Nvidia's Scale-In Architecture Fortifies Agentic AI With Hardware-Based Security
Nvidia has introduced Scale-In, a new infrastructure pillar designed to secure and manage AI agents operating at scale, combining hardware acceleration with software controls to isolate and monitor autonomous systems.
Nvidia Corp. has unveiled Scale-In, an addition to its AI networking infrastructure aimed at securing, managing and operating agentic AI factories. The platform provides secure access, high-performance data movement, tenant isolation, streamlined operations and more consistent performance as autonomous agents play larger roles in AI factory environments.
The shift to agentic AI introduces distinct security challenges that differ fundamentally from traditional AI workloads. "In agentic AI, it's a completely different use case because now we're dealing with many agents that need to run on the infrastructure," said Gilad Shainer, senior vice president of networking at Nvidia. "You need to increase security because you need to make sure that those agents are contained in their sandbox and are not doing the wrong things. You need to build an infrastructure that needs to ensure security all around…what happens in the memory, what happens on storage. This is why we needed to bring an infrastructure which actually enabled the security across everything. And that's what Scale-In is."
Following Scale-In's introduction in August, Nvidia has progressively expanded its agent security offerings. September brought the Open Agent Safety Platform, which pairs OpenShell with Nvidia Sentry to provide out-of-band monitoring and hardware-enforced security measures. An October rollout added the OpenShell Reference Architecture, which incorporates Vera Rubin infrastructure, Vera CPUs and BlueField-4 to enable secure agent execution across enterprise deployments.
In-silicon security for AI networking
Scale-In relies on Nvidia BlueField-4, Nvidia DOCA and Nvidia Spectrum-X Ethernet. BlueField-4 represents Nvidia's fourth-generation data processing unit, a networking platform engineered for agentic AI factories and large-scale infrastructure deployments. According to Shainer, BlueField proves essential for managing agentic AI security.
"You can think about BlueField or Blue-Field-4 as a device that has in-silicon security engines that are working out of band and collecting telemetry from everything, from memory, from storage, from GPUs, from CPUs," Shainer explained. "That's why that infrastructure connects to all other infrastructures and ensures security all around. That out-of-band monitoring enables BlueField to actually understand exactly what agents are doing."
Alongside BlueField-4, Nvidia DOCA serves as a core component of Scale-In. This software platform facilitates rapid development of applications and services that offload, accelerate and isolate data center workloads. Through DOCA, developers construct software-defined, cloud-native, DPU- and SuperNIC-accelerated solutions. Similar to how Nvidia CUDA provides a programming interface for GPU workloads, DOCA grants developers access to DPU acceleration capabilities.
"DPU brings the in-silicon engines security engines, storage accelerations," Shainer noted. "DOCA is the framework, the software, the APIs that enable you to access all of those engines that BlueField-4 provides."
Nvidia has strengthened agentic AI security through OpenShell, an open-source runtime designed to govern agent execution, access and inference routing. Nvidia paired OpenShell with BlueField-4 to create a separate device capable of gathering information outside the GPU and monitoring an environment that agents cannot influence.
"I want to also have another device which is completely separated, completely isolated from the agents, that can monitor everything out-of-band," Shainer stated. "That's what we're doing with BlueField-4, bringing all the security elements together with OpenShell, and now you have a secure infrastructure for agents. This is how the combination of OpenShell and BlueField-4 brings all the security elements that we need for gen AI."
The OpenShell Reference Architecture, introduced more recently, builds upon this foundation by incorporating Nvidia Sentry, which executes on BlueField-4 using DOCA software. Sentry continuously observes agent behavior and can isolate agents that attempt to exceed their defined software boundaries. This combination augments OpenShell's runtime controls with an independent enforcement mechanism.