Key Takeaways
- Xage Security now supports NVIDIA DOCA security enhancements and the new NVIDIA Vera BlueField‑4 STX platform, enabling line‑speed visibility and policy enforcement for AI factory workloads.
- The integration delivers Zero Trust for Agentic AI without requiring changes to host‑based software or protected applications, preserving AI performance while strengthening security.
- NVIDIA Vera BlueField‑4 STX embeds real‑time protection for data, agents, and context memory directly in silicon, creating a secure‑by‑design foundation for autonomous AI interactions.
- Xage’s solution provides end‑to‑end visibility and control across the full AI interaction chain—users, agents, models, tools, APIs, applications, infrastructure, and resources—governing exactly what each agent can access and do.
- Integration points include the Xage Resource Gateway, Xage Agent Sentry, Xage Extended Protection (XEP), DOCA Vault, Fabric policy engine, DOCA Argus, DOCA Flow, and DOCA OvS, enabling hardware‑accelerated enforcement and automated threat response.
- Together, Xage and NVIDIA DOCA form a closed‑loop security model where infrastructure observes runtime behavior, Xage evaluates identity, policy, and context, and enforcement controls allow, deny, contain, or escalate agent activity in real time.
- The combined capabilities enforce least‑privilege, just‑in‑time‑just‑enough access, block lateral movement, contain suspicious agents, and provide audit‑ready visibility, maximizing the efficiency, security, and resilience of AI factories for enterprises, government agencies, and critical‑infrastructure operators.
Introduction to Xage Security and NVIDIA DOCA Integration
Xage Security has announced support for new and enhanced NVIDIA DOCA security capabilities alongside the launch of the NVIDIA Vera BlueField‑4 STX platform. This partnership is designed to bring line‑speed visibility, policy enforcement, and granular control to AI factory environments, thereby supercharging Xage’s Zero Trust for Agentic AI offering. By operating at the infrastructure layer, the combined solution can observe, govern, and control AI interactions at massive scale without compromising the performance or efficiency of the underlying AI workloads. Importantly, the integration does not require any modification of host‑based software or the protected workloads, allowing organizations to adopt the security enhancements seamlessly.
Overview of NVIDIA Vera BlueField‑4 STX and DOCA Security
NVIDIA Vera BlueField‑4 STX is a purpose‑built AI storage accelerator that incorporates NVIDIA DOCA security directly into its silicon. This secure‑by‑design approach embeds real‑time protection for data, agents, and context memory, ensuring that sensitive information and autonomous AI agents are shielded from the moment they are instantiated. The platform leverages DOCA’s programmable data‑path capabilities to enforce security policies at line rate, detect and block threats in real time, and provide the visibility needed to govern complex agentic workflows across distributed AI factories. As a result, Vera BlueField‑4 STX serves as a hardened foundation for deploying trustworthy agentic AI at scale.
Executive Perspective on AI Factory Security
Duncan Greatwood, CEO of Xage Security, emphasized that AI factories are rapidly becoming the new critical infrastructure and therefore demand critical‑infrastructure‑grade security protections. He noted that as autonomous AI agents gain broader access to sensitive data, APIs, applications, and core systems, organizations must achieve unbypassable visibility into and control over what those agents can see, do, and change. By supporting NVIDIA Vera BlueField‑4 STX and the enhanced DOCA security architecture, Xage brings identity‑based Zero Trust to the heart of the AI factory, delivering the foundational security required for safe, mission‑critical AI adoption.
Xage Zero Trust for Agentic AI Capabilities
Xage’s Zero Trust for Agentic AI solution enables enterprises to move AI agents from pilot projects to production environments with confidence. The platform provides end‑to‑end visibility and control across the entire AI interaction chain, encompassing users, agents, models, tools, APIs, applications, infrastructure, and resources. Unlike traditional approaches that focus solely on prompts or model outputs, Xage governs precisely what each agent is permitted to access and which actions it may perform. This granular governance ensures that AI agents operate within strictly defined boundaries, reducing the risk of unintended or malicious behavior.
