. Taken together, they represent a vision of what enterprise AI infrastructure looks like when it stops being a collection of components and starts behaving like a unified, intelligent system.
Building the Thinking Machine: PolicyEngine and TuningEngine
The marquee announcement was the unveiling of two new capabilities of the VAST AI Operating System, VAST Data PolicyEngine and VAST Data TuningEngine. These two new services, slated for release by end of 2026, are designed to work in tandem inside the VAST DataEngine to create what the company calls a “thinking machine” — a system that doesn’t just execute AI pipelines but governs them, evaluates them, and improves on them automatically.
PolicyEngine functions as an inline enforcement layer for agentic workflows, applying fine-grained, tamper-proof controls on what agents can access, what tools they can invoke, and how they communicate with other agents. TuningEngine captures outcomes from those workflows and feeds them into fine-tuning pipelines using methods like LoRA, supervised fine tuning, and reinforcement learning, automatically generating candidate models for evaluation and deployment.
The result is a closed operational loop: observe, act, evaluate, improve. VAST co-founder Jeff Denworth framed it plainly: “Just as people are always learning, so should tomorrow’s applications.” For enterprises trying to deploy AI in regulated or mission-critical environments, the combination of zero-trust governance and automated model improvement in a single platform is a major step toward trusted, autonomous systems.
NVIDIA Collaboration Puts Acceleration Inside the Stack
The software story gets hardware muscle through an expanding collaboration with NVIDIA with an end-to-end, fully CUDA-accelerated AI data stack. The companies introduced CNode-X, a new class of NVIDIA-Certified servers that run the VAST AI Operating System directly on GPU-powered infrastructure. The deeper integration embeds NVIDIA libraries—cuVS for vector search, cuDF-based SQL acceleration via an engine called Sirius, and support for NVIDIA’s Context Memory Storage (CMX) platform—directly into VAST’s core data services.
The goal is to eliminate the fragmented stack problem: separate storage, database, and AI compute tiers that slow enterprise AI pipelines from pilot to production and add operational complexity at every seam. In doing so, it clears the path for agentic AI-enabled workloads to fulfill their promised potential. “CNode-X is CUDA-accelerated at every layer to give AI agents persistent memory so they can work on complex problems over days or weeks, and eventually years, without forgetting—opening the world to the next frontier of AI,” said NVIDIA Founder and CEO Jensen Huang.
CNode-X will come to market through OEM partners, including Cisco and Supermicro, giving enterprises a path to GPU-accelerated VAST infrastructure through vendors they already buy from.
Polaris Solves the Multi-Cloud Management Mess
VAST also announced Polaris, a global control plane purpose-built to provision, operate, and orchestrate distributed AI infrastructure across public cloud, neocloud, and on-premises environments. As AI pipelines span regions and providers between data collection, training, and inference, Polaris offers a centralized service delivery that converts disparate infrastructure instances into one operational environment.
Polaris is built on a Kubernetes-based architecture with a lightweight agent on every VAST node and operates as an intent-driven management layer. Administrators define the desired state of infrastructure, and Polaris coordinates the cloud-native services and VAST software to get there and keep it there. It supports cloud service provider partners, sovereign deployments, and multi-site, multi-cluster configurations under centralized management. It is available as part of VAST cloud deployments.
Video Intelligence Gets a Sovereign Deployment Path with TwelveLabs
VAST’s announced partnership with TwelveLabs, which develops video foundation models, introduced a partnership to help organizations extend video intelligence beyond public cloud deployments. Through the collaboration, the companies will support demand for deploying advanced video intelligence closer to where the data originates and is governed. The pitch is squarely at enterprises and government agencies sitting on massive video archives that can’t or won’t push data to a hyperscaler: media companies, financial services firms running surveillance-based fraud detection, and public sector agencies where data sovereignty is non-negotiable. TwelveLabs gains a deployment path into on-prem and neocloud environments, and VAST gains a compelling vertical use case to anchor its platform story.
Cosmos Gets a Real Partner Program
Finally, VAST formalized its Cosmos partner ecosystem into a unified global partner program, consolidating resellers, system integrators, independent software vendors, cloud providers, and advisory partners under a single framework with structured onboarding, tiered benefits, deal registration, and a centralized partner portal. Cosmos offers a clear engagement model for each partner type, from hardware platform partners validating architectures to consulting firms running deployment practices. H2O.ai and WWT are among the early participants.
The TechArena Take
Today’s announcements represent the most concrete evidence yet that VAST is building out its “thinking machine” vision a coherent, layered way, and is doing so with the right partners. The strategic logic is sound: if AI pipelines are becoming continuous, always-on systems, then the infrastructure layer needs to behave like an operating system—governing, learning, and adapting in real time. PolicyEngine, TuningEngine, and Polaris are all designed to elevate the company’s AI OS position.
Meanwhile, its work with TwelveLabs, the formalization of its partner ecosystem, and the CNode-X collaboration with NVIDIA demonstrate that VAST is assembling a coalition to maximize its reach and impact. Each partnership extends VAST’s reach into a different part of the enterprise buying process, including technical validation, vertical use cases, and channel distribution. Together, they suggest a company that understands the AI OS can’t succeed alone.
.png)