Role of NVIDIA Vera BlueField‑4 STX in Granular Security
Powered by NVIDIA DOCA security, the Vera BlueField‑4 STX platform delivers a foundation for applying granular security policies at line rate. It can detect and block threats to AI workloads in real time while providing the visibility and control necessary to secure agentic workflows at massive scale. By embedding security functions directly into the storage and networking silicon, BlueField‑4 STX eliminates latency typically associated with software‑only security layers, ensuring that protection does not impede the high‑throughput demands of modern AI factories.
Integration Points and Core Components
Xage’s Zero Trust for Agentic AI will integrate with NVIDIA BlueField‑4 STX through multiple touchpoints, enabling end‑to‑end protection for AI factories. The company will deliver AI Security Gateways that run natively within the NVIDIA DOCA architecture via three core components: the Xage Resource Gateway, Xage Agent Sentry, and Xage Extended Protection (XEP). These components work together to secure both the resources accessed by AI agents and the agents themselves, ensuring that policy enforcement occurs as close to the data plane as possible.
Enhanced Integration with DOCA Vault, Fabric Policy Engine, Argus, and Flow
Beyond the core gateway functions, Xage can integrate with NVIDIA DOCA Vault to provide detailed file‑access visibility and control, allowing administrators to monitor who or what is reading, writing, or modifying critical datasets. The Xage Fabric policy engine consumes information from DOCA to make real‑time interaction decisions based on current identity, context, and risk assessments. Furthermore, Xage shares security‑relevant events—such as login attempts, entitlement delegations, and credential changes—with DOCA Argus and DOCA Flow, enriching threat‑detection capabilities on the Vera BlueField‑4 STX platform and enabling more accurate anomaly detection.
Hardware‑Accelerated Enforcement via DOCA OvS and Automated Response
For rapid, hardware‑accelerated enforcement, Xage integrates with DOCA OvS (Open vSwitch) to dynamically configure the BlueField OvS stack. This permits authorized interactions while instantly blocking unauthorized activity at line speed. Xage also serves as the response mechanism for real‑time automated detection and response; upon identifying suspicious behavior, it can isolate questionable AI agents, suspend access for potentially compromised users, or apply additional protections to resources exhibiting anomalous access patterns. This closed‑loop capability ensures that threats are mitigated before they can propagate or cause business impact.
Closed‑Loop Security Model Description
Together, Xage and NVIDIA DOCA create a closed‑loop security model where the DOCA‑based infrastructure continuously observes runtime behavior for visibility. Xage then evaluates identity, policy, resource context, and intended actions to determine the appropriate response. Enforcement controls, executed through DOCA‑OvS, allow, deny, contain, or escalate agent activity in real time, preventing risky behavior from materializing into business risk. This tight feedback loop provides both proactive prevention and reactive containment, essential for maintaining trust in autonomous AI ecosystems.
Benefits: Least‑Privilege, Governance, Threat Containment, and Audit Readiness
The combined solution enforces least‑privilege and just‑in‑time‑just‑enough zero‑trust policies across AI systems, resources, data, tools, APIs, and applications, ensuring that agents only receive the minimum access necessary for their tasks. It also governs interactions between agents, models, APIs, storage systems, data, and other resources, helping to prevent unauthorized activity and lateral movement. By containing suspicious agents and blocking threat propagation, Xage limits the potential spread of compromises. Detailed operational and security information gathered by the platform supports anomaly detection, while automated threat‑response measures enable swift remediation. Collectively, these capabilities maximize the efficiency, security, and resilience of AI factories, providing audit‑ready visibility across the AI lifecycle.
Conclusion: Impact on Enterprises and Critical Infrastructure
With support for NVIDIA DOCA and the Vera BlueField‑4 STX platform, Xage extends its distributed Zero Trust architecture into the next generation of AI factories. Enterprises, government agencies, and critical‑infrastructure operators can now adopt agentic AI with greater confidence, resilience, visibility, and control. The integration delivers critical‑infrastructure‑grade security without sacrificing AI performance, enabling organizations to harness the full potential of autonomous AI while safeguarding mission‑critical data, workflows, and operations.